Professional Documents
Culture Documents
International Journal of
ISSN 2225-658X (Online)
DIGITAL INFORMATION AND WIRELESS COMMUNICATIONS
Volume 3, Issue 1
2013
TABLE OF CONTENTS
Original Articles
PAPER TITLE
AUTHORS
A NEW METHOD TO CHANNEL ESTIMATION IN OFDM SYSTEMS Ali Asadi, Behzad Mozaffari
BASED ON WAVELET TRANSFORM
Tazehkand
PAGES
1
10
19
COLOR
EDGE
DETECTION
Mohamed Almadhoun
25
43
50
67
81
93
103
DEVELOPMENTS,
THREATS
Wolfgang
Kahlig
Kahlig,
Eleonora
112
125
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
In OFDM systems, it is necessary to estimate the
channel to overcome the distortion caused by
channel fadings which can be induced by many
phenomena such as: delay spread, multipath effect,
mobility and Doppler shift. Most of the channel
estimation techniques are proposed in frequency
domain using the pilot symbols. One of them which
has less complicated is well-known as Least-Squares
(LS) method which is widely used in channel
estimation but it is more sensitive to noise respected
to the other reported techniques. In this paper, a new
threshold
based
method
using
wavelet
decomposition will be proposed which is based on
an initial LS estimation technique. The reported
simulation results show that the proposed method
has better performance compared to the other
methods such as Lee Method that has been published
recently.
KEYWORDS
OFDM signal, Channel Estimation, Channel State
(CS), Least Square (LS), Wavelet Transform, Cyclic
Prefix, Channel Impulse Response
1 INTRODUCTION
OFDM is a signaling technique that has been
applied widely in wireless communication
systems due to its ability to maintain effective
transmission and highly efficient bandwidth
utilization in the presence of various channel
impairments which one of them is frequencyselective fading. In OFDM systems the
available spectrum are divided into many
orthogonal
sub-channels,
which
are
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
X(k)
S
/
P
Adding
Pilots
x(n)
x(n)
P
/
S
Adding
Cyclic
Prefix
IFFT
Channel
w(n)
+
X(k)
Demapper
P
/
S
Channel &
Data
Estimation
x(n)
FFT
x(n)
Remove
Cyclic
Prefix
S
/
P
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
1
N
LN 1
k 0
X Ke
2
kn
LN
0 n LN 1
(1)
n N c , N c 1, , 1
n 0,1, , N 1
(2)
CP
OFDM Symbol
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
pilot ,
Xm
data ,
Frequency
Tc
(6)
Frequency
OFDM Symbol
Bc
Time
Figure 4. Comb type arrangement of OFDM symbols
(7)
p
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
(k k p )
(k k p )
H ( k ) [1
]H (k p )
H (k p L ) .
L
L
(8)
1
tb
(9)
x(a, b)
x ( t ) * (
) dt a 0, b .
a
x (t )
j ,k
ck (t k )
(11)
(t ) h1 ( k ) 2 (2t k )
k
X0
X
LP
k j 0
j ,k
j ,k
(t )
(10)
( t ) 2 j / 2 (2 j t k ) .
HP
LP
(t ) h0 ( k ) 2 (2t k )
X1
X2
4 PROPOSED METHOD
Since LS estimator is much sensitive to noise,
we introduce a time-domain method to channel
estimation which its objective is improvement
the performance of LS algorithm. In this paper it
is assumed that the pilot spaces are related to the
maximum delay caused by the channel. The
impulse response of the channel can be
calculated using IFFT based on initial LS
estimated channel as following equation.
hLS ( n ) IDFT { H LS ( k )}
h1 ( n ) h ( n ) w '( n ) 0 n L 1
L n L 1
h2 ( n ) w '( n )
0
L n N 1
(12)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
median ( Di )
(13)
0.6745
hLS ( n ),
hp (n)
0
(14)
hLS ( n )
Di
Pilot
Extraction
H LS ( k )
IDFT
Wavelet
Decomposition
hLS ( n )
h2 (n)
h1 ( n )
Threshold
Computation
Denoising
Algorithm
h ( n )
5 SIMULATION RESULTS
In order to evaluate the proposed method it is
necessary that a simulation must be taken. The
simulation parameters are listed in table (1). In
all simulations the block coding and interleavers
Related Values
N=2048
Nc=512
QPSK , 16QAM
Db4
N=2048
L=10
15, 30
Rayleigh with AWGN
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ACKNOWLEDGMENT
The authors would like thanks IRTC in order to
supporting on this project
7 REFERENCES
1.
Amplitude (dB)
-20
-30
-40
-50
0
Ideal
LS Method
50
100
150
Subcarrier Index
200
250
Amplitude (dB)
-20
-25
-30
Ideal
Proposed Method
-35
-40
0
50
100
150
Subcarrier Index
200
250
Amplitude (dB)
-20
-25
-30
Ideal
Lee Method
-35
-40
0
50
100
150
Subcarrier Index
200
250
6 CONCLUSIONS
In this paper, a new method to estimating the
faded channel based on wavelet decomposition
is presented. In this method, a threshold based
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 1-9
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
1. INTRODUCTION
10
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
RNA is more like DNA, it have some different
changes than DNA. RNA is a chain of nucleotides,
these nucleotides composed of ribose sugar instead
of deoxyribose in DNA, phosphate, and base. RNA
contains thousands of just four bases but with
Uracil in place of Thymine.
Proteins are the basic material that makes anybody.
It is responsible for all functions that occur in the
body, body structure, and all the physical
characteristic of the body. Proteins are the most
important molecule in the body, because any
mutations happened in the proteins may lead to the
protein losing its functionality, and hence its role in
the body, and hence diseases begin. Proteins are
long chains of Amino Acids (AA). There are only
twenty amino acids that made up all the proteins of
the body
Gene expression is the process of converting genes
into proteins in two basic steps: DNA is converted
into mRNA in a process called transcription, and
then mRNA converted to proteins in a process
called translation
DNA microarray is a technology that it is used to
measures the amount of gene expression in the cell.
It is used in the study of behavior of genes under
certain conditions, or at different time frames. This
technology enables scientists to study what happens
to genes in different disease conditions [2].
The result of microarray is a gene expression
dataset which is a tow dimensional array - data
table - where every row represents one gene and
every column represents one condition or sample.
Every cell in this table represent the expression
level Lij of gene i in condition j, as in table 1.
There is many Data mining techniques used for
analyzing gene expression datasets in order to
extract information and finding relations between
these data such as clustering and classification.
Clustering is used to assign data to groups, or
classes like classification, but the difference is that
in classification the classes are predefined, but in
clustering the classes are not determined before
applying the algorithm. Clustering algorithms can
be applied on gene expression datasets in order to
group genes that have the same expression together
and this is helping in the study of diseases such as
cancer. There are three ways of clustering:
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
grouped together, as in figure 2, and this enhances
the results of biclustering process.
Table 1. Gene Expression Data Matrix
Cond 1
Cond 2
Cond 3
Cond 4
Gene 1
Gene 2
Gene 3
Gene 4
Gene 5
| || |
( )
Where
Sub Row Average
| |
( )
| |
( )
( )
2.2 SPHIER
Figure 2. Bipartite Graph after crossing minimization
2. RELATED WORK
Biclustering algorithms can be classified in two
main categories: graphical and non-graphical
biclustering. We focus our discussions on graphical
biclustering which is the main interest point of our
research for general discussion on biclustering
algorithms can be found in [3, 4]. There is another
good reference that focuses on Barycenter crossing
minimization which the main goal of our work is to
enhance Barycenter algorithm [5].
12
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
search technique to find global optimum solution.
He adapts the Tabu search in Marti [9] to solve the
crossing minimization problem and reach to better
order of dataset. Tabu search run in two main steps:
first, construction of initial solution and second,
iterative improvement of this solution until reaching
the optimum solution. Iterative improvement has
two steps: Intensification and Diversification and
each step of them has three steps: normal,
influential, and opposite and the algorithm moves
between these three steps as it insert or remove
nodes depending on their Barycenter and if this
move reduces the number of crossings in the graph
and if there is no enhancement it stops after a
specified number of iteration this process continue
running in this way until reaching the optimum
solution.
2.4 BIMAX
BiMax Algorithm was proposed by Prelic et al [10];
the main idea behind this algorithm is to
transforming the gene expression dataset inputs into
binary dataset (or 0s and 1s) by discretization
process, where 0s means no change in expression
level and 1,s from change in expression level.
The row set is split into GU- the rows that have only
ones in CU, GV - the rows that have ones in CV only,
and GW - the rows that have ones in both. If U and
V do not share any rows and columns of the matrix
E, i.e. GW is empty; the two matrices can be
processed independently from each other. If U and
V have a set GW of rows in common as shown in
Figure 3.6, special care is necessary to only
generate those biclusters in V, that share at least
one common column with CV.
2.1 SAMBA
SAMBA [11] stands for statistical algorithmic for
bicluster analysis. The main goal of this algorithm
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
power of computers increases rapidly, so we think
the time is not factor as efficiency, because at the
end the target of biclustering is to get accurate
results.
SAMBA algorithm is one of famous algorithms that
achieves good statistical results, runs in a good
time, it also used the methodology of bipartite
graph, and used the weight of edges and non-edges
along with statically functions to give weight for
each subgraph, so it differs from the previous the
other algorithms that we discussed, and which using
bipartite graph in that it used the weight of edges
and non-edges in calculations, but other algorithm
depend on edges only in the calculations, and also
depend on local search heuristics to find biclusters.
BiMax algorithm looks like Cheng and Church in
that the two algorithms do not use graph
representation in solving biclustering problem,
BiMax algorithm depended on transforming the
dataset set into binary dataset in a process called
discretization, and then work on this dataset by
divide-and-conquer methodology by partitioning
the dataset to rows and columns, and add them to
clusters depending on the property they share, the
idea of discretization used in other algorithms such
as Tabu search algorithm to minimize the time
taken in crossing minimization process, this
technique decreases the running time.
We can conclude from this discussion that the most
significant algorithms of the algorithms we
discussed is Tabu search, and SMABA algorithm,
because they achieve the best results, and used
different valuable processing techniques.
3. PROPOSED ALGORITHM
There are many techniques used for Biclustering
one of these techniques is constructing a bipartite
graph from gene expression dataset, reordering
nodes in this graph using crossing minimization
technique then apply biclustering algorithm on the
reordered bipartite graph. Most of the algorithms
used to crossing minimization in bipartite graph
divided into three basic steps: the first step is the
construction of bipartite graph, the second step is
the initial ordering, and the last step is iterative
improvement.
(5)
Where:
Let vi represent the ith node in the non-static layer,
and set Ni represents the set of neighbors of vi, also
let rj represent the rank of jth member of the set Ni
[12].
We success to solve the problem of crossing
minimization in a way that achieves better results in
crossings number rather than Barycenter heuristics
[13]. We use the same way of Barycenter heuristics
used in [13] to minimize crossings in bipartite
graph, but we make some changes to the
calculations of weight of each node and using this
weight as a new rank for reordering nodes. We
calculate the Barycenter of each node as in
algorithm 2, then calculating the new weight by
adding the rank of each node to the sum of the
ranks of its neighbors as in algorithm 3, and taking
the position of each node into consideration of our
calculations as in algorithm 4, then using this rank
for reordering the nodes.
The algorithm of biclustering has three main basic
steps:
-
14
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Algorithm 1: Bipartite Crossing Minimization
1. Construct Bipartite graph BG
2. Compute the Barycenter for each
node in the two layers, as in
algorithm 2
3. Compute the new weight for each
node in the two layers, as in
algorithm 3
4. Reorder nodes in the two layers
5. Compute crossing minimization for
bipartite graph
6. Repeat steps from 2 to 5 until
the number of crossing of the
current bipartite graph is larger
than the number of crossings of
previous bipartite graph from
last iteration
Reordered
graph
after
crossing
Output : Biclusters
startRow = 0;
startCol = 0;
Index = 0;
FOR i = 0 TO numCols DO
colSum[i] = 0;
blockSum[i] = 0;
Residue[i] = 0;
FOR r = startRow TO numRows DO
rowSum = 0;
FOR c = startCol TO numCols DO
colSum[c]=colSum[c]+MatrixBG[r][c];
blockSum[c]=blockSum[c-1]+colSum[c];
colCount = c - startCol + 1;
rowCount = r - startRow + 1;
blockMean = blockSum[c]/blockCount;
END FOR
15
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
4. EXPERIMENTAL RESULTS
The proposed algorithm is written in C-sharp
language and the experiments are running on i5 2.3
GHz processor and 4 GB of RAM, and under
windows 7 operating system. We apply the
algorithm on cancer gene expression datasets:
lymphoma dataset [14], gastric cancer dataset [15],
and AML prognosis dataset [16] to test the
accuracy of the proposed algorithm relative to
Barycenter used in SPHier algorithm.
We apply the proposed algorithm and Barycenter
algorithm on the three cancer datasets, and compare
the results of the two algorithms according the
accuracy of crossing minimization.
After applying the proposed algorithm, and
Barycenter algorithm of crossing minimization on
the cancer datasets we get the results in table 2, that
it is represented in figure 3.
In this experiment we are focusing mainly on the
accuracy of the algorithm, because the most
important thing in the gene expression datasets
analysis is the accuracy of results more the time the
algorithm takes, because of the revolution in
computational power of processors solve many of
problems of computations time.
Figure 3 shows enhancements that our algorithm
made in the Barycenter algorithm, the proposed
algorithm achieves better results especially in large
size gene expression datasets with a small number
of conditions.
Because of the decrease in crossings number of
bipartite graph leads to grouping related node
together, the enhancement in crossings number
increases the efficiency of biclustering algorithm
[7].
Table 2. Crossings number of original dataset, Barycenter
algorithm and proposed algorithm
Original
Dataset
Barycenter
Algorithm
Proposed
Algorithm
Gastric
11,048,714,141
11,048,713,498
11,048,654,635
AML
Prognosis
131,451,569,724
131,451,501,126
131,450,516,864
Lymphoma
100,790,992,882
100,790,992,882
100,790,645,942
Standaderd
Deviation
300000
200000
BaryCenter
Algorithm
100000
Proposed
Algorithm
0
7129 12625 22283
Number of genes in thouthands
Figure 3. Comparison between standard deviation of
Barycenter algorithm and proposed algorithm from the
original crossings number
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
proposed algorithm takes much time than
barycenter algorithm, because our algorithm depend
on counting the crossings number for every single
iteration of the algorithm, and this takes long time
to compute.
The area of biclustering is very complex and all
researches in this area is considered as a domain
specific solution, and in some cases its a problem
specific solution, we can find algorithms that find
good biclusters in some cases, and in other cases
these algorithms do not perform well, so there is no
perfect biclustering algorithm or magic algorithm
that can solve all the problems in a prefect way.
We want to state that the proposed algorithm is an
idea for finding the relations between numbers, and
its positions in the dataset, and how these positions
affect the process of crossing minimization. We
take this trend to make it our work in the research,
and we cannot say that it is the perfect solution of
crossing minimization problem, but it is a new
trend of research that we want to go through it.
9.
Number of conditions
Gastric
7,129
30
AML Prognosis
12,625
58
Lymphoma
22,283
29
17
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 10-18
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
14. Raetz EA, Perkins SL, Bhojwani D, Smock K et al. Gene
expression profiling reveals intrinsic differences between
T-cell acute lymphoblastic leukemia and T-cell
lymphoblastic lymphoma. Pediatr Blood Cancer 2006
Aug; 47(2):130-40.
