Professional Documents
Culture Documents
users and with a curve connecting the purpose-space point, we propose a novel mobility prediction approach, i.e. historybased handover prediction. This approach is useful to improve the accuracy of a prediction of users location and optimize the handover configuration parameters. The rest of the paper is organized as follows: In section 2 the approach of the proposed technique is described. Section 3 introduces the handover technique details while section 4 deals with model implementation and simulation results. And finally in section5 the work is concluded. II. A. THE APPROACH OF THE PROPOSED TECHNIQUE
I.
INTRODUCTION
As LTE system is currently under extensively study, the user mobility management is of special interest. Mobility of users with seamless accessibility and without the need to care about the underlying topology is the very source of many challenging issues. There are several management tasks that are deeply influenced by the users mobility. To guarantee a seamless service the correct estimation of users future location is critical. Mobility prediction has been considered as an effective technique for fast and seamless handovers. Different research approaches try to use movement predictions as an addition to classical handover preparation and triggers. They vary from statistical analysis [1, 2] up to handover preparation based on cross-layer optimization [3 4]and complex pattern detection algorithms [5]. However, the weaknesses of those mobility techniques result from their insufficient cost to performance ratio. A simple handover prediction approach is proposed in [6], in which simple moving average for inertial movements and simple mobility pattern matching non-inertial movements. However, almost all works about mobility prediction ignore the fact that users movements are not completely random. Especially in the case of cellular networks this information could be used to optimize handover algorithms. In this paper, a typical urban scenario is considered where the lives of the majority of the residents of the existence of strong regularity; performance in the mobile often with a strong purpose. With the Characteristics of the users mobile of the historical records and the purpose-space point of the mobile
This work was supported by Beijing Municipal Commission of Science and Technology and Beijing University of Post and Telecommunications under 4G projects (Do8o100620802) and supported by National Natural Science Foundation of china(60872050).
Key assumptions The Key assumptions for the proposed mobility model and its analysis are listed as follows. For urban users, whose mobile region only include three points (home workspace and hot) .Hot is the existence region except home and workspace abstract of all existence region (e.g., supermarket, park, square, hospital, etc.). (In this paper, home, workspace and hots are covered by a single cell ) Between the two points which belong to the finite point space set (home, workspace and hot), exist not only a path for the mobile users. Each path has a different weight and markers; and the weight of the path is proportional to the probability numbers of the Users appearing in the path. However, (In this paper, we take the path in which the user appears that the greatest probability, i.e. There is only one path between any two points.) A point which belongs to the finite point space doesnt appear in the middle of the path; the user mobility model describes the law between the two points, and each path across a number of cells of LTE system and the cells have isometric radius.
The simplicity mobile model is shown in Fig.1. In a period of time, all of the user mobility actions include to the regular and irregular action. First let denote the proportion of user mobility in regular terms. A high value of (i.e. >0.7)
TABLE I.
User-ID
Cell-ID
Update-Cycle
Friendly-Rank
The number of the user appear in the cell n Average Cell dwell time The number of handover of the user in the cell The number of Cell reselection of the user in the cell
Figure 1. The simplicity user mobility model
The number of handover or reselection from each cell i (i=1,28).(The celli is the neighbor of the cell)
TABLE II.
USER MOBILITY DATABASE
User ID
Figure 2. The process the database system
Hot-A Router(i)
implies that a high proportion of users mobility is in regular, whereas a low value of (i.e. <0.3) corresponds to users mobility in irregular. If user mobility rule set denotes:
i= N i =1
Proute ( i ) =
employ the route (i). According to the records of the user in database .B , network determines the current pattern of the user is regular or no-regular. If the result of the judgment is regular, the model is valid, i.e. Handover prediction has been simplified and Network performance has been improved. On the country, the model is invalid. B. Personalized mobile user database In the system, the network should recognize the users who frequently visit to a cell, which we call their user-friendly for this cell. And, the network should track and record the movement information of the user-friendly. The MME Calculate on the records, and extracts the user's mobile law at intervals in a period. In this paper, we establish a database to utilize the users mobile law for each user, which is used to record the user's mobile behavior rules. The Process of the system is shown in Fig.2. Initial Records of the database is taken from a table which is finished by the user, when he/she enroll into the networks. The table should include the user's home address and the user's working conditions and his/her preferences, and then according to these information, the approximate laws of the mobile user can be extracted for mobile users database initialization. However, the data denotes the rule of the users mobility just only network guess on the basis of the information which the user submit, not the rules of the user in real life. Consequently, network should track and record the users locomotion activity in a relatively long period, and use the records to optimize the database. Table 1 shows the records of the mobility behavior of a user in real life. And with the records further processing, the results of the processing consist of the basic data of the user mobility database (is shown in table 2). Hot-A (Hot-B, Hot-
