You are on page 1of 34

Protocol Requirements for Cognitive Radio Networks

Physical, data and network layer protocol design in cognitive radio environment

Colophon
Date : Version : Change : Project reference: Freeband reference : Company reference : URL : Access permissions : Status : Editor : Company : Author(s) : Reviewer(s):

20 July 2005 0.2 1.0 AAF/D4.11 AAF /D4.1 N/A Project Deliverable TU Delft Przemysaw Paweczak Mauritius de Graaf (Thales), Anthony Lo (TU Delft), Liang Xia (TU Delft)

Synopsis:
Protocol design requirements and constraints for emergency cognitive radio networks

Copyright 2004 Freeband AAF project partners Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from or via Freeband Communication Management (http://www.freeband.nl).

Abstract
This document focuses on physical, link, network and transport layer protocol requirements for cognitive radio ad hoc networks which should be met to optimally exploit available radio resources and simultaneously provide all the communication services required in emergency situations. We outline specific performance measures which should be considered first while designing appropriate protocols. In this report we also try to state which already designed protocols for wireless ad hoc networks might be implemented in the AAF network without major adaptations. We conclude that the recommended approach for designing protocols for cognitive ad hoc networks is throughout cross layer design.

AAF/D4.11

Preface
The AAF project is part of the Freeband Communication programme, which aims at the generation of public knowledge in advanced telecommunication (technology and applications). Freeband is based on the vision of 4G networks and services. It specifically aims at establishing, maintaining and reinforcing the Dutch knowledge position at the international forefront of scientific and technological developments, addressing the most urgent needs for research and novel applications in the present unfolding of new technology. Freeband comprises more than 25 organisations, including all-important technology providers and many representative end-user organisations. The Dutch Ministry of Economic Affairs is co-funding this programme as part of the BSIK plan. The vision for Freeband for 2010 is to consider communication and information transfer from the perspective of the user, not the provider. The communication infrastructure will become transparent and abundant in all its layers. Freeband addresses the knowledge chain in communication in the direction of the new ubiquitous communication paradigm. Based on this vision key research questions take place in three main themes: Society, Users and Applications: what are the new possibilities in different sectors for ubiquitous communication and ambient intelligence, what do they presuppose as knowledge and how can they be realised? Networking, Service Provisioning and Generic User Interaction: the telecommunication infrastructure viewed from the user's perspective. Enabling Technologies: no new services emerge without adequate technology; conversely, it is the technology that drives the new paradigms! Crisis situations require fast regain of control and todays emergency systems like for example C2000 hardly support data communications. Cellular systems e.g. GPRS, EDGE, HSCSD are not reliable enough in emergency situations. Moreover data communication requires significant radio spectrum, which today is divided into small pieces where each piece is available only for specific group of users. Regulatory bodies in some countries, like FCC in United States, already recognise that new services require completely new approach for spectrum management. Cognitive Radio is such a paradigm shift in spectrum utilisation. This project Adaptive Ad-Hoc Freeband communications (AAF) researches its key technical and legal aspects and realises a working solution of a cognitive radio for emergency services. This document is a deliverable D4.11 of the AAF project. It focuses on the high-level description of protocol design requirements and constraints for emergency cognitive radio networks. It will be regularly updated.

AAF/D4.11

VII

VIII

AAF/D4.11

Table of Contents
1 Introduction 1.1 1.2 1.3 2 Initial assumptions Reasons for this document Structure of the document 1 1 2 2 3 3 3 3 5 5 5 5 6 6 6 7 8 8 9 9 9 11 17 17 18 19 19 19 20 21

System description 2.1 2.2 2.3 Hierarchy in emergency network Required communication services Required functional services

Protocol requirements for CRN 3.1 Routing 3.1.1 Adaptation of existing ad-hoc networks routing protocols to cognitive radio networks 3.1.2 Track one: Cognitive System 3.1.3 Track two: Cognitive Radio System 3.1.4 Requirements for routing 3.1.4.1 3.1.4.2 3.2 3.3 3.4 3.5 Robustness (multiple routes analysis) Proactive, link-state and hierarchical routing as a proposal for routing paradigm

Localizing capabilities Addressing Transport Medium Access Control Discussion

3.5.1 Requirement for separate common control channel 3.5.1.1 3.6 3.7 3.8 3.5.2 Requirements for traffic channels Security Software distribution Additional requirements not connected explicitly with protocol design

3.8.1 Cognitive radio channel propagation model 3.8.2 Co-channel interference 4 Conclusions References

AAF/D4.11

IX

1
1.1

Introduction
Initial assumptions

Cognitive radio (CR) is a promising technology and a new paradigm shift in communication. It tries to utilize free parts of unlicensed spectrum and even uses licensed frequency bands during silent periods of primary, licensed users. It has to be emphasised that term cognitive radio, first coined by Mitola and Maguire Jr in [12] has different meaning that term cognitive radio used in AAF project. Mitola in his PhD thesis [13] explains: The term cognitive radio identifies the point in which wireless personal digital assistants (PDAs) and the related networks are sufficiently computationally intelligent about radio resources and related computer-to-computer communications to detect user communications needs as a function of use context, and to provide radio resources and wire less services most appropriate to those needs. It is a vision of an intelligent wireless black-box with which user travels. Wherever the user goes, cognitive device will adapt to new environment allowing user to be always connected. FCC on the other hand describes cognitive radio as a system which could negotiate cooperatively with other spectrum users to enable more efficient sharing of spectrum. A cognitive radio could also identify portions of the spectrum that are unused at a specific time or location and transmit in such unused white spaces, resulting in more intense, more efficient use of the spectrum while avoiding interference to other users [14]. Here cognition cycle [13] is performed only in radio domain and no other sources of information are used to make decision about a change of transmission parameters.