15. Hippo Y, Taniguchi H, Tsutsumi S, Machida N et al.
Global gene expression analysis of gastric cancer by
oligonucleotide microarrays. Cancer Res 2002 Jan 1;
62(1):233-40.
16. Yagi T, Morimoto A, Eguchi M, Hibi S et al.
Identification of a gene expression signature associated
with pediatric AML prognosis. Blood 2003 Sep 1;
102(5):1849-56.
18
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 19-24
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
Edge detection process has a widespread usage in
computer vision applications. But it has a different
output when its input image changes from color to
grayscale. This changeability of results make us add
modification on edge detection process procedures to
correctly detect all those edges in color images that
cant be detected in gray ones. This research reviews
the proposed solution of Dutta and Chaudhuri on color
edge detection algorithm that works using RGB color
space, detects problem of a huge set of undetected
edges by their proposed algorithm and find solutions
for that, and applies complexity and performance
analysis and experiments to compare the proposed
algorithm with Canny and Sobel edge detection
algorithms.
KEYWORDS
Computer vision, Color edge detection, Enhancement
on edge detection, Complexity analysis, Performance
Comparison.
INTRODUCTION
19
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 19-24
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
2
ENHANCEMENT ON ALGORITHM
(2)
(1)
Pixel(i,j)=2*10+3*210+4*200 = 1450
So it will not consider the divider between two
areas of those two colors as a boundary.
Figure 1 and following images show how applying
algorithm did draw the edge between the two
colors
COMPLEXITY ANALYSIS
(a)
(b)
20
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 19-24
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
This step runs two different loops, number
(n),
perspective
4xn
3- Threshold Technique
which
of
means
time
they
are
complexity,
equal
from
but
when
4- Edge thinning
This operation has 2xn of assignments and
complexity of (n) when n=number of
image pixels.
As for Canny edge detection algorithm, it is multistage starts by detecting edges with miximizing
probability of edge points and minimizing nonedge points, then localization and resulting
resulting one real edge [6], by passing through the
algorithm it follows these steps:
1- Convert to gray: costs n iterations if n is
the number of image pixels
2- Normalize contrast: runs n+255+n, so its
complexity is (n), where n is number of
image pixels.
3- Compute gradients: costs at most is 3 x (k
x n)+ n iterations for the four loops, so
its (n). [k is the number of kernel
elements]
4- Perform hysteresis: costs at most n, so its
(n).
5- Threshold edges: costs n iterations, so its
complexity if (n)
21
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 19-24
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Performance measurement experiment applied the
three algorithms on one image at first, then on two
images, on three, until reaching to apply 10 images
at a time for each algorithm. Applying algorithm
in a loop on more than one image discovers the
difference between them when using algorithm in
video processing.
Time in MilliSeconds
14000
12000
10000
8000
6000
4000
2000
0
# of processed images
Sobel
10
1064.4699 2036.1599 2908.2182 3861.8333 4763.6225 5937.2051 6830.3131 7679.9482 8578.4878 9605.5161
Proposed 2408.2262 3463.8285 5182.9545 6720.1038 8367.0729 9976.6211 11724.613 13379.057 15010.007 16549.499
Canny
758.33158 1345.6685 1750.5082 2319.3174 3066.8976 3335.9226 4047.0674 4639.1153 5192.539 5696.5223
Figure 3. Curves of execution time values for processing multi-images in each algorithm
22
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 19-24
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
The main reason here is that basic operations of
the proposed algorithms need more execution time
80000000
70000000
60000000
50000000
40000000
30000000
20000000
10000000
0
# of processed images
Sobel
10
8967024 17934048 26901072 35868096 44835120 53802144 62769168 71736192 80703216 89670240
Proposed 6531668 13063336 19595004 26126672 32658340 39190008 45721676 52253344 58785012 65316680
Canny
2944842 5889684 8834526 11779368 14724210 17669052 20613894 23558736 26503578 29448420
Figure 4. Curves of number of basic operations for processing multi-images in each algorithm
CONCLUSION
2.
3.
4.
5.
6.
REFERENCES
1.
7.
23
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 19-24
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Wesley, ISBN-10: 0321358287 | ISBN-13: 9780321358288, (2006).
8. Lilja, D.: Measuring Computer Performance: A
Practitioner's Guide. Cambridge University Press,
New York, NY, ISBN 0-521-64105-5, (2000).
24
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
This paper examines the use of social network media at
three aspects in African and Libyan perspective.
Firstly, to use social network media as an open network
learning environment that
provide service for
interaction necessary for learners to support
socialization and collaboration during problem solving.
Secondly, to use social media as a tool to support
blended learning in e-learning system and encourage
non-native English students to express their ideas and
fill the gap of communication problems. Thirdly, to
analyze the interaction of the learner in social media
threaded messages and its relation to group and
individual performance using different social schema
and social network analyses. Quasi-experimental
results indicate that there is an increase on the
cognitive level of students at different level while
qualitative results reveal that it helps deepen learning,
memorable, have freedom to express opinions and
lessen pressure and increase communication and
socialization.
KEYWORDS
blended learning, collaboration, e-learning, network
learning, socialization, social network media
1 INTRODUCTION
Learning English is not mandatory in Libya, few
years back the language has never been integrated
in the curriculum and English was static and never
been practice outside the school. Thus, foreign
lecturers who were tasked to deliver computer
science and information technology education
suffered communication gap among learners and
used different method to lessen the problem. One
method is to employ e-learning systems that
allows students to learn freely at their home and at
school whenever possible. Given the emerging
Abdalla Radwan
Computer Science Department
Sirte University Libya
radwan2004@hotmail.com
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
the goal is not that easy, since it will entail proper
communication and conversation among members
in the group. One should not feel superior or
inferior among others; otherwise a catastrophic
effect in the learning process will occur.
In the field of e-learning, collaboration usually
takes place using pre-programmed agent or
animated character agent by virtually deceiving
the learner that someone is helping him in the
learning process, but in reality it was preprogrammed based on learners prior knowledge
and personal profile. Many researchers have been
trying to develop collaborative software and
integrate to e-learning module but none of which
have surpassed the power of human socialization.
For example, early e-learning system with
animated agent cannot joke, greet and show facial
expressions because of its limited domain,
complexities, issues and constraints [23]. These
simple gestures can build rapport and develop
personal and affective relationships among
members and somehow affect the overall
performance of the learning group. Thus, instead
of pre-defined and pre-programmed socialization;
social networking has been adopted for
collaboration.
The growth of social networking has created new
opportunities for collaboration in problem solving
[13],[24]. In 2009, University of Cambridge
posted a problem in mathematics using blogs and
solved within six weeks known as Polymath I [25].
Social networking media let people rendezvous,
connect or collaborate, support network of people,
share content and services that are more adaptable
and responsive to changing needs and goals [26].
Social media have already led to widespread
adoption of portfolios for learners bringing
together learning from different context and
providing an on-going record of lifelong learning,
capable of expression in different forms.
The rapid diffusion and public acceptability of
social network media such as Facebook, Yahoo
Messenger, Windows Media Live and Skype have
enable users to connect with people more than
ever before. Student used social media at school
for various purposes such as socialization, sharing
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
divided into 5 groups consisting of 5 or 6
members; overall, there were 9 groups in the
study. Group members were randomly selected as
suggested by [29],[30].
The topics included in the course Algorithm in elearning module have been selected/driven by
either by the problems practical importance or by
some specific characteristic making the problem
an interesting research subject. The following
topics have been included in the module such as
sorting techniques, searching algorithms, string
processing, graph problems, combinatorial
problems. These topics are all suitable for
collaborative problem solving.
27
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
40% is out-messages. In-messages are the number
of messages contributed by all groups related to
the problem at hand while the out-messages are
interjected social interaction. Though outmessages are threaded communication that are not
related to problem, it becomes integral in the
overall collaboration by providing informal
communication. Informal communication entice
the collaboration by making jokes, creating
relationship, encourage to talk and share opinions,
thus, allowing learner to increase their level of
collaboration. Among the 9 groups studied, Group
9 emerge with 101 messages (66 in-messages and
30 out-messages), Group 1 contributed
82
messages (62 in-messages and 20 out-messages)
while Group 6 contributed 82 total messages (34
in-messages and 48 out-messages). It appears from
the table that Group 9 is the most serious group
during collaboration as indicated by 66 inmessages. Group 6, 4 and 2 are the most least
sociable group with 5,10,15 out-messages reported
respectively. This is perhaps attributed to random
selection of participants. Group 8 has a 51 outmessages making it the most sociable group as
indicated by many times used of social interjection
like greetings to one another, asking how is family
life, asking about weather, news and others,
evident to Arab learners everyday life.
Group 1
(5 )
Group 2
(5)
Group 3
(5)
Group 4
(5)
Group 5
(6)
Group 6
(6)
Group 7
(5)
Group 8
(5)
Group 9
(6)
62
20
82
24
15
39
26
25
51
60
10
70
34
48
82
28
5
33
26
20
46
18
51
69
66
35
101
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
due to elaborative nature of discussion
nevertheless it was coded using the guided
example provided by the two coding schema.
Veldhuis-Diermanse Schema
Table 2 summarized the coded learning processes
using Veldhuis-Diermanse Schema where type of
learning has been divided into four subgroups;
Cognitive, Affective, Meta-cognitive and Others.
Other category is the number of out-messages.
Among the 344 in-messages from Table 1, it was
further breakdown to Cognitive category with 155
or 45% of the total in-messages , Affective with 91
or 26%, and Meta-cognitive with 98 or 29%. For
cognitive category, Group 1 has 30 messages,
Group 9 with 27 messages and Group 4 with 25
messages, they were ranked 1, 2 and 3
respectively. In this category, member of the
groups debated and discussed, used external
information and experiences and linked internal
information or concepts found. It involves
knowledge and the development of intellectual
skills by recall or recognition of specific facts,
Group 1
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
Group 8
Group 9
30
17
15
62
12
4
8
24
10
10
6
26
25
15
20
60
18
6
10
34
13
7
8
28
14
8
4
26
6
6
6
18
27
18
21
66
Cognitive (155)
Affective (91)
Metacognitive (98)
Total (344)
Anderson Schema
Table 3. Units of Meaning Coded for Tutoring and Communication Process (Anderson Schema)
Type of Tutoring Process
Group 1
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
Group 8
Group 9
15
20
27
62
7
7
10
24
2
14
10
26
13
18
29
60
8
12
14
34
8
9
11
28
7
9
10
26
3
5
10
18
16
20
30
66
29
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Facilitation has 120 total messages out of 344 inmessages or 35% have been coded into this
category. Group 1 and Group 9 coded with 20
messages respectively followed by Group 4 with
18 messages. In this category, indicators show that
student making agreement/disagreement, seeking
understanding/consensus, reinforcing student
contributions and assessing the efficacy of the
collaboration process.
Table 4.1 Blooms Cognitive Level Coding of Different Groups (after reading e-learning module)
Cognitive Level
Group 1
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
Group 8
Group 9
Evaluation
Synthesis
Analysis
Application
Understanding
Knowledge
1
0
1
1
1
2
0
0
0
0
1
1
0
0
1
0
0
1
1
1
0
1
2
1
0
0
0
0
1
1
0
0
0
1
1
1
0
0
0
0
0
1
0
0
0
0
1
0
1
1
1
1
2
3
Cognitive Level
Group 1
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
Group 8
Group 9
Evaluation
Synthesis
Analysis
Application
Understanding
Knowledge
2
2
3
3
3
3
0
1
1
1
2
2
0
1
1
2
3
3
3
1
2
2
3
3
1
1
1
2
3
3
1
1
1
1
2
3
0
0
1
1
2
2
1
0
1
1
2
1
1
2
3
2
3
3
Table 4.2 Blooms Cognitive Level Coding of Different Groups (after collaborative problem solving)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Knowledge remain at 1; a perfect score for pre and
post-exam. For knowledge category, students
exhibit memory of previously learned materials by
recalling the content of the e-learning module, so it
is not surprising that in the pretest, these categories
have high result.
Understanding category also increased due to prior
knowledge taken by the students from the elearning module, students were able to organize,
compare and interpret data given a change of the
input of the algorithmic problems of the course
presented to them during collaboration. Initially,
Synthesis and Evaluation have the lowest value
but it is where the cognitive level significantly
increased. The highest cognitive level in Blooms
taxonomy is Evaluation where student are
expected to demonstrate the ability to validate data
based on predefined criteria. Student for example,
were able to prove that the time complexity of
sorting technique varies accordingly as input
changes. Another criteria in analyzing the
algorithm is generality, where certain inputs can
be changed into different format and re-evaluate
their coded program to fit the new input. In the
Synthesis, students can change the program from
one data types to another, e.g. from integer, to
character, or string or another data types in their
coded program. These changes
were not
Knowledge
Understanding
Application
Analysis
Synthesis
Evaluation
Before
After
Before
After
Before
After
Before
After
Before
After
Before
After
Member 1
0.33
0.98
0.33
0.66
0.33
0.66
0.00
0.33
0.33
0.33
0.00
0.33
Member 2
0.33
0.98
0.33
0.98
0.00
0.66
0.00
0.33
0.33
0.66
0.00
0.66
Member 3
0.66
0.98
0.33
0.98
0.33
0.66
0.00
0.66
0.33
0.66
0.00
0.66
Member 4
0.98
0.98
0.33
0.98
0.66
0.99
0.00
0.98
0.33
0.66
0.33
0.33
Member 5
0.66
0.98
0.00
0.98
0.98
0.99
0.33
0.98
0.00
0.66
0.33
0.66
Member 6
0.66
0.98
0.33
0.98
0.98
0.66
0.00
0.33
0.33
0.33
0.00
0.33
Average
0.60
0.98
0.28
0.93
0.55
0.77
0.06
0.60
0.28
0.55
0.11
0.50
31
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Table 4.4 Comparative Group Performance of All the Group in 2nd Semester
( T-test Before and After Collaboration)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Sadiya for Group 6, Mona for Group 7, Fathiya
and Saber for Group 8 and 9 respectively.
Participants tend to communicate to the person
whom they perceived intelligent or have
exceptional characters among the groups.
As deadline approaches, collaboration tend to
change and communication effort become more
dense. Figure 2 shows how interaction seems to
change more drastically throughout the ending
part. The central participant in the preliminary
collaboration phase have established their position
except for Group 9 where initially it was Saber but
was taken gradually by Mosbah. It is very
interesting to note that other central players of
other groups seem to collaborate with Mosbah,
making him the most active and the most
33
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
contribution, everybody is engaged. . For this, in
the beginning student gets to know each other and
slowly
increases
the
collaboration
and
communication. As time goes by and each
become relax to deal one another. Surely, there
was an increased of learning since no one among
the groups lose interest during their collaborative
problem solving. The interaction pattern clearly
showed transformation of membership as learners
Measures of Centrality
There are many important properties within a
social network that relate to the identification of
the power of relationships in the network structure.
It was revealed that certain participants in
collaborative online learning communities have a
greater power within the community than others.
This was evident in the study when comparing the
fact that some participants attracted a number of
responses to their queries while others attracted no
responses and quickly become isolated. Taking
into account the preliminary group collaboration
of Figure 1 and the post group collaboration of
Figure 2, the following centrality has been
computed Table 5.1 and 5.2 for degree centrality,
Table 5.3 and 5.4 for betweenness centrality and
Table 5.5 and 5.6 for closeness centrality.
Degree Centrality refers to the number of
connections that a node contains and indicates the
level of activity of a node within a group. The
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
degree centrality an indication that all groups were
actively involved and working very close while
Group 9 shows a very good relationship to each
other as their degree centrality becomes all equal
at the end of the collaboration. It means that there
is no competition among the members as they are
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
observation that collaboration takes place and
students gradually contributing to problem solving
as the maximum betweenness and average
decreased drastically. It means that students
actively participated in problem solving.