Cetc.) denotes the Cell-ID of the cell in which the user often appears. The probability of the user appearing in and the user detach the route, are represented as and , respectively. In this context, = Proute(i ) .The results of optimization should to the greatest extent simulate the rules of the user mobility behavior in his/her everyday life. C. The method to update the database The different law is shown by the users moving at intervals, and the network updates the database when necessary. The network automatically updates data (add or clear the route and cells) at regular intervals or by unscheduled event trigger. Fig.3 illustrates how the database updating functionality works. III. THE HANDOVER TECHNIQUE DETAILS
A. The users mobility pattern to decide The completion of the users mobility pattern to decide action is in the side of the network; it doesnt need interaction between the side of the user and network. e.g.; the information of the reselection or handover can include the locating information of the previous cell. According to the information and the location of the user, network can make use of the lookup table to search a route to match the route of the current user from the database. The users mobility pattern to decide the process is represented in Fig.4. The users movement action starts from some a point which belongs to {home, workspace and hot}, after N cells, and ends up with another point. In the process, the results of the table lookup consist of the optional set {route (i)}, and the next cell in each route consists of Potential handover set
e.g. From event A1 occuring to event A2, UE measurement total 5 times. The value of Ec N 0 is v1 , v2 , v3 , v4 and v5 , respectively. According to the comparative result decide the event 1D. Reporting event 1D occurs when the comparative result is gradually decreased, and the CPICH Ec / N0 < Thaway (of
Ec N 0 < Thpre1
move pattern is regular and for the PHO ( route (i )) next reporting event 1D doesnt occur, contemporary a cell CPICH E c / N 0 > Th pre 2 , where and if all PHO ( route ( i )) C HO ( ss )1
Max [(1 ) Proute
(i)
received energy chip divided by the power density of a cell ci , denoted by E c / N 0 ( c i ) , of the common pilot channel (CPICH) is used as the signal strength. New reporting events are introduced to send measurement information for handover prediction as follows. Reporting events A1 and A2 occur when the CPICH E c / N 0 of the current cell becomes smaller than a certain threshold and
ci is assigned to bHO . If the current move pattern is regular and reporting event 1D occurs for all PHO ( route (i )) , then
next
Thalert
inform the UE to start the measurement, record E c / N o and periodic send the report each time for PHO until it receives a measurement report caused by an event A1 from that UE. If the reporting event A2 occurs, eNoteB starts comparing the record,
moving terms is random. Early works [6] proposed many valid methods, and this paper offers some useful supplementary methods. IV. NUMERICAL RESULTS AND DISCUSSIONS
There are two sides used to measure the performance of the proposed technique which are known as the number of handovers and the rate of Ping-pang handovers. And we first tested the fast handover procedure with the proposed technique. Then for performance evaluation of the proposed technique we also implemented LTE system without the proposed technique and made a comparison between two systems with Tau performance simulator [7]. A. simulation Models The simulation model of the cellular network area consists of 28 macrocells that have the same radius of 1km. The signal level the UE received from eNodeBi , Ec N0 (ci ) , is calculated
proposed technique when the hysteresis threshold is small as shown in Fig.6. The more regularity of user mobility, the greater performance gain has been gotten with the proposed technique as expected. At = 0.9 and = 0.05, the proposed technique has gotten 30% performance gain. But, at = 0.5 and = 0.2 the proposed technique obtains almost no any performance gain. This is because the proposed technique is only for the greater regularity mobility. The rate of ping-pong handovers shows similar results. In Fig.7, we fixed the time condition fixed 60s for ping-pong handover; if the stochastic of user mobility was increased, the rate of ping-pong handovers was also increased as expected. The handover with the proposed technique showed lower rate of ping-pong handovers for every ping-pong handover condition. This result shows that the proposed technique is also useful to reduce unnecessary and ping-pong handovers owing to the correct handover prediction. V. CONCLUDING REMARKS
di is the
distance between the UE and eNodeBi , and N i is the sum of zero-mean gaussian background noise and interference noise. We choose six UEs with the different mobility laws to test the proposed technique; detail parameters as shown in Table 3. simulation results As for the number of handovers, the proposed technique was about 10% better than that of the handover without the
TABLE III.
USER MOBILITY LAWS PARAMENTERS
This paper proposed a handover optimum approach with simple handover prediction based on acts of users of mobile history. The proposed prediction technique uses the user mobility database and simple mobility pattern matching as well. A valid update database approach is proposed, to ensure the database time in tracking the user mobility actions. The numbers shows that the proposed technique has a smaller number of handovers and lower rate of ping-pang handover than the handover process without the proposed technique. The comparison with other prediction techniques and further comprehensive performance analysis on the handover preparation are now in progress for refining the proposed technique. REFERENCES
[1] Roy,A.;Das,S.K; Misra,A,Exploiting Information Theory for Adaptive Mobility and Resource Management in Future Cellular Networks,IEEE Wireless Communications,2004 , pp.59-65. Cheng,C.; Jain,R.; Berg, E.v.d.:Location prediction for Mobile Wireless Systems,in Furht,B.(Hrsg.). Wireless Internet Handbook,CRC Press,Boca Raton ,2003 , pp.245-264. J. McNair,et al., A survey of cross-layer performance enhancements for Mobile IP networks, Computer Networks, Volume 49, Issue 2, Elsevier,2005,pp.119146. J. Jung et al., Mobility Prediction Handover Using User Mobility Patternand Guard Channel Assignment Scheme, LNCS, Vol.3264, 2004. Poon, W.T.; Chan, E. Traffic Management in Wireless ATM Network Using a Hierarchical Neural-Network Based Prediction Algorithm, Proceedings of the International Conference on Computers and their Applications, ICSA ,2000. Tae-Hyong; Qiping Yang; Jae-Hyoung Lee; Soon-Gi Park;Yeon-Seung Shin;A Mobility Management Technique with simple Handover Prediction for 3G LTE Systems Vehicular Technology Conference,2007.VTC-2007 IEEE 66th,2007,pp 259-263. Telelogic AB Inc., Telelogic TAU SDL Suite Ver.5.0, 2006. See http://www.telelogic.com
UE
[2]
[3]
[4]
[5]
[6]