Figure 1.1: Sources on which decision is made in two proposed cognitive radio systems A) Mitola radio B) FCC cognitive radio

This simply means that FCC approach is a simplified view of Mitolas cognitive radio where only one parameter is considered while making decision about future transmission and reception (see Figure 1.1). In AAF project we use term cognitive radio in the way FCC describes it. This means that each node working in the AAF network, with the respect to its cognitive capabilities only spectrum sensing is supported and will be able to change only its physical parameters (without possibility of changing whole architecture completely by means of i.e. software download). So wherever we use term cognitive radio we mean FCC cognitive radio device.

AAF/D4.11

Another distinction which has to be made is about possible parts of spectrum to choose from for transmission by cognitive radio node. When we assume during design process that only unlicensed bands are available for the radio node (unlicensed band cognitive radio) then most of the constraints discussed in this document will become less strict e.g: no need for deferring when new user is sensed because each one has equal rights for accessing the channel. One of the examples of such system is the IEEE 802.15.2 [19]. However when we want to utilize also licensed bands (licensed band cognitive radio) then design process will become much more complicated due to constant control and management of licensed band and large set of upgrades to existing protocols. Work on coexistence of various standards is ongoing e.g. in IEEE 802.19 group [20]. Track II of the AAF project focuses on licensed band cognitive radio. One of the positive sides of such approach is that future designed protocols for licensed band cognitive radio will work efficiently also in unlicensed bands.

1.2

Reasons for this document

To make cognitive radio a reality we must introduce various changes to existing protocols and design new ones. Simultaneously we are designing an emergency system in which specific communication parameters should be fulfilled. Those parameters should be reflected in implemented protocols. This document aims at pointing out those constraints.

1.3

Structure of the document

This document is organized as follows. Chapter 2 briefly describes system requirements and outlines communication services for emergency devices. Chapter 3 is the main part of the document which states what specific parameters should be fulfilled and how the protocol structure should look like to make cognitive ad hoc network reality. Chapter 4 concludes the document.

AAF/D4.11

2
2.1

System description
Hierarchy in emergency network

In [2], [5], [6], [18] authors divide network into three functional components: Extended Area Network (EAN), Jurisdiction Area Network (JAN) and Incident Area Network (IAN). Since EAN and JAN might be build on wire infrastructure and based on existing protocols, AAF WP4 will focus only on the protocol requirements for nodes in IAN domain. More reasons for this approach can be found in [18].

2.2

Required communication services

In AAF project we require a finite set of services. Since each next-generation communication system for emergency situation has the same requirements, no matter on what physical interface build, this list is in line with the communication services presented in [5]. The required services for emergency communications are:

Video call: used to communicate between emergency workers and between emergency workers and
control centre. Video call constitutes of simultaneous image and voice transmission. By supporting video call we simultaneously support still picture transmission which will inform control centre about current situation on the disaster scene.

Voice call: Standard communication service. Each call should have various priorities assigned to
packets transmitting voice traffic.

Text messages and alarm services: Text messages will be used to inform emergency nodes about
minor events. Alarm services will be sent through text messages with highest priority. More detailed description of required services can be found in [44].

2.3

Required functional services

Below we outline most essential functional parameters each CR node should have. Those requirements are a basic substrate for protocol design.

AAF/D4.11

Localizing capabilities: It is compulsory to provide each emergency node with localizing capabilities
because command centre has to know the exact positions of each emergency worker and information about node position should be utilized by routing protocol and spectrum sensing algorithm. One of the approaches is to use GPS service by each CR device. Unfortunately this solution has two important ramifications: first it has rather low accuracy (measurement mismatch might reach meters) and second GPS signal propagates poorly in indoor environment (for other reasons the reader is referred to [38]). Some localizing techniques for ad hoc networks have been proposed which for example utilize information from the network layer [38], assume that some of the nodes know their position [39] or know the distance between nodes [40]. However when we decide that common control channel (see below) will be based on UWB technique than each node will be capable of measuring the position by itself with high accuracy.

Distributed sensing: In [41] authors through simulations have proved that cooperation in spectrum
sensing between CR nodes decreases probability of interference to primary users. Since AAF CRN cannot interfere with licensed users too this requirement is essential for exact operation of our designed network (one of the performance measure that has to be assumed in this project is the value of sensing error probability).

Robust control channel: Since CRN node has no certainty that the channel in which it communicates
presently will be available in the future, reference channel must be available through which nodes will exchange information about occupied channels. We must remember that when control channel will be unavailable then whole network will become inactive.