Analyzing Group 9, preliminary shows that
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
shortest paths to all others they are all closed to
anyone else and they are in excellent position to
monitor the information flow of the network.
Group 9 has all equal value at post collaboration
38
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Table 5. Qualitative Themes
Theme
-Learning is fun, better, better, better! Thinking more!
-Memorable! Because of video, I remember because it is interactive and can repeat the simulation.
-We are free, we can say what we want, and get helped, No pressure.
-Support socialization, allowed us to chat, practiced English
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
collaboration. The inclusion of interactive
simulations in the e-learning improve the quality
and outcomes of the study. By knowing the
concepts, learners actively collaborate and
negotiate among themselves as shown in the social
network analysis.
We are free, we can say what we want, and get
helped, No pressure.- The pouring of in and outmessages during collaboration relate this, they
have freedom to express their opinions in the
social media without limitations. Many students
try to communicate with other learners in and out
of their circle to seek help. Students can
collaborate without the pressure and watchful eye
of their instructor. This is illustrated through the
following excerpts:
Fatma claimed that Libya is free so are
we to express our opinion, I studied in my
house and chat with my friends, it make
sense since we can tell to our parents that
we are studying instead of just searching
and chatting non-sense, I learned a lot so
it is not surprising that I have a good mark
in the exam.
Rowida also said that We can say what
we want, seek help to our friends, and
collaborate with the groups. I get some
inputs and explanation from the other
group, and it helps me a lot. Sometimes we
have a good output if no pressure is given
to us, just ample time to solve the
problem.
40
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
4 CONCLUSIONS
This paper has presented the results of an approach
to content analysis of messages exchanged during
collaborative problem solving of approximately
three weeks duration. This analysis has enable the
tentative identification of patters of individual and
group learning during the activity. The students
interactions have been analyzed using the
Veldhius-Diermanse schema and Anderson
schema and relate the results to group and
individual
performance.
The
cognitive
performance level of students increased
dramatically after collaboration.
The social network analysis likewise shows the
density of collaboration among learners. This
provides little insights into a key aspect of the
individual and group processes and argued that
there is a need to perform more analysis to
understand fully the richness of these learning
interaction. It also concluded that the more
threaded communication the better, but it doesnt
guarantee that such quantity is transformed into
effective collaboration. The out-messages for
example are communication but not related to the
problem at hand and yet, it was concluded that this
provides informal communication and support
socialization.
In the interview, four themes have been selected
based on the written recall of events, experiences
and observed during the study. This recall has the
potential to access aspects of learning that are not
directly available in discussion or transcripts. But,
having many messages, performing saturation or
filtering is difficult. In summary, the study
successfully reported the used of event recall
among participants, the used of several content
coding analysis and the used social network
analysis.
2.
3.
4.
5.
6.
7.
8.
9.
5 REFERENCES
10.
1. Ballera, Melvin and Aziza Ehmaid Omar. ( 2012).
Exploring Social Networking Technology and
Multiple Pedagogical Agents: How, When and to
What Extent they Facilitate Learning in E-Learning
System. Proceedings of 2012 International
11.
41
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 25-42
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
42
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
With a massive arrival of information
technologies in the past quarter century there
has been a change of the set processes in the
society. These changes concerns also
education. Online courses are nowadays
well established forms of learning and
teaching all over the world. They are used as
supporting courses of traditional classes, as
complementary courses of hybrid or blended
courses or pure online courses. Although
they cannot fully replace a teacher, they are
now becoming a powerful tool in the
educational process, both formal and
informal. Nevertheless, in all cases they
must be designed well to attract and
motivate students to work and study the
uploaded information. Therefore, this article
focuses on the analysis of online materials
and their impact on the process of learning.
KEYWORDS
eLearning, Internet, Learning Management
System, online courses, online material,
learning, survey.
1 INTRODUCTION
In the Czech Republic there are
nowadays 26 public universities, two
state universities and more than 40
private universities. The Faculty of
Informatics and Management (FIM) is
one of the five faculties of the University
of Hradec Kralove educating specialists
in the area of informatics, economics,
43
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Management System) called Blackboard
Learn .
At present, more than 220 e-courses are
offered, with more than 45 of these
being foreign language courses. Several
of them can be taught completely online.
However, most of them are blended as
there are usually a few tutorials. At
tutorials (face-to-face teaching) students
usually discuss the problems they come
across when doing different types of
tasks or writing assignments. Moreover,
sometimes there are regular classes and
e-courses are used as reference courses
(i.e. students can once again read the
information obtained in class and do
some additional exercises to practise
their knowledge) for further self-study or
revision of the lecture. For more
information see Frydrychov Klmov &
Poulov [1].
The creation of online courses is, of
course, very demanding for teachers and
therefore a number of trainings focused
on the use of ICT, course design in the
virtual learning environment, and also on
the methodology of online materials
design from the pedagogical and didactic
point of view were prepared for them in
the past years. The eLearning course
should provide students with information
needed for their successful study in the
form of study materials. Furthermore, it
should enable students to verify the
acquired knowledge with the help of a
number of tests, provide them with a
simple tool for communication with a
teacher and other students and a
possibility to submit and possibly revise
their assignments or term projects. In
addition, it should enable teachers to run
the educational process in an effective
way and help with record keeping.
44
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
burdensome, and it is difficult to
understand, respectively, on the
contrary, it seems trivial and
obvious.
In the preparation for teaching some
of the less frequent motifs, based on
the teachers need to be curious, to
discover, to be playful and
constructive, the need for security,
comfort, contact, the desire for
autonomy, and the need to manage
time ...
The learning texts should be
effective; they should include
headlines and subheading in order to
be well arranged, and clearly
structured.
Study materials should respect the
hermeneutic approach.
Courses and learning materials
should be an intellectual challenge.
For a detailed description of the creation
of any online study materials see [3] or
[4].
2 RESEARCH AND ITS FINDINGS
In order to find the latest attitude of FIM
students to the study materials, the
authors made a survey in the winter
semester of 2011. The survey
concentrated on the eLearning form of
study with respect to the study materials.
A form of electronic questionnaire was
chosen for this research.
Overall, 484 students participated in the
survey. 66% were men and 34% women
(Fig. 2), which in fact reflects the nature
of the faculty since most respondents
were from the field of informatics:
applied informatics (AI), information
management (IM) and information and
knowledge management (IZM) - 360,
other respondents studied financial
management (FM 27), sport
management
(SM
17)
and
management of tourism (MCR 56)
see Fig. 3.
45
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
from 19 to 41. More than 85% of
respondents were under 25 years old.
Positive:
I am (generally) satisfied.
It would be good to introduce more
e-subjects.
It depends, but generally, I am
satisfied.
46
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Negative:
The Blackboard is not working
properly at the moment.
Some
e-materials
could
be
structured in a better way.
Study materials are rarely balanced.
Everyone works with the electronic
materials differently, both students
and teachers. The structure of
lectures is sometimes not clear,
interesting, following only the
printed version. As for the
presentations, rules of a good
presentation are often broken.
Moreover, some teachers do not
know how to work with the tools
available to them, which is again at
the expense of quality, not
mentioning the outdated material.
The quality of individual lectures of
individual courses sometimes differs
but most of them are OK.
I would welcome more examples or
step-by-step tutorials.
There could be fewer learning
materials but it is more important to
get students involved and motivated
in these materials.
In most cases students evaluate the
online
materials
positively,
but
sometimes with some reservations, such
as:
Yes and no some materials are too
concise, other are too extensive. Only in
a few cases the study materials are
balanced. But in most cases I am
satisfied.
An overwhelming majority of students
(96%) found the electronic texts easy to
47
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
replies (almost 50%) concerned their
accessibility. Students really appreciate
accessing the study materials any time
and anywhere, most preferably, from the
cosiness of their homes. Other positive
factors of the electronic materials are as
follows:
an easy access to study materials;
a
possibility
of
copying,
downloading and printing;
low/no costs of learning materials;
ecological approach;
well-written and structured;
a chance of self-study;
individual pace of study;
a chance to return to study materials
and complete ones learning`
self-tests, a chance to check ones
knowledge;
interaction;
no need to take notes during the
face-to-face lectures.
3 CONCLUSION
From the above described results of the
survey, it is clear that students welcome
an opportunity to work online although,
predominantly,
in
the
so-called
combined/blended/hybrid
form
of
learning. The reason is that they can
complete their knowledge and thus
finish their learning process, which they
did not manage to do during the face-toface classes. They are not forced to do it
immediately at school or in the library
because they can access the online
course from the cosiness of their homes
any time they feel like that. In addition,
if they did not understand anything
during the lecture, they can contact their
teacher online and ask him/her.
In the online course students mostly
exploit the study materials. Therefore,
tutors should pay careful attention to
48
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 43-49
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
If such a material is developed, then both
teaching and learning is a challenge.
Moreover, such written educational
materials might increase students
interest in online studies, respectively
his/her motivation for self-study.
4 ACKNOWLEDGEMENT
The paper is supported by the project
Excellence, no. 2216: The ICT
reflection within the cognitive processes
development .
5 REFERENCES
1.
2.
3.
4.
5.
49
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
The separation of the searched data from the
rest is an important task in data mining. Three
separation/classification
methods
are
presented. We use a singularity exponent in
classifiers that are based on distances of
patterns to a given (classified) pattern. The
approximation of so called probability
distribution mapping function of the
distribution of points from the viewpoint of
distances from a given point in the form of a
scaling exponent power of a distance is
presented together with a way how to state it.
Considering data as points in a metric space,
three methods are based on transformed
distances of neighbors of a given point in a
multidimensional space via functions that use
different estimates of scaling exponent.
Classifiers data separators utilizing
knowledge about explored data distribution in
a space and suggested expressions of the
scaling exponent are presented. Experimental
results on both synthetic and real-life data
show interesting behavior (classification
accuracy) of classifiers in comparison with
other well-known approaches.
KEYWORDS
Nearest neighbor; fractal set; multifractal; IINC
method; correlation dimension
1 INTRODUCTION
One task is that we have a large amount
of data after extensive search, but we are
interested only in their small portion and
we wish to separate them from the rest
with a minimal error. These tasks are also
rather common in particle physics where,
during measurements, an extensive
amount of data (events) is registered but
only their small portion contains
information which has not been
previously known. This separation task
can be generalized in such a way that
there are several classes of data in which
we need to sort them.
Often, in classification problems, the only
known fact is the learning set, i.e. the set
of points each of known class. The
problem is how to estimate the probability
to which class a query point x of the data
space belongs. The different approaches
to the classification can be divided into
parametric and nonparametric methods.
Parametric methods include neural
networks of different kinds [6], decision
trees or forests [2], also known as the
CART method, the support vector
machine SVM [1] and many more.
50
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Nonparametric methods are mostly based
on the Bayesian approach [6], [10], and
the k nearest neighbors (k-NN) method
[5], [6], [24].
Comparing
different
classification
methods applied to different data sets, one
can find a very interesting fact that
sometimes a simple approach outperforms
the other methods including those which
are very sophisticated. This is the case of
the 1-NN method and the k-NN method,
for example. These methods are
thoroughly
elaborated since Cover
and Hart [5], and many variants have
appeared up to now, see e.g. [8], [24].
Simple methods like 1-NN can be
improved by some preprocessing, in fact,
by some modification of the learning set.
Best known are bagging and bootstrap
approaches. Other approaches set some
weights to learning set samples or to
individual
features
(coordinates),
eventually modify features of individual
samples (move samples) to get better
results [8], [24]. Here we describe new
algorithms in a pure form without any
preprocessing.
Classifiers directly or indirectly take the
distribution of data points around a given
query point into account. To express the
distribution of points from the viewpoint
of distances from a given point, the
probability distribution mapping function
was introduced [16]. The approximation
of this function in the form of a suitable
power of the distance is presented. How
to state this power the distribution
mapping exponent is described. This
exponent is used for probability density
estimation in high-dimensional spaces
and for classification. It will be seen that a
core notion in the transformation
mentioned above is a slightly redefined
singularity or scaling exponent to fit the
notion of distance between points. The
scaling considered here is related to
51
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
3 BASIC TYPES OF CLASSIFIERS
Classifiers are based on several
principles:
A cut of distributions of individual
features (variables) of a pattern; this is
rather close to the Bayes classifier,
used in physics.
Bayes classifier. For each variable
empirical distribution densities (histograms) are estimated for each class.
For a query pattern with unknown
class, the corresponding density is
estimated from empirical or empiricbased histogram for each feature.
Then, for each class a product of these
densities is computed and the largest
product corresponds to the class to
which the pattern belongs.
Distance based classifiers. These
classifiers need the pattern space be a
metric space. Therefore,
some
transformation is often used and
metrics introduced. The simplest is
the well-known 1-NN (nearest
neighbor) classifier that associates the
class of its nearest neighbor to the
query pattern. Several (k) nearest
neighbors (k-NN) can be used;
sophisticated variants use a special,
even adaptive metrics.
Cluster-based classifiers try finding
clusters according to given criteria,
and in a cluster, a threshold is used for
classification as in Random Forests.
Kernel-based classifiers first identify
parameters of kernels and then
estimate a class according to common
densities.
GMDH classifiers and approximators
that are special forms of neural
networks where neurons have two
inputs only and each neurons
response is a full quadratic
polynomial of two input signals [22].
1
{number of pairs (i, j) : X i X j < r}
N2
.
In a more comprehensive form one can
write
C I ( r ) = Pr( X i X j < r ) .
Grassberger and Procaccia [11] have
shown that for small r the CI(r) grows like
power C I ( r ) ~ r and that correlation
exponent
can be taken as the most
useful measure of the local structure of
the strange attractor. This measure allows
distinguishing between deterministic
chaos and random noise.
The correlation integral can be rewritten
in the form
CI (r ) = limN
2
h(r X j X i ) ,
N ( N 1) 1i< j N
(1)
ln C I ( r )
ln r
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Thus, if the correlation integral as a
function of r is depicted in log log
coordinates, there appears a nearly
straight line with a slope equal to
correlation dimension v. This was shown
in [11] and the method based on this fact
is
called
Grassberger-Procaccias
estimator of v. There are other methods
for estimation of correlation dimension .
One of the most cited is Takenss (viz lit.
odkaz 27) estimator [27].
5 DISTRIBUTION MAPPING
FUNCTION
Here we single out
three notions
introduced or used in [14]-[15]. The
probability distribution mapping function
(DMF) is a mapping of the probability
distribution of points in n-dimensional
space to the distribution of points in onedimensional space of the distances. The
distribution density mapping function
(DDMF) is a one-dimensional analogy to
the probability density function. The
power approximation of the probability
distribution mapping function in the form
of (distance)q is introduced, where we call
the exponent q the distribution mapping
exponent (DME).
These notions are local, i.e. are related to
a particular (query) point. We show that
the distribution mapping exponent q is
something like a local value of the
correlation dimension according to
Grassberger and Procaccia, [11]. It can be
also viewed as the local dimension of the
attractor by Froehling [9] or singularity
eventually scaling exponent (exponent)
in the sense of Stanley and Melkin, [26].
To be more exact, let us introduce two
definitions
to study a probability
distribution of points (patterns) in the
neighborhood of a query point x in
n-dimensional Euclidean space En.
D ( x, r ) , where D(x, r) is a
r
D ( x, r )
rq
const
for r
0+ . The
53
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
appears here. Note the slope that is
q = 2.5.