No clock synchronization: Clock synchronization is very hard to achieve in ad hoc networks so we


assume that all designed protocols will work without any synchronization.

Immediate channel switching: The time in which CR node will switch from one channel to another
has to be minimal such that the latency introduced by switching might be negligible for all protocols above physical layer.

Please note that aforementioned requirements are given from the protocol point of view. Requirements from system perspective for a mobile AAF node can be found in [44]

AAF/D4.11

Protocol requirements for CRN

Below we describe required features each Cognitive Radio Network (CRN) node in IAN should have.

3.1

Routing

3.1.1 Adaptation of existing ad-hoc networks routing protocols to cognitive radio networks According to [4] the work in the AAF project will fork into two tracks named Cognitive System and Cognitive Radio System. First track aims at results/products on a short time notice and focuses on existing standards IEEE 802.11 and IEEE 802.16. Research will be made only on layers above link layer and work will concentrate on utilization of existing cognitive capabilities of aforementioned standards (however it is questionable whether such capabilities already exist since IEEE 802.11k group responsible for radio resource measurements in WLAN is still working on a draft version of its document). Second track focuses on long term scale and just like in track one utilizes both IEEE standards. However it will try to alter many parameters in physical and link layer such that full radio cognition is met. Below we will distinguish those two tracks while describing issues of routing protocols adaptation to cognitive radio ad hoc networks.

3.1.2 Track one: Cognitive System We can exploit the information from physical and link layer to help routing protocol in making various routing decisions. By exploiting radio layer information routing protocol can:

Differentiate routes depending on channel type due to changing propagation characteristics of


various radio links. This leads to better QoS when compared to algorithm taking into account number of hops only.

Increase nodes connectivity due to wider set of available radio links and available longer
transmission distances: any cognitive radio node is capable of transmitting with broad set of frequencies, i.e. UNII and USM band [4] or UNII, USM and TV band [1]. By utilizing simple measure that the higher the frequency the shorter the transmission distance, routing algorithm may decide which radio link should be used for specific hop. It has very important implications to emergency network since high frequency signals have bigger problems with thick objects penetration. It is why routing has to utilize the channel information and send high priority packets on highly resilient channels (lower frequency channels).

Detect faster link failiures. Perform more efficient multicast due to increased connectivity.

AAF/D4.11

3.1.3 Track two: Cognitive Radio System It is already known that physical and link layer protocols designed for standard fixed bandwidth ad hoc networks must be changed and adapted to cognitive radio environment to effectively utilize spectrum information [2]. The role of those modified layers of the protocol stack is to manage radio resources in the way appropriate for the nodes in the whole CRN. The remaining layers might be adapted explicitly to cognitive radio networks. Indeed in [1] authors claim that higher layers [above link layer] will implement standard protocols not specific to cognitive radios. However it is valuable to examine in the AAF project the impact of cognitive radio capabilities on routing protocols in ad-hoc networks (application layer is beyond the scope of the AAF project). Especially the project should answer the question what is the benefit for routing protocols from introducing cognitive capabilities to network nodes in terms of:

Time constraints: route setup time and end-to-end latency; Casting issues (multicast, broadcast, geocast [5] and unicast); Throughput: overhead value, overall transmitted traffic value, packet loss value; Route quality: route length, route discovery and reconstruction time.
Above mentioned performance metrics is based on parameter list presented in [3].

3.1.4 Requirements for routing Apart from hypothetical changes to existing ad-hoc network routing protocols network layer in IAN should have features common to any routing protocol in any emergency network, for example [5], [6]. Below we will outline them in detail.

3.1.4.1 Robustness (multiple routes analysis) Emergency nodes cannot loose a connection with the IAN. Each node sends crucial information to the JAN and loosing it even for a small quant of time has far reaching consequences (more detailed reasons for the connectivity constraint are outlined in [2]). To perform simple analysis of impact of multiple routes on routing protocol robustness we will model a network in terms of reliability theory [11]. Suppose that failure probability of one link x between two nodes is given by standard exponential density function F (t ) = 1 exp( x t ) , where x [0,1] is the failure rate of the link x and t represents time. To
x

increase routing robustness we introduce another link y with the same failure probability F by Fs
x, y

(t ) . When

assuming independent failures of link x and y overall path failure probability between two nodes is given

(t ) = 1 exp((x + y )t ) (we omit mathematical derivation of the expression). For a general case

of routing packets between two nodes simultaneously through n routes when each route has equal failure

AAF/D4.11

F x (t ) overall path failure probability is given by Fsx , y ,..,n (t ) = 1 exp( it ) . We now i =1 x , y ,...n introduce gain function Gn ,n1 (t ) = Fs (t ) Fsx , y ,...n1 (t ) which states what would be the benefit from
probability introducing another redundant route to the path between two nodes Figure 3.1 depicts gain functions for six-route path.

0.25

0.2

0.15 Gn,n-1(t)

0.1

n n n n n

= = = = =

2 3 4 5 6

0.05

10

20

30

40

50 t

60

70

80

90

100

Figure 3.1: Set of gain functions; n=6, n=0.02.