54
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
6 DECOMPOSITION OF THE
CORRELATION INTEGRAL TO
LOCAL FUNCTIONS
The distribution mapping exponent
(DME) reminds one of the so-called
correlation dimension by Grassberger and
Procaccia [11], and corresponds to the
generally used definitions of power
scaling laws especially to singularity
exponent. It can be seen that the
correlation integral is a distribution
function of distances between all pairs of
points of the data points given. The
probability distribution mapping function
is a distribution function of the distances
from one fixed point x. In the case of
finite number of points N, there are N(N 1)/2 distances between pairs of points and
from them one can construct an empirical
correlation integral. Similarly, for each
point there are N - 1 distances and from
these N - 1 distances one can construct an
empirical
probability
distribution
mapping function. There are exactly N
such functions and the mean of these
functions gives the correlation integral.
This is also valid for N going to infinity.
Let us return to Eq. (1). In this section we
show that the correlation integral is the
mean of distribution mapping functions.
Theorem. [15] Let the correlation
integral, i.e. the probability distribution of
distances lij between all pairs of points
from the learning set, be CI(r) and let
D(xi, r), where lik is the distance of k-th
neighbor from point xi, be the distribution
mapping function corresponding to point
xi. Then, CI(r) is a mean value of D(xi, r):
N
C I ( r ) = lim
1
N N
D ( xi , r ) .
(2)
i =1
Proof.
Let h(x) be Heaviside step function. Then,
the correlation integral is
N N 1
CI (r ) = lim
1
N N ( N 1)
h(r lij )
i =1 j =1
and also
N 1
C I ( r ) = lim
1
N N
i =1
1
N 1
h ( r lij )
j =1
55
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Pr (T ( x) = c | T ( xi ) = c ) = K
1
.
ri q
A question arises where do these
considerations come from? The answer is
simple:
Let us consider a (fixed) query point x.
Construct a ball B(x, r) with center x and
radius r. Let the volume of this ball be
V(x, r) and it holds V ( x, r ) = S n r n . Sn is a
constant
dependent
on
space
dimensionality n. Let there be i points in
B(x, ri). We are interested in the density
of points in the neighborhood of point x.
i
i
It
is
p ( x, ri ) =
=
.
V ( x, ri ) Kri n
Considering this density constant (in the
sense that we get the same density for
different values of i, i.e. for different radii
ri), radii ri would grow proportionally to
the n-th root of i, ri n i . But as
shown above the space can have
(locally
or
globally)
effective
dimensionality q lesser than n. Thus, we
should compute density of points using
the volume of q-dimensional ball using
i
i
formula p( x, ri ) =
=
, where
Vq ( x, ri ) S q ri q
Pr (T ( x) = c | T ( xi ) = c ) = K
1 / ri q
xi U c
.
(3)
Note that the sum goes over indices i for
which the corresponding samples of the
learning set are of class c, c = 1, 2, C,
where C is the number of classes. It can
be seen that any change of distance ri of
1 / ri q
p ( x | c) =
xi U c
N
i =1
1 / ri
(4)
p (c | x) = lim
1 / ri
xi U c
N
i =1
(4)
1 / ri
56
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
1 / ri q
lim
p (c | x ) =
x U c
N
lim
i =1
1 / ri q
1
q
xi U c ri
is faster the larger DME q is. Usually, for
multivariate real-life data the DME is also
large (and the correlation dimension as
well).
d ( x, ri q , c) / ri q . For
lim
i =1
9 THREE METHODS
d ( x, ri q , c) / ri q = p(c | x) lim
N
i =1
1 / ri q
(6)
because d(x, ri , c) = d(x, z, c) = p(c|x) for
all i (uniform distribution has a constant
density).
Given the learning set, we have the space
around point x sampled by the
individual points of the learning set. Let
pc(ri) be an a-posteriori probability that
point i at distance ri from the query point
x is of the class c. Then, pc(ri) is equal to
1 if point xi is of class c and pc(ri) is
equal to zero, if not, i.e. if the point is of
the other class. Then, the particular
q
realization
of
p (c | x )
i =1
1 / ri q
is
sum
xi U c
p(c | x) lim
i =1
1 / ri q = lim
1 / ri q
xi U c
57
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
0.5 4
0.5 2
0.5
A UC
0.4 8
0.4 6
0.4 4
0.4 2
0.4
0
10
15
20
25
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
is rather a small difference, only 1.93 %,
showing that error in correlation
dimension estimate need not be critical.
Table 1.
Parameters of DME distribution for different data sets
and color notation for Fig. 3. Data sets are from UCI
MLR.
mean
Dimen mean sigma/ DME/
sion
DME Mean dim.
Data
Color
Higgs
Red
23
1.75 0.098 0.076
German
Heart
Adult
RKB
Ionosphere
Aquam.
Violet
Green
Blue
Coral
20
13
14
10
33
2.71
2.42
5.28
4.94
2.06
0.079
0.074
0.158
0.185
0.139
0.136
0.186
0.377
0.495
0.062
Pr (T ( x ) = c | T ( x i ) = c ) = K
p (c | x ) = K
1 / ri q
xi U c
1/ i
xU c
59
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
This approach is very simple and very
effective; no estimate of the singularity
(fractal) exponent is needed.
(7)
If x1, x2, xN are independent and
identically distributed samples of a
random variable with probability density
function
f,
then kernel
density
approximation of f is
f h ( x) =
1
Nh
K
i =1
x xi
h
60
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
1
(s) =
< ,
s
t =1 t
where is Riemann'
s zeta function.
10 GENERALIZATION
11 DISTRIBUTION MAPPING
EXPONENT ESTIMATION
ri q = ki , i = 1, 2, ..., Nc,
c = 0 or 1,
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Legend:
62
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
12.2 Data from the Machine Learning
Repository
Data sets intended just for running with a
classifier were prepared by Paredes and
Vidal and are available on the Internet.
We used all data sets of this corpus. Each
task consists of 50 pairs of training and
testing sets corresponding to 50fold cross
validation. For DNA data, Letter data
(Letter recognition), and Satimage
(Statlog Landsat Satellite) [21], single
partitioning into a training and testing set
according to the specification in [21] was
used. We also added the popular Iris data
set with tenfold cross validation. The
results obtained by the QCregre approach,
in comparison with data published in [24],
are summarized in Table 2. Each row of
the table corresponds to one task from
[21]. For tasks where the data are not
available, only the results for 1-NN
method with L2 metrics were amended. In
the QCregre method, we used a rather
complex strategy of robust modification
of linear regression, as described above.
The interesting point is the experiment
with the simplest strategy of using a half
of the samples nearest to the query point.
For some tasks we obtained very good
results.
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
13 CONCLUSION
15 REFERENCES
1.
64
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
2.
3.
4.
5.
65
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 50-66
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
(Review) Nature Vol. 335, 29 Sept. 1988, pp.
405-409.
27. Takens, F.: On the Numerical Determination
of the Dimension of the Attractor. In:
Dynamical Systems and Bifurcations. Lecture
Notes in Mathematics, vol. 1125, pp. 99 106.
Springer, Berlin (1985)
28. Zipf, G.K.: The Psycho-Biology of Language.
An Introduction to Dynamic Philology. The
MIT
Press,
1968.
(Eventually:
http://en.wikipedia.org/wiki/Zipf'
s_law)
66
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
KEYWORDS
67
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
68
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
69
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
70
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
71
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
72
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
3. IMPROVING
RELIABILITY OF TREE
COMMUNICATIONS
To realize our goal, we keep the same
principles of this solution and propose
two extensions called Sol1 and Sol2. Our
key idea consists in (1) determining
a priori one or several backups per
parent and (2) taking into account these
potential parents in the coloring
SERENA. Indeed, we extend the set of
conflicting nodes with the backup and
any other node that may belong to one of
the 5 sets previously defined if the
parent was replaced by the backup.
Consequently, the parent replacement by
a predetermined backup does not induce
any color conflict and the initial coloring
remains valid. Depending on the number
of backups, we propose two solutions.
3.1. First Solution (Sol1): Tree Coloring
with One Backup per Parent
73
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
The second solution Sol2 consists in 3hop coloring where the set of conflicting
nodes of each node contains its
neighbors up to 3-hop. Consequently, a
node has as potential parent backup all
its 1-hop neighbors. The rule R2 is taken
into account with the initial parent.
However, it is impossible to define an
order between the color of a node and its
parent backups which are in fact its 1hop neighbors. So, Sol2 implies that
there is no order between the color of a
node and the color of its parent backup.
Remark 3: Notice that applying this
solution, routing loops are possible.
However, we address this issue by one
of the two following hints:
We assume that the routing protocol
handles the loops with a detection
and resolution technique.
Or, when a node chooses a parent
backup, it chooses only a node that
has a smaller identifier.
74
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
4. COMPARATIVE STUDY
In this section, we present a table that
compares
qualitatively
the
aforementioned two solutions with the
basic solution SERENA. We then
present the simulation results.
75
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
(a) Sol1
(b) Sol2
Fig. 5. Number of colors in cases density=10
with/without Rule R2.
76
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
77
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
78
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
6. REFERENCES
[1] Minet, P.; Mahfoudh, S. SERENA:
SchEduling RoutEr Nodes Activity in
wireless ad hoc and sensor networks. In
Proceedings of IWCMC 2008, IEEE
International Wireless Communications
and Mobile Computing Conference,
Crete Island, Greece, August 2008.
[2] Xu N., A Survey of Sensor Network
Applications, IEEE Communications
Magazine, vol. 40, 2002.
[3]
Garcia-Hernandez
C.F.;
Ibarguengoytia-Gonzalez P.H.; GarciaHernandez J.; Perez-Diaz J. A., Wireless
Sensor Networks and Applications: a
Survey,
International
Journal
of
Computer Science and Network
Security, IJCSNS07, vol. 3, pp. 264273, 2007.
[4] Heinzelman W.R.; Kulik, J.;
Balakrishnan H., Adaptive protocols for
information dissemination in wireless
sensor networks, 5th ACM/IEEE
conference on Mobile computing and
networking,
MobiCom99,
Seattle,
Washington, United States, August
1999.
[5] Heinzelman W.; Chandrakasan A.;
Balakrishnan H.; Energy efficient
communication protocol for wireless
sensor
networks,
International
Conference System Sciences, Hawaii,
January 2000.
[6] Senouci S.M.; Pujolle G., Energy
efficient routing in wireless ad hoc
networks,
IEEE
International
Conference on Communications ICC04,
Paris, France, June 2004.
[7] Levis P.; Clausen T.; Hui J.;
Gnawalli O.; Ko J., The Trickle
Algorithm, RFC 6206, MARS 2011,
http://tools.ietf.org/html/rfc6206
[8] Carle J.; Simplot-Ryl D., EnergyEfficient Area Monitoring for Sensor
79
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 67-80
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
80
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
The tremendous and rapid development
in sensors technology allowed their
application in various fields requiring
monitoring, such as, transportations, rare
species surveillance, agriculture, military
activities, medical field, etc. Due to their
intrinsic constraints and limitations,
several dedicated MAC protocols have
been designed for wireless sensor
networks and whose main objectives are
bandwidth optimization while keeping
very low energy consumption. One of
the newest proposals is ANT which is
emerging as a widely used MAC
protocol for wellness and sports sensor
devices.
In this paper, we give a comprehensive
overview of ANT/ANT+ and a
comparison between ANT and two
established standards: Bluetooth Low
Energy (BLE) and 802.15.4/4a. We
detail also some practical experiments
we conducted to evaluate some
characteristics of ANT.
KEYWORDS
ANT, 802.15.4, BLE, WSN, MAC protocol.
1 INTRODUCTION
Wireless Sensor Networks (WSNs) have
been proposed for many applications
81
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Networks with low bandwidth (LRWPAN).
ANT+ adds network management layers
which enables interoperability, defines
data structure, provides definition of
node profile and manages network keys.
In this paper, we propose to present a
theoretical overview of ANT and ANT+
and some practical tests conducted with
real material.
This paper is organized as follows: In
the beginning, we present the basic
principles of ANT protocol and ANT+.
Then, we present
a
comparison
between ANT, BLE [24] [25] [26] [27]
[28] and 802.15.4/4a. Next, we present a
practical study of ANT. In the end, a
conclusion comes to take stock of our
study and discusses the prospects for
research on which we move.
2 PRESENTATION OF ANT
ANT is a protocol designed for wireless
personal area networks (WPAN) with
low power consumption, low processing
power and low baud rate. The small size
of the ANT protocol stacks allows the
integration into components such as:
PDAs, watches, phones, Smartphone,
etc. It operates in the 2.4GHz ISM
(Industrial, Scientific and Medical radio
bands) band and allows for a flow baud
rate of 1Mbps with transmission
duration less or equal than to 150
microseconds / frame for 8bytes of data.
ANT provides a management of
Physical, Data link, Network and
Transport OSI layers (Fig 1). It's a very
compact stack requiring minim material
resources.
82
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
2.2 ANT channels
2.2.1 Parameters
The ANT protocol uses 2.4 GHz
bandwidth and divides it into 125
channels of width equal to 1 MHz and a
baud rate of up to 1 Mbps.
In order to create a connection between
two nodes, they must have the same
channel configuration. Then the slave
needs to know several parameters we
cite here after [21].
a- Network
ANT can set up several private and
public networks. In order to establish a
communication between two nodes, they
must be in the same network, which
enables a network service that can be
shared with multiple source nodes. On
the other hand, we can establish a private
network to ensure confidentiality and
restrict access. ANT node can participate
in several networks since the channels
can be assigned independently to
multiple networks. This parameter
contains two information: Network
number and Network key.
Network number is coded on 8 bits and
varies between 0 and a maximum value
fixed during the implementation of
ANT. The default value of the network
number is 0.
The Network Key is coded on 8 bits and
configured by the application. Each
network must have a network key. The
key and the network number allow the
deployment of networks with a varying
level of security and access control.
83
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Table 1. Transmission types and characteristics
Broadcast
Acknowledged
Burst
Forward direction
-Send the data in each time period.
-The data are not acknowledged.
-If the higher layer does not send
new data, ANT retransmits the last
data received by the HC layer.
-The frames are sent only once.
-The master may decide to send a
frame with the request of
acknowledgement during the step of
establishing a two-way connection.
-The HC layer can choose to send
all data frames with a request of
acknowledgement or it mixes the
broadcast transmission and data
transmission with acknowledgment.
-Communication by request of
acknowledgment of data uses more
bandwidth and consumes more
energy.
-This type of transmission is
important for critical data.
-There
is
no
automatic
retransmission of data packets
without acknowledgment.
-If the ANT protocol doesnt have a
new data in the interval time, the
last frame is retransmitted.
-This mechanism is used by the
master to transmit a large amount of
data to the slave.
-The burst transmission begins in
the transmission period and it is
expected to send a set of frames that
must be acknowledged by the slave.
-All lost frames are automatically
retransmitted.
-The transmission time is not
limited
- Data rate is up to 20kbps.
-The channel that uses this type of
transmission has priority over all
other channels
Reverse direction
-Send at the end of the period
of transmission of the master.
-Frames sent by this type of
communication are never
acknowledged.
84
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
d- Channel type
This parameter defines the type of
transmission that will occur in the
channel. It is coded on 8 bits and can
define eight types of communication.
e- Channel period
This is the basic frequency used by the
master for transmission of data. By
default, this parameter represents the
interval time between two successive
transmissions of data transmission with
broadcast transmission (Figure 3). The
frequency of data frames in a channel
varies between 0.5Hz and 200Hz and the
default frequency is equal to 4Hz. The
channel period is coded on 16 bits.