It is easily observed that one additional route to the path between nodes increases reliability by certain extent. However the more routes we introduce to the path the smaller the gain is. Abovementioned analysis is very simplified (exponential distribution representing link failure is not a good reflection of reality because usually link state does not decrease with time). Nevertheless this analysis outlines three important features which robust routing technique should have:

Provision of only one additional route to the path strongly decreases failure probability; Number of routes must be choose optimally in the sense that even small number of additional routes
provide good robustness;

Routing algorithm must choose paths optimally since most of the paths will have different n parameter
(if more routes traverses through the node the higher the probability that it might fail e.g. due to excessive battery exploitation).

3.1.4.2 Proactive, link-state and hierarchical routing as a proposal for routing paradigm Routing protocols designed for emergency ad hoc networks should have three essential features. First, routing protocol should be proactive. It means that every node maintains up to date routing table in which

AAF/D4.11

path to every destination in the network is stored. The reason behind that constraint is that update of routing tables in reactive protocols is time consuming. In emergency situation every packet should be transmitted with minimum latency and exchange of signalling messages to find a route to a new emergency node in IAN might take time. Unfortunately proactive routing protocols introduce higher signalling traffic in comparison to reactive ones. Moreover constant updates of routes information decreases battery lifetime. However as [7] states reactive routing protocols maintain multiple routes that might never be needed and we find it as a very valuable feature. Second, routing should be based on link state because of rapidly changing physical layer parameters not only due the mobility and transmission range changes but especially because of cognition capabilities of the nodes. It means that paths between nodes should not only be computed in terms of number of hops but also in terms of such parameters like link reliability. This assumption is in line with the features of cognitive system (see paragraph 3.1.2). And third, routing should introduce hierarchy (in terms of emergency group organization) to support various mobility groups. Thanks to this feature routing could be performed more effectively between rescue groups (for example fire brigades may have higher mobility than police). One of the candidates that might support all those features is LANMAR protocol [7]. Another candidate is the OLSR protocol, strongly supported by IETF [37]. Although above features should be obligatory we might find other classes of routing protocols valuable for emergency scenarios like for example hybrid protocols such as ZRP [3] which combines features from both proactive and reactive routing protocols.

3.2

Localizing capabilities

We have already outlined the reasons for introducing localizing capabilities of emergency nodes (see section 2.3). Knowing the physical position of the node we can route packet directly to the area where the specific node could be found. Thanks to this characteristic it is not necessary to maintain large routing tables. With the four aforementioned requirements for routing we might find a hybrid location aware routing (also known as location proxy routing protocol or Grid routing [8]) a valuable solution: routing on long distances will use geographic coordinates and short distance routing will use link state routing tables.

3.3

Addressing

Nodes addressing should be performed automatically through decentralized DHCP servers. Addressing scheme should be in line with the IPv6 protocol (but this is not a strict constraint). It is believed that IPv4 will be superimposed by IPv6 around 2025 [10], which is also a probable date for introduction of ad hoc emergency networks to the market. The main drawback of this assumption is that each node should have gateway capability of translating IPv6 packets to IPv4 and vice versa (since some nodes might still be in line with IPv4 addressing only). AAF project should answer the question what is the introduced delay by IP

AAF/D4.11

packets translation gateways. It should be also answered if it is obligatory for each node to have a gateway capability, just like we have stated before, or only one gateway node (exactly the same which is responsible for connection with JAN) should be the IPv4/IPv6 translator and all non supported connections should be routed to it.

3.4

Transport

Although the research in AAF project focuses on first three bottom layers of the protocol stack it is worth knowing what is the behaviour of TCP protocol in cognitive radio environment. The authors in [7] identify two important factors that influence performance of TCP in wireless ad hoc networks: mobility and MAC performance. In both cognitive and non-cognitive ad hoc networks nodes mobility occurs so both types of networks will perform identically when considering this parameter in isolation. The main difference is in medium access control procedures. We can identify TCP throughput as an important factor in ad hoc networks so it should be also identified during research. MAC protocol designed for IEEE 802.11 and IEEE 802.15 will decrease throughput of simultaneous TCP flows since only one flow can have an access to the channel at a time. But we have already stated that crucial changes in protocol stack will be performed in physical and link layer. Those changes will decrease channel occupation probability and simultaneously will increase number of simultaneous TCP flows. We might improve TCP performance by informing transport layer about the link status by means of some signalling protocol (e.g. ICMP [7]). This approach is also equal to cognitive system approach (see paragraph 3.1.2) where each layer cooperates with other.

3.5

Medium Access Control

Below we will describe some of the multi carrier MAC protocols and outline their important features relevant to cognitive radio network.

3.5.1 Requirement for separate common control channel Below we outline some most important factors which should be considered during design phase of common control channel:

While using 802.11 MAC scheme RTS/CTS control packets in separated control channel packets will
encounter less number of collisions (DATA and ACK packets will only be transmitted in group frequency bands [1]). Moreover we will encounter smaller probability of hidden and exposed terminal phenomenon. Whats also important is that all group bands are assigned for data traffic transmission so we will encounter no throughput decrease due to transmission of signalling traffic.

AAF/D4.11

Dimensioning of control channel: The gain from introducing more traffic channels to CRN is not
linearly proportional [32] due to the fact that common control channel becomes saturated with signalling packets. We might construct dynamic common control channel which will have different bandwidth depending on the current number of users in all bands in CRN. Such behaviour will decrease overall SINR (Signal to Interference Ratio) in traffic channels (assuming that UWB channel resides in the same part of the spectrum as most of the traffic channels in CRN).

Time of traffic channel switching in CR node must be negligible: Channel switching cannot
introduce additional latency to transmission (see also paragraph 2.3).

Each CR node can only transmit or receive at the same time and can either send/receive data or
send/receive signalling: This assumption comes from the fact that usually not every node can have spectrum sensing capability (in other words equipped in two antennas one for transmission/receive and one for spectrum sensing/signalling

Either UWB or dedicated ISM band: Using a UWB as a common control channel has this advantage
that we can simultaneously provide nodes with the information about position. Such information might be used by all of the protocols (especially by routing and MAC). See below.

Use of localization capabilities for cognitive radio: We should utilize information about position of
the node in the MAC protocol. One of the proposed schemes: if nodes know the exact locations of transmitting stations of the licensed bands (this information will be stored in node's internal memory and periodically updated by means of software download) it can always compare its position with stored coordinates and utilise this information as one of the criterion in frequency selection process (see below)

Multiple criterions for channel switching: The one and only approach for design of physical and link
layer is through cross layer design because cognitive radio MAC (CR MAC) protocol must utilize information from physical layer to assign resources for wireless nodes. Assigning policies must be based on various requirements and data for making decision must be provided only by physical layer. Some of those policies are: usage diversity, spatial reuse, common frequency use, minimize use of licensed bandwidth. Table 1 presents most recent Multi Channel MAC protocols proposed in the literature.

Assigning particular bands groups to particular group of users: We define a band group as a set
of frequencies which might be utilized by only some users due to the construction of RF front end.

Even when saturated common control channel must inform about licensed users occurring in
the area: Every packet about primary user of the used licensed band must be transported immediately to each cognitive node in the receiving vicinity of the primary node (node that has always the right to access the channel as the owner). When we assume that information about occupied channels is transported through CSMA/CA protocol of the IEEE 802.11 we should construct a signaling scheme in which packets with information about available licensed users must be always served first no matter how many other packets are waiting in the queue.

10

AAF/D4.11

Allow cooperation between legacy ad hoc equipment: We should construct protocols which should
allow communication between legacy equipment (IEEE 802.11 nodes) and CRN nodes. This idea is open for discussion. Moreover we assume that not every node in CRN has sensing capability so we have to periodically inform some nodes about available channels.

Enhanced Multicasting capability: TBC


3.5.1.1 Discussion First we remark that N antenna protocols (for example [30], [31]) are not applicable for a CR emergency node since they are energy consuming and may lead to increased size of the node. Second, we have already emphasized that clock synchronization in ad hoc network is difficult (see paragraph 2.3) so we cannot use MAC protocols which exchange packets in predefined period of time (MMAC). So by the rule of elimination the only multi channel MAC protocol which we might consider as a base for a research in AAF is DCA and Hui et al. algorithm.

Figure 3.2: Throughput versus the distance in UWB and IEEE 802.11a/g [43]

The elementary adaptation will consider forcing nodes from leaving licensed band. That is why we might introduce two levels of RTS packet: RTS which inform about licensed user availability always transmitted with highest priority and with lowest possible inter-frame space timer, and standard RTS frame. Because in legacy IEEE 802.11 CSMA/CA scheme transmission is continuous, which makes inter-packet sensing difficult, a sender can divide its sent data into small segments. Between each segment transmission node will perform channel sensing if licensed user is already available [36].

AAF/D4.11

11

Another possible approach, not listed in the table, is to utilise frequency hopping [28], [29] in which each node changes frequency channels with a predefined pattern. When two hosts agree to exchange data with a RTS/CTS scheme they stop hopping and communicate on the negotiated channel. Decision about a channel for communication can be performed either by a sender or receiver. In the proposed protocols nodes have only one half-duplex transceiver. Unfortunately this approach cannot be utilized in CRN due to the fact that each of the nodes does not know how many channels are available in particular time. Without that information frequency hopping is practically impossible. We have assumed that common control channel will be built on top of UWB physical layer. However it has to be noted that throughput of UWB decreases heavily with distance [42], [43] (see also Figure 1.1). This would simply mean that maximal distance between neighbouring nodes of CRN will be limited to UWB transmission distance ergo less than 10 m. Such limitation is caused by low power levels in which UWB nodes are allowed to transmit. This topic is open for discussion.