85
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
2.2.3 Establishment of a channel
If two nodes want to communicate, they
must undergo the following process:
First, the master sets the type of
network, the frequency, the message rate
and the channel ID. Secondly, it
establishes the channel by transmitting
the channel ID with the data in a
particular time interval (channel period).
ANT channel will be maintained
indefinitely at the given time interval
and frame baud rate. The higher layers
of the master will eventually provide
new data to the ANT protocol for
continuous transmission. Thirdly, the
slave opens its channel to search for a
specific master or to search a subset of
masters (the type of research depending
on the configuration of slave channel).
When the slave receives the first
message, it selects the type of network,
operating frequency, and frame baud rate
of the master. With this information it
can synchronize with the master and
begin receive data [11][21][17].
In ANT, after the synchronization step,
in the ideal case, we don't have a
transmission failure if the slave node is
in the transmission ranges of the master
nodes. If the slave node is connected to
many master nodes, it switches between
channels to receive the information.
2.3 Device pairing
Pairing action consists of establishing a
relationship between two nodes that
wish to communicate together (master
and slave). This relationship may be
permanent,
semi-permanent
or
transitory.
86
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
87
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Table 2. Comparison between ANT and 802.15.4/4a
Frequency band
Channel
numbers
Max Packet
Play load
(bytes)
802.15.4
3
frequency:
868, 915
and
2400MHz
27
802.15.4a
ANT
BLE
3 frequency:
Sub-Ghz, low
band
and high band
2400MHz
2400MHz
- 16 with
UWB
- 14 with chirp
signal
8 channels
40
20
30
100
Range (meters)
100
Min
configuration
Transceiver
Node Type
Battery life
6 months
Synchronization
Mode of
transmission
Transmit only or
transceiver
We have two types
of node: central
node (for collecting
the information
from simple nodes )
and simple node
(transmit the
detected
information to the
central
node)
Up to 3 years
The transmission of
the parameters of
synchronization in
all transmitted
packets
Best effort
Transceiver
Single mode
(BLE) or dual
mode
(Bluetooth &
BLE)
1 year
By the master
(AFH)
advertising
event &
connection
event
88
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
The USB2 stick is characterized by
[13]:
It uses the ISM band and uses the
frequency from
2403 to
2480MHz,
It
supports
the
Broadcast,
acknowledged, or burst data
transmissions,
Its minimum message rate per
channel is 0.5 Hz,
Its burst transfer rate is limited by
20kbps,
It uses up to 8 channels,
It has the scanning mode,
It has up to 3 public, managed
and/or private network keys.
These USB sticks will work in the
2466MHz frequency, the frequency of
transmission is equal to 4 Hz.
We have configured the ANT USB to
use the 2466MHz frequency to transmit
data with a transmission frequency equal
to 4 Hz, these values are the default
values of the ANT product.
We have transmitted 1000Kbytes of data
with burst transmission. Our goal is to
test and verify the correct functionality
of ANT protocol.
The Figures 6 and 7 will help us to find
the center frequency and the spectrum
large that are used by the node. The
Figure 6 represents the signal
transmitted by the ANT USB and
detected by the oscilloscope. This
spectrum is periodic and shows the
effect of this transmission in the
networks. The device uses the full
bandwidth from 2464 to 2468MHz
frequencies to transmit the data and we
have a peak in 2466MHz, so ANT USB
uses a bandwidth with 1 MHz of width
and a center
2466MHz.
frequency
equal
to
89
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
we use burst transmission ANT uses all
the bandwidth and stops other
transmission.
1.
91
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 81-92
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
13. ANT usb2 stick data sheet, taken from
http://www.thisisant.com/
14. ANT+ Devices profiles, ANT presentation,
taken from http://www.thisisant.com/
15. The opportunities, ANT presentation, taken
from http://www.thisisant.com/
16. ANT+ Alliance - an overview, ANT
presentation,
taken
from
http://www.thisisant.com/
17. Proximity search, Application notes, taken
from http://www.thisisant.com/
18. Implementing A Receiver for Transmit-Only
ANT Devices, Application notes, taken
from http://www.thisisant.com/
19. Interfacing with ANT General Purpose
Chipsets and Modules, Applica-tion notes,
taken from http://www.thisisant.com/
20. ULP - advantage ANT, ANT presentation ,
taken from http://www.thisisant.com/
21. ANT Message Protocol and Usage,
Application
notes,
taken
from
http://www.thisisant.com
22. WiSpy
:
Spectrum
Analysis,
http://www.metageek.net/products/wi-spy/
23. kit de developpement : ANTAP2DK1
(Nordic p/n nRF24AP2-DK1), Dy-nastream
Innovations
Inc
,http://www.thisisant.com/pages/products/an
t-dev-kit
24. Comparing
Low
Power
Wireless
Technologies, http://www.digikey.com ,
consulted in march 16, 2013
25. Specification:
Adopted
Documents
https://www.bluetooth.org/Technical/Specifi
cations/adopted.htm, consulted in march 16,
213
26. Bluetooth
low
energy,
http://www.litepoint.com/whitepaper/Blueto
oth%20Low%20Energy_WhitePaper.pdf
27. Specification of the Bluetooth System, Covered
Core Package, Version: 4.0; The Bluetooth
Special Interest Group: Kirkland, WA, USA,
2010.
28. Bluetooth low energy wireless technology
backgrounder,
http://www.nordicsemi.com/jpn/nordic/content_d
ownload/3045/40678/version/1/file/Bluetooth_le
_technology_backgrounder_4.pdf
92
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ghamri@ensiie.fr
ABSTRACT
In the past few years, the smart grids have
attracted more and more attention. The fact
that these systems are based on the
pervasive use of new Information and
Communications Technologies (ICTs) opens
new perspectives in an aim to improve the
electrical grid performances. One of the
most commonly deployed technologies in
such a real environment is the wireless
sensor networks (WSNs) technology. WSNs
are thus used to monitor and control the
electrical grid components. In this paper we
will give an overview on WSNs applications
and challenges for smart grids. One
promising research topic, when speaking
about WSNs in smart grids, is the cosimulation which allows studying and
evaluating any new technique in a near real
grid behavior. Hence, in this paper, we will
also present and discuss the architectures,
the advantages and the drawbacks of the
most important smart grids co-simulation
platforms used to combine the network
behavior with the power grid systems
models.
KEYWORDS
Co-simulation, network simulation, electric
power simulation, smart grid, wireless
sensor networks.
1 INTRODUCTION
The huge development of consumer
electronics (PDAs, cell phones, laptops,
93
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
components, the grid self-healing
improving and adjusting automatically
the electricity distribution. All these
functionalities become feasible thanks to
the advent of Information and
Communications Technologies (ICTs).
ICTs
guarantee
the
efficient
communication
and
coordination
between all power grid participants
(suppliers, consumers, equipments, etc.).
The power grids equipped with ICTs are
called smart grids.
One
of
the
most
appropriate
technologies for the smart grids control
and monitoring is the wireless sensor
networks (WSNs) technology [2].
Hence, WSNs are deployed all over the
electricity grid parts (generation plants,
substations, power lines, consumers
side, etc.). Then, in this paper, we
propose to present the different
challenges to be raised by WSNs to meet
the smart grids requirements. We also
undertake a study on how a WSN based
smart grid solution is validated through
what we call the co-simulation process
(i.e. combining the network behavior
with the power grid systems models in
order to assess the performance in both
network communication side and
efficient power management side).
The rest of this paper is organized as
follows. In the section II, we focus on
the different WSNs applications in the
smart grids [3] [4] and we present the
new challenges facing the WSNs to
satisfy the smart grids requirements. In
section III, we present the need of using
co
simulation
to evaluate the
performance of WSNs communication
behavior in the context of the smart grids
by considering the characteristics of the
electrical components. Different cosimulation platforms will be presented
with a discussion on the advantages and
94
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Temperature sensors for high
voltages lines;
etc.
2.1 Applications of WSNs in Smart
Grids
The electrical power grid can be
partitioned into three main conceptual
segments: the energy generation, the
power transmission & electricity
distribution, and the consumption side.
1) WSNs for energy generation segment:
In the conventional power grids, energy
generators are monitored by a limited
number of high-cost wired sensors
deployed in some critical locations. One
of the main objectives of smart grids is
the usage of intensive renewable energy
sources (wind, photovoltaic cells, etc.)
and their ease of integration in the
generation process. These generation
sources are generally situated in remote
and hard to reach areas within harsh
environment
which
requires
a
continuous monitoring of such zones via
low-cost sensors. Then WSNs represent
the ideal technology to monitor such
energy generators [5] [6] [7] [8].
2) WSNs for transmission and
distribution segment: The transmission
system
(towers,
overhead
and
underground power lines) and/or
distribution
system
(substations,
transformers and wiring to the
consumers) failure may cause blackouts
and may even present a danger to the
public
security.
Hence,
these
components must be monitored in near
real time to troubleshoot any problem. In
addition to the continuous monitoring,
this segment must be protected against
external attacks due to the ease of access
to any component. Once again, WSNs
provide promising solutions for the
95
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
2) Reliability and robustness of WSNs:
WSNs failures arising in power grids
may create serious problems. In fact, the
environmental conditions (interferences,
humidity, dust and vibration) may cause
the sensor nodes defection and affect the
network topology. The problems in
WSNs can be caused by the sensor
nodes constraints (limited battery and
CPU, and reduced memory capacity).
3) Real time constraints: Considering the
real time requirements in smart grids
which must react very quickly to a
failure situation, the communication
network must transmit the data in
a near real time to correct the anomaly
arising in the network and restore the
normal functioning of the electricity
grid.
After giving an overview of WSNs
applications in smart grids and some
challenges related to the sensor networks
deployment, the following section will
be dedicated to the description of smart
grids co-simulation platforms.
3 SMART GRIDS CO-SIMULATION
The electric power grid is among the
most critical infrastructures for a nation.
In
fact,
this
infrastructure
is
interdependent
with
other
vital
infrastructures and disruptions in the
power
grid
can
have
severe
consequences for other critical systems
such as the natural gas and the water
supply systems. To avoid any disruption
in that grid, any new mechanism must be
well studied and experimented before
deploying it in the real infrastructure
which is the role of simulation.
96
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
functionalities in the other as a module.
This can avoid the problem of
synchronization and gives a perfect
emulation of the real world.
3.2 Smart
Platforms
Grid
Co-Simulation
97
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
2) The ns-2/OpenDSS co-simulator: In
2010, Godfrey et al. proposed a cosimulation platform [17] using the ns-2
and the OpenDSS (Open Distribution
System Simulator). They studied the
impact of cloud transient (solar
ramping) phenomenon, when clouds
pass in front of photovoltaic (PV)
panels, on the voltages variation. The
communication network is needed to
discharge distributed storage batteries to
compensate the voltage reduction.
OpenDSS is a simulation tool for electric
utility distribution systems. It provides
energy analysis tools for power delivery
and is designed to simulate discrete
events.
In the co-simulation, when the storage
controller detects the solar ramping, it
will attempt to dispatch the storage units
to maintain a smooth voltage. OpenDSS
provides the time of the PV ramp event,
the topology of the storage batteries and
the power load profile for the storage
units. ns-2 simulates the arrival of
controller messages to each storage unit,
which
is
assumed
to
respond
immediately. The arrival times are sorted
and fed back to the OpenDSS engine as a
script. The system voltage can be
computed at any time according to the
storage units reaction determined by
message arrival times.
The PV panels form a WSN that
communicates using the IEEE 802.11
standard. The model used in the ns-2 is
configured to simulate the IEEE 802.11
operating in 915 MHz Industrial,
Scientific and Medical (ISM) band.
Figure 1 gives an overview of the data
flow between the OpenDSS and the ns-2
during the co-simulation. The messages
sending time and the storage units
coordinates are transmitted to the ns-2
module, which computes the messages
arrival times. These times are merged
98
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
scheduler. In the ns-2 side, a new class is
implemented as a PSLF interface that
initializes the power system dynamic
simulation rounds in ns-2. This class
collects power system data from PSLF
after each dynamic round and saves it in
ns-2 for future use by other network
components. In PSLF, a new stand-alone
dynamic model implemented and
attached to the simulated system. It has
two functionalities; the first is reporting
back the power system state data to the
ns-2 after a demand. The second
functionality is suspending PSLFs
simulation after each dynamic round and
waiting for ns-2 commands to continue.
The second function is crucial for
scheduling co-simulation by a global
scheduler.
Besides the new interface, other
components are implemented in ns-2 to
facilitate the power system applications
management.
Slave
agents
are
implemented in application layer that
stand agents interacting with power
system
and
communication
infrastructure (intelligent electronic
devises or digital relays). A master agent
is also implemented in the same layer to
coordinate the slave agents tasks. This
agent can stand for a central control
module or an operation center. Classic
transport protocols are modified to
support power data transmission. After
describing their co-simulation platform,
the authors studied an agent-based
supervisory backup relay protection
scheme on this platform. This case study
allows the co-simulation framework and
the protection scheme validation.
4)
The
ns-2/Modelica
based-on
simulator: In [19], the co-simulation is
used to evaluate a new communication
network for smart grids named
PowerNet. This network includes
interoperable heterogeneous networks
99
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
The simulator choice is explained by the
fact that ns-2 provides tools to model
different networks (wired, wireless, local
and wide-area networks) and it can
evolve easily. For the power side,
Modelica is an object-oriented modeling
language for large-scale complex
physical systems.
5) Other co-simulation libraries: In
addition to the co-simulators presented
above, many other tools are available for
building new co-simulation platforms,
such as the Toolkit for HYbrid Modeling
of Electric power systems (THYME) [20]
which is a C++ library for building
simulator that integrates power system
modules
with
existing
network
simulators (OMNET++, ns-2). THYME
applications include: the study of control
system based on demand/response, the
impact of specific communication
technologies and intelligent sensors on
situational awareness, control or both,
and the design and analysis of wide area
diagnosis technologies that depend
critically on communication networks.
This tool includes an extensive set of
modules representing the power system
elements and behaviors.
THYME is not a simulator but a set of
libraries integrated in the adevs
simulator. As it is presented in Figure 2,
THYME provides the power model and
the power flow data, while adevs gives
the simulation environment (algorithms
and interfaces). The unlimited set of
network simulators that can be
integrated with THYME offer the
possibility
of
evaluating
WSNs
performance deployed in different
environments and with different network
sizes.
Table
1
summarizes
the
advantages/drawbacks of each cosimulation platform described above.
Before choosing which co-simulation
platform is adapted to a given smart grid
application using WSNs as ICT, we first
have to determine the requirements of
such an application based on the
advantages and drawbacks of each cosimulation platform.
For instance, if we choose ns-2/adevs
co-simulation,
we
avoid
the
synchronization problem given that
adevs is integrated as a module in ns-2,
but the complexity of the power
behavior
formalization
decreases
drastically the performance of such a
platform. Moreover, in the case of large
scale sensor nodes deployment, the ns-2
simulator is not appropriate.
In OpenDSS/ns-2 co-simulation, the
scripts writing is facilitated by the
separate simulators running. But the
intercommunication technique is not
defined. In the ns-2/PSFL platform, the
use of new interfaces in both simulators
facilitates the communication of
simulators in spite of the large number
of modules to implement.
The ns-2/Modelica platform can
integrate WSNs with existing networks
technologies
-such
as
WIMAX,
WHANs, WPANs, etc.- which reduces
the deployment cost. Finally, cosimulation with THYME offers a large
100
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
number
of
WSNs
deployment
opportunities using different network
simulators.