12

AAF/D4.11

Table 1. List of main multi channel CSMA MAC protocols

Name of the protocol /author

Number of transceivers

Number of channels and their size

Separated control channel its and size

QoS support

Parameters considered during channel negotiation

Mobility support/considerat ion

Power control

Who makes the decision

Remarks

Hui et al. [15]x

2 half-duplex

N/fixed

Yes/fixed

Through dynamic control channel, differentiated IFS and adaptive backoff time

Code assignment list: ID of the neighbour node, code used by neighbour node, type of service, channel occupation time; free channel list

No

No

Receiver

MAXM [16]

TBC

N/fixed

No

TBC

Neighbour ID

No

TBC

TBC

AAF/D4.11

13

DCAPC [17]

2 half-duplex

N/fixed

Yes/fixed

No

Power vector, channel usage vector

Only in simulations

Yes

Receiver

MMAC [27]

1 half-duplex

N/fixed

Yes/fixed

No

List of channels with least scheduled traffic; channels can have one of three priorities; number of pending packets per node (but not considered in the protocol implementation)

Only in simulations

Yes (but not considered in the protocol implementation)

Receiver

Needs clock synchronization; authors discuss possible enhancements for multiple flows transmission

ODC [35]

1 half-duplex

N/fixed

No

No

Depending on the traffic conditions in each channel; two new packets as an extension to IEEE 802.11: DEP and ARR

No

No

Transmitter

14

AAF/D4.11

Naspuri, Zhuang and Das [30]

N half-duplex

N/fixed

No

No

Node is searching for an idle channel with all its antennas; if total received signal strength is below a certain threshold then channel is marked as idle, among multiple idle channels the last used is preferred

Only in simulations

No

Transmitter

Exchanged packets are not IEEE 802.11 compatible

Naspuri and Das [31]

N half-duplex transceivers

N/fixed

No

No

TBC

TBC

TBC

transmitter

DCA [32]

2 half-duplex transceivers

N/fixed

Yes/fixed

No

Preferred channel list is sent by transmitter with RTS packet, receiver responds with the channel number on the CTS packet

TBC

TBC

Receiver

AAF/D4.11

15

Jain et al. [33]

N half-duplex transceivers

N/fixed

Yes/fixed

No

Transmitter sends a list of free channels visible by him and sends this information to receiver in the RTS packet; receiver is also performing sensing and compares its list with send by the receiver and chooses one of those which is comparable with his own.

Only in simulations

No

Receiver

Extension of [30] in sense it transmits IEEE 802.11 like packets

PCAM [34]

3 half-duplex transceivers

N/fixed

No but protocol assumes additional channel for receiving and transmitting broadcast information

No

No channel negotiation is performed; each node has pre-assigned primary channel and each node knows the number of channel on which it will communicate with the others; secondary channel is not fixed and it is used for switching between N possible channels

Only in simulations

No

N/A

Protocol assumes that distributed protocol exists which pre-assigned channels for each node; broadcast channel is used to exchange AODV routing protocol datagrams

16

AAF/D4.11

3.5.2 Requirements for traffic channels Below we outline some most important factors which should be considered during design phase of traffic channels:

Differentiated classes of service by means of IEEE 802.11e: This document explicitly states how
many different communication services are envisioned to be used in the emergency CRN. Each of these services has specific priority attached to it which should be reflected in the MAC protocol. Table summarizes some of the protocols that support different classes of services. Unfortunately none of those protocols tries to use existing IEEE 802.11e solution.

Power control: Due to the fact that no synchronization is allowed in the network we cannot assume
sleeping mode (doze mode when using IEEE 802.11 terminology) because nodes will not know when to receive beacons. On of the proposed solution is to assign different transmitting powers to different packets. Moreover we can periodically change transmitting powers of some of the data packets [36].

Mobility support: There is still a lack of MC MAC protocols which consider different types of nodes
mobility. One of the research areas in AAF WP4 should be on the impact of various mobility types on elementary performance metrics (throughput, latency).

3.6

Security

One of the factors which should be considered during design process of CRN emergency network is security of the network infrastructure and security of transmitted information. Without proper network security terrorists responsible for the disaster would be able to eavesdrop emergency information and utilize it for future attacks. Moreover the network elements due to their poor security could become a target of attack itself. Because cognitive radio constitute a new approach for building wireless networks it simultaneously opens a door for new methods of attacks on their physical structure. Below we outline some of the possible methods of attacks on CRN and ways of prevention:

Licensed user emulation attack: Because cognitive radios cannot be completely sure whether a
licensed spectrum is free and available for transmission they simply defer from licensed bands and utilize other non-licensed parts of the band if they are not sure if it is really free. Suppose that attacker knows in which specific area CRN works. Knowing which licensed bands CRN might use attacker can simply transmit signal in the licensed band emulating real transmission and thus limiting overall CRN capacity. Until now we don't know any method of prevention against this attack.

AAF/D4.11

17

Common control channel jamming: One of the possible solutions for common control channel
deployment is the UWB. In this case potential attacker can simply transmit periodical pulses which have the same spectrum as common control channel of CRN but with higher power than legitimate users. Throughout jamming of just one channel attacker blocks the possibility of communication between all CR nodes. This is the reason for building sophisticated UWB transmission methods for control channels utilizing UWB. It has to be underlined that a need for special care of control channel is the same for any type of approach (dedicated channel, channel hopping etc.).

Attacks on spectrum managers: We cannot allow having one central spectrum manager responsible
for assigning frequency bands for nodes (see paragraph 2.3) because it constitutes a single point of attack. Whenever the spectrum manager is not available for CR nodes the communication process becomes impossible. That is why information about spectrum availability should be as distributed and replicated as possible. This constraint is in line with the requirement for more accurate measurements of spectrum availability (see paragraph 2.3). One of the preventing ways for this attack is to use specific pilot channel in each license band. It would inform secondary users about the reservation of the nodes.

Eavesdropping: Usually in the infrastructure-based corporate WLAN it was assumed that signal will
not leave building due to his short distance and will be limited to eavesdropping and sniffing. However cognitive radios are allowed to work in the bands lower than UNII and ISM. This means that they can perform longer transmission distances with the same powers. It also allows for easy physical data collection from locations far distanced from CRN location where attackers invisible to emergency services. This yields a need for strong data encryption at the physical level. Frequent leaving and joining the emergency network must be preceded by authentication process. It is open for discussion which layer should be responsible for this step. Currently the most possible approach is that application layer will perform all the necessary authentication procedures. Moreover the entire WEP infrastructure should be the basis for authentication procedures in CRNs.

3.7

Software distribution

One of the most probable approaches for constructing nodes in CRN is by the means of programmable devices such as FPGAs or ASICs [21], [22]. This makes each software node programmable and allows for simple on-line reconfiguration without any changes in hardware. Usually it is assumed than in specific period of time node does not have a specific capability to perform specific task and must be reprogrammed to fulfil it. This software can be downloaded with the use of three techniques [23]:

Over-the-air software download (OTA): users request for software and download it from software
servers (base stations) or from other users in peer to peer fashion.

Internet download: software defined wireless terminals request for specific software from software
repositories in the Internet

18

AAF/D4.11

Software download from the dedicated hardware: users upgrade their software by connecting their
terminals directly to the physical sources of software like flash drives, optical storage or magnetic disks. We must emphasise that none of those techniques may be exploited in cognitive radio networks. Rapid changes in radio environment do not allow for long time procedures of downloading. Emergency nodes cannot wait for completion of software download process (which might last up to minutes with OTA technique [24]) or know when to install new software from various storages. Software defined terminals have yet another very important ramification. SDR forum describes 7 stages of software update for radio terminal [25] two of them are compilation of software and verification of the completed module. This process also takes time (which depends on the physical architecture of the node) and it is not allowed in the CR nodes. Thus only time constrained software download techniques might be considered. The most promising approach is to download only the parameters for reconfiguration of the terminal [26]. Those parameters will inform CR node which software should be changed. This software will then be uploaded directly from the internal memory of the mobile terminal. Software modules are already tested and verified so the installation time will be drastically decreased. Information about reconfigured parameters will be transmitted in the spectrum sensing frames in the common control channel. However it is still open for discussion whether CR terminals should be based on SDR approach because it might complicate already difficult design process of CRN for emergency network.

3.8

Additional requirements not connected explicitly with protocol design

3.8.1 Cognitive radio channel propagation model It has to be emphasized that in simulations and performance analysis of cognitive radio system we cannot assume that power of the transmission decreases with distance according to the exponential function with the exponent coefficient the same for every channel available for cognitive radio when CRN would utilize TV band its exponential coefficient might be two times smaller than in ISM band. This assumption is crucial for accurate simulation results.

3.8.2 Co-channel interference It has to be discussed whether we assume that two neighbouring channels co-interfere. If we do so then MAC protocol should take this information into consideration while performing decision on channel switching (one of the possible approaches would be to choose channel with maximal distance in terms of carrier for all neighbouring nodes).

AAF/D4.11

19

Conclusions

The rapidly changing radio environment, more radio channels to utilize, number of parameters to choose during decisions taken by MAC and routing protocols, etc. makes design of CRNs very challenging. In this deliverable we have outlined some specific parameters and constraints which have to be taken into consideration while designing protocols for layers above PHY. Many protocols have the same design requirements (like robustness, no clock synchronization or localizing capabilities) which simplify design by small fraction. Moreover we can state that UWB as a common control channel might become a good solution for realizing certain functions outlined in this document. We also outline that cooperation between physical and link layer is essential for accurate operation of CRN. We have to emphasize that new requirements might occur during design process so this document will be constantly updated.