Advantages
Drawbacks
ns-2/ Modelica
THYME/ OMNET++
or NS-2
NS2/adevs
NS/ OpenDSS
NS2/PSLF
4 CONCLUSION
Due to the importance of the power grids
and the impact of their failure on the
Human life, new researches attempt to
modernize and smarten these grids so as
to avoid all the problems experienced
with the conventional electric power
systems. The so called smart grid is a
combination of the existing power grid
with
new
Information
and
Communications Technology
(ICT).
In this paper we presented the
application of a technology extensively
used to monitor and control the smart
grids behavior: the wireless sensor
networks
technology.
We
then
enumerated the different challenges that
should be met by the WSNs to satisfy
the smart grids requirements.
2.
101
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 93-102
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Networks.
IEEE
Communications
Magazine, pp.102114 ( 2002).
3. Erol-Kantarci, M. H., Mouftah, T.:
Pervasive Energy Management for the
Smart Grid: Towards a Low carbon
Economy.
Pervasive
Communications
Handbook, Eds. S. I. A. Shah, M. Ilyas, H.
T. Mouftah, CRC Press (2011).
4. Isaac, S. J., Hancke, G. P.: A Survey of
Wireless Sensor Network Applications from
a Power Utilitys Distribution Perspective.
AFRICON (2011).
5. Lu, B., Habetler, T.G., Harley, R.G.,
Gutierrez, J.A., Durocher, D.B.: Energy
evaluation goes wireless. IEEE Industry
Applications Magazine, vol.13, no.2, pp.1723 (2007).
6. Shen, L., Wang, H., Duan, X., Li, X.:
Application of Wireless Sensor Networks in
the Prediction of Wind Power Generation.
4th International Conference on Wireless
Communications, Networking and Mobile
Computing, vol., no., pp.1-4 (2008).
7. Ipakchi, A., Albuyeh, F.: Grid of the future.
IEEE Power and Energy Magazine, vol.7,
no.2, pp.52-62 (2009).
8. Al-Anbagi, I., Mouftah, H. T., ErolKantarci, M.: Design of a Delay-Sensitive
WSN for Wind Generation Monitoring in
The Smart Grid. IEEE Canadian Conference
on Electrical and Computer Engineering
(CCECE), Niagara Falls, ON, Canada
(2011).
9. Leon, R. A., Vittal, V., Manimaran, G.:
Application of Sensor Network for Secure
Electric Energy Infrastructure. IEEE
Transactions on Power Delivery, vol.22,
no.2, pp.1021-1028 (2007).
10. Lim, Y., Kim, H.M., Kang, S. A.: Design of
Wireless Sensor Networks for a Power
Quality Monitoring System. Sensors 2010;
10(11):9712-9725 (2010).
11. Ullo, S., Vaccaro, A., Velotto, G.: The role
of pervasive and cooperative Sensor
Networks in Smart Grids communication.
15th IEEE Mediterranean Electro-technical
Conference
(MELECON),
pp.443-447
(2010).
12. Yang, Y., Lambert, F., Divan, D.: A Survey
on Technologies for Implementing Sensor
Networks for Power Delivery Systems.
IEEE Power Engineering Society General
Meeting, vol., no., pp.1-8 (2007).
14.
15.
16.
17.
18.
19.
20.
102
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
ABSTRACT
1 INTRODUCTION
KEYWORDS
Energy Model; Fault-Tolerance; Markov
Chain; Mobile Robot; Wireless Sensor
Networks.
103
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
appropriate robot (a maintainer robot). In
the two distributed algorithms, the
management responsibility is distributed
over the robots, and each robot operates
as both a manager and a maintainer. In
fact, in the fixed distributed algorithm,
the area covered by the sensor nodes is
divided into equal-size subareas. Each
robot is assigned to a particular subarea
as both a manager and a maintainer
robot. Using this algorithm, a sensor
node selects its manager robot based on
its position in the network. If a manager
robot receives a failure report, it moves
to the failure location and replaces the
failed node. In the dynamic distributed
algorithm, the manager robot is selected
as the closest robot, and the failure is
handled by the manager robot that
received the report.
In their study, the authors of [1] have
shown that the centralized manager
algorithm has a low message overhead
and the smallest average robot traveling
distance per failure among the three
algorithms.
In this paper, we present a first strategy,
called Centralized Maintenance Strategy
(CMS), to repair failures in the WSNs
based on the Centralized Manager
Algorithm [1]. We detail our
improvements on the original algorithm
in Section 3.
However, failure handling according to
CMS still presents some drawbacks in
WSNs. In fact, the connectivity and
coverage of the network can be greatly
affected if the failure involves a critical
node during the repair time of nodes. In
this paper, we focus on the improvement
of the CMS. Our goal is to provide a
fault-tolerant WSN for real-time
applications. Our idea is to anticipate the
sensor failures. To reach this objective,
we need an energy model to predict the
energy consumption of sensor node. In
104
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
spent for radio communications by the
sensor radio interface, denoted in this
, and the sensing energy
paper by
(the energy it costs to take
measurements) denoted by
.
We focus first on the evaluation of
.
We claim that the status of the MAC
layer of the radio interface of a sensor
node can be modeled by a Markov
Chain. The states of this Markov Chain
depend on the MAC layer used by the
sensor node. For example, if we use the
IEEE 802.11 MAC layer we have three
modes of operation (states): Transmit,
Receive and Idle. Each state corresponds
to a different power consumption level
[8].
: node is transmitting
- State 1
.
a frame with transmission power
- State 2
: node is receiving a
frame with reception power
.
- State 3
: even when no messages
are being transmitted over the medium,
the node stays idle and listening the
.
medium with idle power
If the sensor node uses IEEE 802.15.4 or
SMAC as a MAC protocol, we have a
fourth state:
- State 4
: The radio is turned off,
and the node is not capable to detect
signals. We suppose that the node uses a
in this state which is largely
power
smaller than in any other state.
In conclusion, we model any MAC layer
of a sensor node as a discrete time
Markov Chain with four states:
,
,
and
.
the random variable related
We note
to the state of the MAC layer at time .
We choose a time step (unit) such that
the duration of any action (e.g;
transmission/reception of a frame) is a
multiple of this time step and we
suppose that all state transitions occur at
the beginning of the time step.
The notation
means that the
MAC layer is in state at time step .
Let
be the probability that a node in
state will enter in state at the next
transition.
(1)
Then, we note by
the transition
probability matrix (the element of row
column represents the probability ).
be the
Let
probability vector with
represents the probability that the MAC
layer is in state
at time step
(
).
Then, we have:
(2)
And:
(3)
(4)
the energy consumed by
Let
radio communication at the
transition given that the node is in state
at time step :
(5)
Where:
is obtained from equation 3 and
(the element of column of the vector
) is the energy consumed to transit
from state to state .
105
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
(6)
Where:
is the length in seconds of a
time step.
We note
the energy consumed
for radio communication by a node at
transition:
the
(7)
(8)
In addition to radio communication
energy, the sensor consumes an amount
of energy for event sensing. We note by
the average consumption
energy in sensing mode (to take
measurements).
(9)
Where:
: is the average rate of events detected
by sensor node, : is the sensing mean
time of an event and
: is the
energy consumed by node in sensing
mode (the energy it costs to take one
measurement).
the sensing energy
We note
consumed by the sensor until the timestep :
(10)
(11)
3 COORDINATION STRATEGIES
To coordinate the movement of
maintainer robots upon occurred failures
in the WSNs, [1] has proposed a
centralized manager algorithm to detect,
report and handle failure. We start with
this algorithm that we modify to obtain
the Centralized Manger Strategy in the
following subsection. Then, we propose
CMSA to achieve fault-tolerance and
provide a continuous service for
applications supported by WSNs.
3.1 The Centralized Maintenance
Strategy
In our study, we assume a connected
network and an area covered by static
wireless sensors. We suppose also that
all nodes know their location by means
of a localization technique like the GPS
(Global Positioning System) [9].
At the beginning, robots are uniformly
distributed over the considered area. One
robot is selected to act as a central
manager. Failures are detected by the
guardian nodes and reported to the
central manager robot. The manager then
forwards the report to a selected
maintainer robot to handle the failure.
Like the centralized manager algorithm,
we distinguish three stages in this
strategy:
In the first stage (the initialization stage),
we have an exchange of three types of
messages: (i) Sensor node broadcast
messages: this message containing the
identity of sensor node and its location is
periodically broadcast to the one-hop
neighbors sensor nodes. Upon receiving
such messages, any sensor node of the
network can determine the list of its
106
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
guardees (its one-hop neighbors). Hence,
the
guardian-guardee
relationship
between sensor nodes is established, (ii)
Manager broadcast messages: The
manager robot broadcasts periodically its
identity over the network (to all sensor
nodes and maintainer robots) and (iii)
Maintainer reply messages: When a
maintainer robot receives a manager
broadcast message, it responds with a
message containing its identity and
location. Hence the manager can
determine the location of all maintainer
robots in the network.
To establish a guardian-guardee relation
between nodes in WSN, the centralized
manager algorithm [1] supposes that a
guardee node selects the nearest node as
a guardian. This choice presents some
drawbacks. Indeed, if a failure occurs on
the guardee node that is the relay point
for the guardian to the manager robot,
the report failure cannot be sent to the
manager robot. In our strategy, we
propose that any node is guarded by all
its one- hop neighbors. In this case we
guarantee that at least one report failure
is received by the manager.
After initialization, any sensor node
periodically sends its broadcast message
(message type 1) to its one-hop neighbor
nodes. If a guardian has not received any
messages from a guardee for a certain
amount of time, the guardian deduces
that the guardee has failed and sends a
report failure containing the identity and
the location of failed node to the
manager robot (the failure detection and
reporting stage).
To report failures, Mei and all in [1]
have used a geographic routing protocol
(knowing that we assume network
connectivity and area coverage). But
upon multiple simultaneous failures, this
assumption is not realistic. So, using the
nearest node to the manager as a relay
107
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
measure
the
different
transition
probabilities (
). Once
the
transition
probabilities
are
determined, they are sent with the value
of the residual energy
and the node
location to the manager robot. With the
reported measurements, the manager can
compute the lifetime of sensors using the
analytical model. Indeed, it is equivalent
corresponding to:
to finding the time
(12)
Where:
is calculated according
to the equation 11.
Therefore, with our proposed analytical
model, the manager robot can first
estimate the N expected sensor failures
in the network designated by
). To determine the maintainer
(
, the
robot of an estimated failure
robot manager identifies the list of
robots able to handle the anticipated
failure at its required replacement time
with a null off-service time and
selects among them the nearest robot as
the
maintainer robot. To ensure a null
off-service time for the failure
, the
must be
selected maintainer robot
scheduled to repair the expected failure
at:
(13)
With:
-
and .
current position of
represents the speed of
108
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
used in simulation is equal to (0.1404
Watt, 0.1404 Watt, 0.0018 Watt,
0.000018 Watt). References values are
taken from a ZigBee node implementing
IEEE 802.15.4 medium access.
4.1 The Network Dysfunction Time
In a first scenario, we suppose that
sensor failures are caused only by energy
depletion. Experiments are run with
different bit rates per sensor node: 1, 2
and 3 kb/s. With CMS and CMSA,
simulation has shown that we must have
at least 2 robots to guarantee the network
connectivity and coverage of the
considered area. We designate by
network dysfunction time, the sum of
duration for which the network
connectivity and the coverage are not
guaranteed. In other words, its the
period during which at least one node is
failed in the WSN. Figure 1 represents
the variation of the percentage of the
network dysfunction time when CMS
and CMSA are used for different number
of maintainer robots with different node
bit rates (1, 2 and 3 Kb/s).
109
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
second scenario where we add random
failures generated according to a Poisson
distribution with an average of 2 failures
per hour. Therefore, two types of failures
can occur: failures caused by energy
depletion which are anticipated and
repaired before they happen and
unpredictable failures which are detected
by guardian nodes and reported to the
manager.
We represent in figure 2 the network
dysfunction time given by CMS and
CMSA for different number of
maintainer robots and a fixed node bit
rate equal to 3Kb/s. This figure shows
that CMSA reduces considerably the
network dysfunction time using the
minimal number of robots. In fact, the
network dysfunction time provided by
CMSA is due only to the repair time of
unpredictable failures generated during
the simulation, which explains the low
percentage of the network dysfunction
time compared to CMS.
110
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 103-111
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
network, the reaction time upon a failure
cannot be tolerated as simulation results
have shown. To remedy this problem,
we have proposed an anticipated version
of Centralized Manager Strategy:
CMSA. In this technique, the manager
receives the transition probabilities
matrix from each node sensor in the
network. The transition probabilities are
used by the analytical model to estimate
the lifetime of sensor nodes. Before the
estimated sensor lifetime expires, the
manager sends an order to a maintainer
robot to replace the corresponding
sensor. Simulation results have shown
that using this technique, we obtain a
null dysfunction time of the wireless
sensor network with a minimal number
of robots.
In a future work, we propose to improve
the two distributed strategies given in [1]
to coordinate the movement of robots
based on fixed distributed manager
algorithm and dynamic distributed
manger algorithm. Thanks to our
analytical model predicting the lifetime
of a sensor node, these strategies will be
made able to support real-time
applications.
5.
6.
7.
8.
9.
6 REFERENCES
1.
2.
3.
4.
111
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
KEYWORDS
Rules for legislation, visualization,
structuring, object-orientation, UML
1 INTRODUCTION [ 1]
The law primarily has a linguistic
representation
especially
within
language-constructs. This is changing,
and the representation is becoming
more multi-layered, both in terms of
the form (text corpus as a legal
information system, citizen portals,
formalization,
structure,
images,
graphics, input forms, etc.) as well as
the legal regime (international law, EU
law, national law, etc.). A linguistic
transformation is taking place in the
concretization of general norms by
112
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Table 2
E. g. the German Tenancy Law:
Table 3
E.g. the French Code Napoleon:
Object Orientation is an
attempt to develop structures of
organi-zation as well as
software by classification of
information
using
their
characteristics and the possible
operations. An object-oriented
design is part of object oriented
methodology and forces the
analyst to think more in terms
of objects and less in
procedures. An object contains
encapsulated
data
and
procedures which are grouped
together to represent an entity.
Therefore
the
interface
between the objects is exactly
defined.
Table 4
E.g. the Italian Tenancy Law:
113
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
but standar-dized model of
complex systems.
3.1.2
Fundamental Aspects.
114
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
3.1.3
Objective Facts
115
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
3.4 Strategies
would be identical with:
The further automation, streamlining
and simplification will form the basis.
Fundamentally, all rules should be
formulated in the manner of a daily or
with a "check list" and should be
simply readable. The principle of the
avoidance of lengths and nesting
should be considered consistently, as
most of the Legistik demand. New
techniques are necessary, like charts
and tales, displayed clearly on Excelsimilar pictures. The visualization [3]
by surveying pictures and the means of
structure-diagrams, flow charts and
graphic diagrams must be raised to the
principle.
Also, the important logic - notation
with conscious and visible marking of
the logical AND, the logical OR
and the building of parenthesis is of
decisive meaning.
It must be
considered, that the combination
between sentences and/or parts of
sentences is made with different
conjunctions. So an AND (=UND)
connects more intensively than an
OR (=ODER). A reversal of this
rule is possible, as shown in the
following example:
If the event A happens and
the event B happens or
event C happens, then event
XY would happen.
[3]
116
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
The
last
connection-word
(UND/AND) represents only a
linguistic usage and not a logical
AND.
The logical content should mediate us
really a logical enumeration, therefore
the above-mentioned AND is just a
connection for the enumeration. The
law is intended to inform you, - in
spite of this not-logical AND
(und) that this law is made for
AND
fr die genossenschaftlichen
Nutzungsvertrge
But if you are following the exact
logistical rules, you get: [4]
1. (1) Dieses Bundesgesetz gilt
nur, wenn es sich handelt um
(this law is only valid, if)
die Miete von Wohnungen
OR (VEL)
einzelnen Wohnungsteilen
OR (VEL)
Geschftsrumlichkeiten aller
Art
OR (VEL)
fr die genossenschaft-lichen
Nutzungsvertrge
] Rechtsmodellierung im E-Government,
Fallbeispiele zur Legistik (2005), Diss, Kepler-Universitt
Linz
117
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
die Miete von Wohnungen
OR (VEL)
die Miete von einzelnen
Wohnungsteilen
OR (VEL)
..
DANN
(THEN)
ist
das
Bundesgesetz anzuwenden.
(Just in these cases you get legal
force)
Thereby it would be clear, that there
is only the logical listing OR (=
VEL) and not the excluding OR (=
AUT) and not a logical connection
with AND. (The logical AND would
necessitate,
that
you
SIMULTANEOUSLY must have all
the above conditions, as
118
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Abstraction and Application
In law, abstraction should give a
focused view on the essential, i.e. the
concrete
elements
and
their
characteristics receive a certain
conceptual meaning.
Generalization and their formal
representation are the basis for the
multi-layered representation of the law
(text, commentary, legal ontologies,
logical structures, etc.).
The application is an application
software program to solve user
problems. The application is the
practice of theory and abstraction.
The software can be very complex
reflecting its importance in the
knowledge society. The app is an
application for smart phones and
tablet computers
(e.g. RIS: app to search in the
consolidated version of the federal
law).
While the abstraction has a very long
tradition, the application and the app
are a recent development. Complex
processes should be supported by
means of a user-friendly smart phone
application, which in turn requires an
abstraction (and simplification). In law,
the challenge of the app is presenting
complex search queries and documents
in a simple and user-friendly form.
One of the goals of the IRIS2013 was
to highlight the interaction between
abstraction and application in law and
to find appropriate practical solutions.
The IRIS conference is well known for
its interdisciplinary approach and the
involvement
of
administration,
business and civil society.
Figure 3: Action
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Jurisprudence is underdeveloped,
which limits itself to dogmatic)
4.2.2 Initiatives
5 CONCLUSION
120
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Additional References
1. Schweighofer, E.: The leading topic of
IRIS2012, , august 23th, Salzburg, (2011)
2. Kahlig, W.: Speech at Kepler-University,
Linz, Einfachere oder gerechtere Gesetze
(2005)
3. Lachmayer, F.: Mglichkeiten einer
Verwendung normentheoretischer
Analysen fr die Gesetzgebung, Studien zu
einer Theorie der Gesetzgebung, SpringerVerlag Berlin-Heidelberg-New York
(1976)
4. Kahlig, W.: Rechtsmodellierung im EGovernment, Fallbeispiele zur Legistik,
Diss, Kepler-Universitt Linz (2005)
5. Adamovich, L.: Probleme einer modernen
Gesetzestechnik, in: Winkler, G. /
Schilcher, B.(Hsg.), Gesetzgebung, Wien
(1981)
6. Heindl, P., Kahlig, W.: Mietrecht
anschaulich, Manz, Wien, (2001)
7. Heindl, P., Kahlig, W.: Mietrecht
anschaulich, Manz, Wien (2002)
8. Heindl, P., Kahlig, W.: Wohnrecht
anschaulich, Manz, Wien (2003)
9. Heindl, P., Kahlig, W., Stingl W.: Wohnund Steuerrecht anschaulich, Manz, Wien
(2004)
10. Heindl, P., Kahlig, W., Stingl W.: Wohnund Steuerrecht anschaulich Gesamtausgabe, Manz, Wien (2004)
11. Heindl, P., Kahlig, W.: Mietrecht
automatisch, Manz, Wien CD-Ausgabe
(2004)
12. Heindl, P., Kahlig, W.: WEG2002
automatisch, Manz, Wien CD-Ausgabe
(2004)
13. Kahlig, W.: Mietrecht einfach, Eppenberg,
Wien (1997)
14.Kahlig, W.: Handbuch des
Immobilienwesens, CONTAKT,
Loseblattausgabe (2004)
15. Kahlig, W.: Steuer-Handbuch des
Immobilienwesens, CONTAKT,
Loseblattausgabe (1996)
16. Stingl,W., Kahlig, W.: Beibltter im
Steuerrecht anschaulich, Manz, (2004)
17. Stingl, W., Kahlig, W.: Beibltter im
Steuerrecht automatisch, Contakt (2005)
18. Lachmayer, F., Reisinger, P.: Legistische
Analyse der Struktur von Gesetzen, Manz,
Wien (1976)
19. Lachmayer, F.: Grafische Darstellungen als
Hilfsmittel des Gesetzgebers,
Gesetzgebungs-theorie, Juristische Logik,
Zivil- und Prozessrecht, Springer-Verlag
Berlin-Heidelberg-New York (1978)
121
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Reform der Standards fr die Gestaltung
von Rechtsvorschriften (2003)
38. Kahlig, W.: Content Visualisierung,
Johannes Kepler Universitt Linz (2003)
39. Kahlig, W.: Netzplananalysen von
Gesetzesentwrfen, Krntner
Landesregierung, Klagenfurter LegistikGesprche (2003)
40. Kahlig, W.: Visualisierung des
Wohnrechts im Spannungsfeld zwischen
dem Rm. Recht und den Kantschen
Thesen, Universitt Salzburg, IRISKongress (2004)
41. Kahlig, W.: NetzplanVisualisierung,
Universitt Innsbruck, Arbeitswissenschaftliches Institut (2004)
42. Heidl P., Kahlig, W.: Legistik im 3.
Jahrtausend, Prtschach, Bundestag der
Immobilientreuhnder (2004)
43. Kahlig W.: Rechtsmodellierung am
Beispiel des Wohnrechts, Johannes
Kepler -Universitt Linz (2004)
44. Kahlig, W.: Begriffsvisualisierung am
Beispiel des Wohnrechts, Universitt
Salzburg, IRIS-Kongress (2005)
45. Kahlig, W., Stingl, W.: Das Wohnrecht im
3. Jahrtausend, Winterseminarwoche des
VI (sterr. Verband der Immobilientreuhnder), Schladming 2005
46. Kahlig, W.: Das Wohn- und ImmobilienSteuerrecht im 3. Jahrtausend, FHW,
Lehrgang fr Immobilientreuhnder
(2005)
47. Heindl, P., Kahlig, W., Stingl, W.: Das
Wohn- und Immobilien-Steuerrecht
anschaulich Der neue, vereinfachte
Zugang zu den Wohnrechtsgesetzen,
Gewerbehaus der Handelskammer Wien
(2005)
48. Kahlig, W., Stingl, W.: Vorteile im
Immobilien-Steuerrecht durch
CONTHAUS-Software, IBM-Symposium
Messegelnde Wien (2005)
49. Kahlig, W.: Ab wann werden nur mehr
Computer Gesetze machen?, Universitt
Innsbruck (2005)
50. Kahlig, W.: Automatisierte Verfahren im
Immobilien-Wohnrecht, FHW,
Fachhochschule Wien, (2005)
51.Kahlig, W., Stingl, W.: Das ImmobilienSteuerrecht,
Bundestag der Immobilientreuhnder in Loipersdorf (2005)
52.Kahlig, W.: Wohn- und ImmobilienSteuerrecht fr Praktiker Der neue
Zugang zu den Wohnrechtsgesetzen,
Landesinnung Tirol, Innsbruck,
Fachgruppe (2005)
53. Kahlig, W.: Wohn- und ImmobilienSteuerrecht fr Praktiker Der neue
Zugang zu den Wohnrechtsgesetzen,
122
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Konzessionsprfung fr Immobilientreuhnder, WIFI, Wien (2007/08)
72. Kahlig W.: Vereinfachung von Gesetzen,
Vorbereitungskurs fr die Konzessionsprfung fr Immobilientreuhnder, WIFI,
Wien (2007)
73. Kahlig, W.: Das strukturierte Wohn- und
Immobilien-Steuerrecht, Bundestag der
Immobilientreuhnder in Wien (2007)
74. Kahlig, W.: Strukturiertes Wohn- und
Immobilien-Steuerrecht, FH-Wien (2007)
75. Kahlig, W., Stingl W.: Automatisierungsverfahren im Immobilien
Steuerrecht, Vortragsreihe der
ERSTEBANK, Wien (2007)
76. Kahlig W.: Juristische Visualisierung mit
UML, Universitt Innsbruck (2007)
77. Kahlig, W.: Automatisierungsverfahren fr
Immobiliengesetze, Vortragsreihe der
Wiener Landesinnung Fachgruppe fr
Immobilientreuhnder (2008)
78. Kahlig, W.: Unified Modeling Language
fr juristische Anwendungen, Universitt
Salzburg, IRIS- Symposium (2008)
79. Kahlig, W.:
http://lawgical.jura.unisb.de
/index.php?/entry/343IRIS2008-Wolfgang-KahligUnified-Modeling-LanguageUML-fuer-juristischeAnwendungen.html
80. Kahlig, W.:
http://www.jurablogs.com/de/b
ersicht-iris-2008
81. Kahlig, W.: Restrukturierung des
Rechtsgebudes mit formalen Mitteln,
Universitt Innsbruck (2008)
82. Kahlig, W.: Kahlig C.: Entwicklung des
Wohnrechts, Festveranstaltung Hotel
Marriott, Wien (2008)
83. Kahlig W.: Vereinfachung von Gesetzen,
Vorbereitungskurs fr die
Konzessionsprfung fr
Immobilientreuhnder, WIFI, Wien (2008)
84. Kahlig, W.: Ontologie und Logiken in der
Versicherungswirtschaft, Universitt
Wien, Juridicum (2008)
85. Kahlig, W.: Legal Cognitive Decision
Processing Towards a Dynamic Legal
Taxonomy, London East University,
London, GB (2008)
86. Kahlig, W.: Juristische Visualisierung
beyond UML, Universitt Innsbruck,
Arbeitswissenschaftliches Institut (2009)
87. Kahlig, W.: Visualisierung des
Wohnrechts Gesamtbild und Statistiken,
Universitt Wien, Juridicum (2009)
88. Kahlig, W.: Strukturimmanente
Denkanstze in der Legistik, Universitt
Salzburg, IRIS-Kongress (2009)
123
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 112-124
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
106.Kahlig, W.: IT-untersttzte Rechtsvisualisierung am Beispiel des WEG,
Universitt Innsbruck (2012)
107.Kahlig, W.: Recht und Logik, Techn.
Universitt TU Mnchen, Germany (2012)
108.Kahlig,, W.: Vereinfachung von
Gesetzen, Vorbereitungskurs fr die
Konzessionsprfung fr Immobilientreuhnder, WIFI Wien (2012)
109.Kahlig, W.: Strukturierung des
Rechtswissens, Universitt Innsbruck
(2012)
110.Kahlig, E., Kahlig W.:
Legistical Efforts By E-Government,
Dubai, I.A.U. University (2013)
111.Kahlig, W.: Strukturierung des
Rechtswissens -ex ante und ex post,
Universitt Salzburg, IRIS-Kongress
(2013)
124
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
1 INTRODUCTION
Wireless networks are susceptible
and exposed to attack because of its
borderless nature. It is easy to penetrate
any wired network via wireless network as
Access Point (AP) is bridging between
wireless and wired network. Packet
sniffing can be done passively because of
the hub-based configuration of the APs.
This activity initiates more dangerous
attacks such as Man In The Middle
(MITM) attack. MITM attacks such as
session hijacking and MAC spoofing are
some of the critical threats for wireless
125
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
threats. The statistic from the report shows
that the pattern of attacks changes from
time to time which basically follows the
development of the internet. From 1997
until 2003 most of the incidents reported
were due to spam and virus cases, but from
2004 until 2011 there is tremendous
incline in the attempt to intrude and
intrusion of the network attack. In 2003
only 60 cases due to intrusion were
reported, but in 2004, 368 cases were
reported. In the same report, the
inclination is very obvious from 2007 until
2011 when attacks due to intrusion
increase sharply by 861%.
In this paper we present current
and persistent threats to wireless networks
and accompanied by some active
researches from academia regarding the
advancement and some counter measures
to the threats.
The second section of the paper
discusses some current threats in wireless
networks and researches of developments
and countermeasures to the attacks. The
third part discusses security risks to the
wireless network due to the advancement
of cloud computing, and the fourth section
presented some new hacking tools
available on the market. The fifth section
is regarding the latest developments of
wireless networks, section six presents a
case study in Malaysia, findings and
analysis in section seven, some proposed
countermeasures and mitigation in section
eight, criteria for secure wireless network
in section nine and conclusion in the last
section.
2
THREATS
NETWORKS
OF
WIRELESS
2.1
Vulnerability
Scanning
and
Password Cracking
Vulnerability scanning is a process
when hackers use certain tools such as
Kismet and InSSider to scan the network.
The objectives of this activity are to:
a) find the vulnerabilities and security
level of the network
b) determine signal strength
c) determine the accessibility of the target
network
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
From the war driving experience more
than 50% of the residents have no security
or implementing weak security protocol.
Researches to attempt cracking
WEP and WPA/2 were done in [5-8].
Series of attempts by Fluhrer, Mantin and
Shamir (also known as FMS), and later in
2004 a person under pseudoname KoreK
made a second attempt and succeeded.
Tews, Weinmenn and Pyshkin (known as
PTW) launched new generation attack in
2007 followed by a Chopchop attack.
Based on these successful attempts,
cracking tools were developed. In WEP
mode even though the length of the
passphrase is increased or complicated,
only 30 minutes were taken to break the
code. Nevertheless, users which are using
WEP because of the convenience of
setting and interoperability, are advised to
set proper passphrase which will take
longer time to break and will create noise
in the network.
Even though WPA/WPA2 is said
to be robust, it is still protected by a
passphrase which can be cracked by using
Dictionary Attack. In [8], a new
proposed space-time trade off solution is
used where the Pair Master Keys (PMK)
are pre calculated for each passphrase in
the library and store them into another
library called Hash Library. This research
also
suggested
employing
cloud
computing to generate possible passphrase
and to take GPU parallel computing into
consideration to effectively calculate PMK
and proofread the Hash Library.
This section concludes that with
the matured amount of wardriving and
password
cracking
activities
had
developed the advanced and powerful
tools for network scanning and password
cracking.
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
of data gathered from sniffing activities
was reduced, so the system was capable to
listen to the traffic in the larger radius.
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
executed algorithm might be jeopardized
and users might get false alarm of RAP or
wrong information about the illegitimate
APs.
In [16], an end user solution has
been proposed in order to detect RAPs
especially in public hotspots. In this
proposed work, a flexible and practical
solution is suggested, especially for mobile
users or travellers to protect their
credentials where the security monitoring
is not reliable. The technique proposed
was based on the knowledge that in the
existence of RAP, the client has to
communicate with a remote server through
an evil twin AP and a normal AP. In this
case, compared to the normal scenario, the
twin evil case has one more wireless hop.
To distinguish these two cases (one and
two wireless hops), Inter-packet Arrival
Time (IAT) statistic, which is a time
interval between two consecutive data
packets sent from the same devices has
been adopted. A prototype system called
Evil Twin Sniffer (ET Sniffer) which has
been evaluated in real wireless network
including 802.11b/g was introduced by
using two novel algorithms which are
Trained Mean Matching (TMM) and Hop
Differentiating Techniques (HDT). The
analysis from the experimental set up
yields ET Sniffer can detect illegitimate
APs quickly with high accuracy and
resilient
to changes in wireless
environment.
In [17], Somayeh et al proposed a
novel method for RAP detection on the
client-side which is able to detect both
MITM attack and evil twin attack. This is
a client side detection method where
clients mobile will be warned on the
safety of the network by indicating
different colours of lights namely green
(safe), yellow (RAP is present) and red
(MITM attack). In this research, a method
to decide whether the network is exposed
to RAP, tricked by RAP or is it a safe
network is designed by comparing SSIDs,
MAC and IP address of the public APs.
Green light will be indicated if the network
is safe to connect to, yellow indicates the
existence of RAP and red light is the sign
Denial of Service
One of the weaknesses of wireless
network is the restricted usage of
bandwidth. This situation is giving great
advantage to hackers to launch Denial of
Service attack by replaying packets in
order to generate noise or by sending deauthenticate packets to the legitimate users
in the subnet. In fact, according to Belly
Rachdianto [18], this is the technique used
by a large group of hackers, Anonymous
to bring down governments web servers in
the recent attacks.
Practically, jamming attack is a
popular attack in military battlefield
usually by exploiting the weaknesses of
Wireless Sensor Network (WSN) which
have limited power resources. Jamming
attack can be launched either from outside
or inside the network by using jamming
devices to block the wireless medium and
prevent
wireless
devices
from
communicating with each other in order to
degrade the performance or to paralyze the
network. Some researches presented
herewith are advancements in jamming
attack which could be launched from
internal and external of the network.
Liu et al. in [19] have proposed
mobile jamming attack to Wireless Sensor
Networks (WSNs). In this attack, the
mobile jammers are placed randomly in
the network, listen to the network traffics
and detect the cluster head based on
received signal strength (RSS). According
to [19], the attack to the cluster head node
is an effective method because if the
communication between this node and sink
node is interrupted the base station may
not get any information. The results show
that mobile jammers has longer lifetime
129
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
corrupted and less energy is needed for
mobile jammers to bring down a WSN.
In selective jamming/ dropping as
proposed in [20], the targeted channel
which was defined by separate frequency
band, time slot or Pseudo Noise (PN) code
will be under attack or the attacker will
select data of high importance by
observing specific packet identifiers. Due
to the nature of selective jamming, this
type of attack is not easy to detect as
aggregate behavioural metrics are unable
detect attacks of selective nature, when
only a small fraction of packets were under
attacks.
Random
Packet
Distribution
(RPD) is a type of external jamming attack
when the specified attack model is wellprotected with various security measures
[21]. RPD jammer is put at the area of APs
in a Wireless Mesh Network (WMN). At
this point, the jammer will destroy the
packets by transmitting a short period of
noise signal when packets transmission is
detected. According to [21], any kind of
encryption and authentication may not
mitigate the risk of such attack because the
attacker does not need to participate in the
network.
In this investigation, the analysis of
damages caused by this attack was
conducted using Qualnet Network
Simulator. Performance metric such as
goodput, delay, jitter and Mean Opinion
Score (MOS) were analysed. The
simulation also showed that real time
applications such as (Voice over Internet
Protocol) VoIP will be adversely affected.
MOS score shows that at 5% attack rate
can affect the voice quality. Based on the
analysis of TCPs throughput in [21],
packet lost rate at 20% can bring down a
network, which is equivalent to 40% RPD
attack rate.
In another research, Distributed
Jammer Network (DJN), utilizes the
advantages of nano-technology, where the
jammers are not visible in the naked eyes
is proposed by Hong et al [22]. These low
powered jammers are claimed to have less
self-interference. DJN has few distinguish
advantages which are:
2.6
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
Bluetooth technology which enabling
file transfer and other data communication
between devices such as printer, PDA and
smartphones are vulnerable to threats
which are classified as in ABOTT (A
Bluetooth Threat Taxonomy) which is
listed below.
a) Surveillance Bluprinting, bluefish, sdp
tool
b) Range Extension BlueSnipping,
bluetone
c) Obfuscation- Spooftooth
d) Fuzzer BluePass, BlueSmack
e) Sniffing BlueSniff, Wireshark
f) Denial of Service Signal Jamming,
Battery exhaustion
g) Malware BlueBag, Caribe
h) Unauthorized direct data access
Bloover, BluBug
i) Man in the middle BlueSpoof
2.7 Threats in RFID
RFIDs are used as smart tags
which are widely deployed and it also
exposed to many kind of threats; which is
similar to attacks in WiFi. Some of the
common attacks are sniffing, tracking,
spoofing, tracking, replaying and Denial of
Service attack. The most critical attack in
RFID is replay attacks which enable the
attacker to intercept and retransmit RFID
queries.
3
HACKING TOOLS
The attacks in wireless network are
aided by tools and some of the tools have
been used in attacks explained in section
two. Furthermore, the devices proliferation
for wireless hacking tools online which
require minimal technical skills and easy
to execute such as Raspberry Pwn, Nokia
N900 PwnPhone, and Wi Fi Hot Spot
Honeypot trigger more hacktivism. In the
recent Hacker Halted Conference on 19
November 2012 in Kuala Lumpur, Wayne
M Burke, the CEO of Sequrit [30], has
presented some of the mobile hacking
tools as said above and found that those
tools are being sold from as low as USD
99.99. These tools are merely for the
security professionals and for the
educational purposes, but also have great
potential to be misused, abused and
exploited
by hackers.
The
vast
development of these mobile devices
which have permanent connectivity to the
internet can be used to spy, drop malicious
code (malware), monitoring person by
GPS, redirected the email and read
messages.
4.1
Raspberry PwnPi
131
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
This tool is a fully-featured
security penetration and auditing platform
[31]. It has comprehensive hacking tools
which will lure the hacker to attack
wireless
medium
which
include
information gathering, network mapping,
penetration
and
vulnerability
identification. It also releases free
community edition, which is definitely an
excitement to all the hackers.
Nokia N900 PwnPhone
Nokia N900 is mobile-phonebased-pentesting platform which includes
Aircrack-NG, Metasploit, Kismet and also
featuring oneclick evil AP, WEP Cracker
and a packet capture. This comprehensive
device is available in the market at the
price of USD 995.00.
4.2
4.3
Mobile Vulnerabilities
Recently, it also has been reported
that there is a flaw in wireless chipset
design
provided
by
a
wireless
semiconductor manufacturer, Broadcom
LATEST DEVELOPMENTS
According to [34], found that there
is a tremendous increment in terms of sales
in tablet and smartphones market. In 2012,
821 million of these devices were sold
worldwide, but in the first quarter 2013
itself the sell is 1.2 billion. There is a 46%
increase of sales and it is predicted that
these numbers will be increasing from
time to time. The huge number of sales in
these mobile devices is supported by
flexible
operating systems
mainly
Android. Android out-casted other players
such as Apple and Microsoft by
conquering 72% of the worldwide market.
In fact, during the third quarter of 2012
Android device sales exceeded PC
shipment by 40%.
5.1
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
security settings of online shopping sites,
which was the main target for SQL
Injection and cross scripting threats. This
is a significant threat to the local network
as those devices were synchronized with
the local network.
BYOD is also the uprising trend in
schools and hospitals. The main concern in
applying BYOD in these environments is
the proper and solid segregation between
authorized and unauthorized territory for
the users. Students especially, by nature
are very curious homo-sapiens, so the
tendency for them to test their level of
authorized access. If they reach the core
network there are many possibilities of
malicious activities such as data
modification, forgery, cheating and any
other possible unethical doings. The
network must be very properly designed,
so that all the users are allocated
reasonable bandwidth, unless the users
will get frustrated and bring their own AP
to access the network.
5.2
Devices Tethering
In [35] states one useful and
flexible feature of tablets and smartphone
is the capability to tether or distribute their
bandwidth of cellular provider to any other
nearby devices such as laptops or any
other mobile devices. The 3G/4G cellular
network provider will charged this service
with certain amount of bill. The main
concern here is whether the end devices
are consistently secured. The risk of illegal
signal interception (MiTM) is high if the
level of security is not properly set. Some
apparent threats are DoS, virus threats and
compromised privacy due the possible
illegal GPS chips activation. The worst is
when central corporate network is
bypassed through an unsecured mobile
connection. Another issue that might arise
is the lacking of encryption levels in jailbroken phones.
5.3
Convergence of WiFi and
Cellular Network
In the wider scale, there is a
massive integration of WiFi and cellular
133
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
6
WIRELESS NETWORK
SECURITY: A CASE STUDY IN
MALAYSIA
In Malaysia, cyber space is
governed and monitored by three
incorporated bodies which are Malaysia
Communication
and
Multimedia
Commission (MCMC), Cybersecurity and
Malaysia Computer Emergency Response
Team (MyCERT). These bodies are in
affiliation with Ministry of Science,
Technology and Innovation (MOSTI).
MCMC is a body in which regulating rules
and is authorized to take action against any
wrongdoings subjected to cyber security,
while Cybersecurity and MyCERT are the
bodies investigating and receiving incident
reports or breaches.
According to a Security Assurance Analyst
from Cybersecurity, incidents reported due
to wireless and mobile devices threats are
relatively small, which is only four cases
reported in 2012. The reported incidents
are due to scam, password cracking,
bandwidth leeching and compromised
access point. However, it does not indicate
the wireless threats are not significant in
Malaysia but there is a huge possibility
that other incidents were not reported
accurately in which did not mentioning the
134
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
As indicated in Figure 2, intrusion
and fraud have inclining pattern towards
2011. It can be deduced that malicious
activities that causing monetary loss and
privacy intrusion developed rapidly and
this is might due to the vast development
of Wifi, especially public Wifi. An
example of intrusion is network and
system intrusion and examples of fraud
cases are web defacements and website
phishing. In many organizations, such as
hotels and hospitals public Wifi is offered
to ensure comfort and accessibility to the
users, but the network might not properly
segmented and isolated that will cause
outsiders reaching the internal network
intentionally or not. But in 2012 it is
shown that cases due to these two
categories were declining. This is due to
the
awareness
of
the
network
administrators to increase the encryption
level of their own network.
The investigation by Cybersecurity
when doing penetration testing in some
companies in Malaysia yields some
common threats in wireless network. The
most popular threat found is WEP
cracking. It was found that the network
administrator tend use default setting or
easy-to-guess password to secure their
wireless network. Rogue access point
(RAP) is another common threats found in
the investigation. RAPs were installed to
the local network and it bridges the local
network to the outer network. Some
authorized officers might intentionally
brought AP for his personal use, without
knowing that a hacker can get access
easily from his RAP.
In some companies, they tend to
ban the officers to have access to some
websites that might contain virus, so this is
when the officers will bring their own
broadband dongle and plugged it into the
laptops and desktop. This action might
transfer the virus into the companys
system and infected the whole network.
Another finding is signal leakage which is
due to inappropriate planning of signal
radius. While doing network scanning
135
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
which are affordable and easy to use and
carry and the vast availability of hacking
software on the internet.
RAPs are usually found in public
hotspots. With the availability and
competitive price of physical access points
and with a little practice and training,
anyone could install a RAP to any public
hot spot. Furthermore, RAP is almost
invisible especially when the attacker uses
his own laptop as a soft access point. This
threat is actually leads to another
dangerous attack which is Man In The
Middle (MITM). In wireless network
MITM is dangerous because it is a passive
and silent attack which listens and sniffs
users conversation and most probably will
intercept and modify the packets being
relayed. MITM is almost impossible to
prevent as long as the attacker find a
hole to access the network, legally or
not. MITM can lead to another disastrous
attack, for example session hijacking.
People are always the weakest link
in security. The threats posed by social
engineering are almost destructive as other
technical threats. The art of deception is
the most useful skill in social engineering,
while insiders ignorance and negligence
always open the door for the opportunist
attackers to cause security breaches.
On the other hand, a lot of
researches on improvisation of the attacks
and counter measures against the attacks
have been carried out. For example, Denial
of Service attacks especially jamming
attacks have gone through many phases of
advancement in terms of capability and
invisibility,
while
countermeasures
techniques to evade RAPs are also on the
rise.
The recent developments of
smartphones such as tethering capabilities,
convergence with 3G/4G and BYOD
policy add more security vulnerabilities.
The service and users must be more
responsible
to
ensure
secured
communication by adhering to strict policy
and technical countermeasures. The power
to choose, buy and use is in consumers
136
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
8.3
8.4
8.5
Policy Enforcement
By
having
clear
and
understandable policy or agreement of
prohibited actions in wireless network
such as packet sniffing or placing any
device which can act as a RAP might help
to reduce malicious activities in wireless
network. As to handle threats that might
occur because of social engineering, a
written agreement of policy between
employees and employers must be taken
place from the first day of work.
Employees might be given certain unique
password to login into companys WiFi
network and repeatedly reminding them
not to share the password and other
sensitive credentials to others.Precautions
must also be taken when the users are
using wireless network outside of their
homes or offices, such as hot spots to
safeguard confidential credentials from
any attackers. Here are some advices:
8.6
137
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
registered and an automated system must
be deployed to enabling APs updating
activities.
Role Based Access Control
(RBAC) is a system that will assign a role
to any device based on how they were
authenticated. RBAC also defines devices
accessibility level and what they can
access from a network. In order to avoid
misuse and abuse of the network and other
security breach, a secured wireless
network management should be able to
monitor users activity in real time. This is
important as there will be hundreds or
thousands of devices connected to the
network, so it is important to monitor and
manage the applications and programs that
they are using in your network. Network
administrator also can even remotely wipe
the device if it lost or stolen.
All devices are also need to be
registered and assigned unique IP from the
network. To avoid hassle, network access
control can allow the users to register
themselves to the network. As for the
network administrators, this is when they
can check the latest updates and validate
security level of the devices.
9.3
CONCLUSIONS
As presented in this paper, wireless
networks are susceptible to attacks and
prone to many types of threats. This is due
to its seamless nature and its popularity.
Users in public hotspots are claimed to be
more exposed to such threats because of
the ease to deploy MITM by sniffing
techniques and planting RAPs and most of
the time victims do not aware that their
data or privacy has been compromised due
to passive nature of the attacks.
Vast availability of the hacking
tools and devices opens up wide path for
hacktivism, while the advancement of
cloud computing making Wi Fi more
vulnerable to threats. The security of cloud
computing as well must be managed very
well to avoid leakage and breach of
information.
Wireless network is an easy access
to local system if the network is not
properly segregated. The developments of
wireless network give flexible access
especially under BYOD policy. For this to
work in a secure environment set of rules
must be followed and adhered.
Acknowledgement: This research is in
affiliation with Ministry of Higher
Education, Universiti Teknologi Malaysia,
Malaysian Japan International Institute of
Technology
(MJIIT)
and
i-Kohza
Computer System and Network (CSN).
.
11
REFERENCES
1.
2.
3.
138
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
139
International Journal of Digital Information and Wireless Communications (IJDIWC) 3(1): 125-140
The Society of Digital Information and Wireless Communications, 2013 (ISSN: 2225-658X)
30.
31.
32.
33.
34.
35.
36.
37.
38.
BYOD
http://www.securedgenetworks.com
The Security Implications of Tethering a
Mobile Phone To Your Computer
http://www.webologist.co.uk
Republic
Wireless
http://www.republicwireless.com
TM WiFi www.tm.com.my
11 Features Every Secure Wireless
Network
Should
Have
http://www.securedgenetworks.com
140