20

AAF/D4.11

References
[1] R. W. Brodersen, A. Wolisz, D. Cabric, S. M. Mishra, D. Willkomm, CORVUS: A Cognitive Radio Approach for Usage of Virtual Unlicensed Spectrum, University of California, Berkley, 2004. http://bwrc.eecs.berkeley.edu/Research/MCMA/CR_White_paper_final1.pdf [2] F. Brouwer, M. de Graaf, H. Nikookar, F. Hoeksema, "Adaptive Ad-hoc Free Band Wireless Communications", Project plan for AAF, Twente, 19 May 2004. [3] C.-K. Toh, Ad Hoc Mobile Wireless Networks, Prentice Hall, 2002. [4] AAF WP Leader Discussion, meeting minutes, Technical University Delft, 24 February 2005. [5] L. Xia, Network Architecture for EUROPCOM, EUROPCOM report, Technical University Delft, 23 March 2005 [6] Safecom: The State of Public Safety Communications, International Symposium on Advanced Radio Technologies, 2 March, 2004 [7] I. Chlamtac, M. Conti, J. J. -N. Liu, "Mobile ad hoc networking: imperatives and challenges". Ad Hoc Networks, vol. 1, no. 1, pp. 13-64. July 2003. http://www.ece.ncsu.edu/wireless/Resources/Papers/adhocSurvey.pdf. [8] G. Pei, M. Gerla, X. Hong, LANMAR: Landmark Routing for Large Scale Wireless Ad Hoc Networks with Group Mobility, IEEE/ACM MobiHOC, 2000. [9] W.-H. Liao, Y.-C. Tseng, J.-P. Sheu, "GRID: A Fully Location-Aware Routing Protocol for Mobile Ad Hoc Networks", Telecommunication Systems (18), 2001. [10] WIkipedia entry: IPv6. http://en.wikipedia.org/wiki/IPv6. [11] Wikipedia entry: Reliability engineering: http://en.wikipedia.org/wiki/Reliability_engineering. [12] J. Mitola III, G.Q. Maguire Jr, "Cognitive radio: making software radios more personal". IEEE Personal Communications, vol. 6, no. 4, pp. 13-18. August 1999.

AAF/D4.11

21

[13] J. Mitola III, "Cognitive Radio: An Integrated Agent Architecture for Software Defined Radio". PhD Dissertation, Royal Institute of technology (KTH), Sweden, May 2000. http://www.it.kth.se/~jmitola/Mitola_Dissertation8_Integrated.pdf. [14] FCC ET Docket No. 03-108, Facilitating Opportunities for Flexible, Efficient, and Reliable Spectrum Use Employing Cognitive Radio Technologies, December 2003. http://hraunfoss.fcc.gov/edocs_public/attachmatch/FCC-03-322A1.pdf. [15] T. Hui, Y. Yang, H. J. Dong, Z. Ping, A MAC Protocol Supporting Multiple Traffic over Mobile Ad Hoc Networks, IEEE VTC Spring, 2003. [16] S.-H. Hsu, C.-C. Hsu, S-S. Lin, F.-C. Lin, A Multi-channel MAC Protocol Using Maximal Matching for Ad Hoc Networks, ACM MobiHoc 2004. [17] Y.-C. Tseng, S.-L. Wu, C.-Y. Lin, J.-P. Seng, A Multi-Channel MAC Protocol with Power Control for Multi-Hop Mobile Ad Hoc Networks, International Conference on Distributed Computing Systems Workshop, 2001. [18] M. de Graaf, F. Bouwer, System Architecture, AAF project deliverable, 26 April 2005. [19] IEEE 802.15 WPAN Task Group 2. http://www.ieee802.org/15/pub/TG2.html. [20] IEEE 802.19 Coexistence Technical Advisory Group. http://grouper.ieee.org/groups/802/19. [21] J. Mitola III, "Software Radio Architecture: Object-Oriented Approaches to Wireless Systems Engineering", Wiley 200. [22] M. Dillinger, K. Madani, N. Alonistioti, "Software Defined Radio: Architectures, Systems and Functions", Wiley 2003. [23] L. B. Michael, M. J. Mihaljevic, S. Haruyama, A Framework for Secure Download for Software-Defined Radio, IEEE Communications Magazine, July 2000. [24] M. Dillinger, R. Becher, Decentralized Software Distribution for SDR Terminals, IEEE Wireless Communications, April 2002. [25] SDR forum, Requirements for Radio Software Download for RF Reconfiguration, 13 November 2002. http://www.sdrforum.org/public/approved/02_a_0007_v0_00_dl_req_01_22_03.pdf.

22

AAF/D4.11

[26] H. Wang; T. Cai; J. Song, "Analysis of common structure and software download for software defined radio", ICCT 2000, August 2000. [27] J. So, N. Vaidya, Multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using a Single Transciever, IEEE/ACM MobiHoc, 2004. [28] Z. Tang, J. J. Garcia-Luna-Aceves, Hop-Reservation Multiple Access (HRMA) for Ad-Hoc Networks, IEEE INFOCOM, 1999. [29] A. Tzamaloukas, J. J. Garcia-Luna-Aceves, A Receiver-Initiated Collision-Avoidance Protocol for Multi-Channel Networks, IEEE INFOCOM, 2001. [30] A. Naspuri, J. Zhuang, S. R. Das, A Multichannel CSMA MAC Protocol for Multihop Wireless Networks, IEEE WCNC, 1999. [31] A. Naspuri, S. R. Das, Multichannel CSMA with Signal Power-based Channel Selection for Multihop Wireless Networks, IEEE VTC Fall, 2000. [32] S.-L. Wu, C.-Y. Lin, Y.-C. Tseng and J.-P. Sheu, A New Multi-Channel MAC Protocol with OnDemand Channel Assignment for Multi-Hop Mobile Ad Hoc Networks, International Symposium on Parallel Architectures. Algorithms and Networks, 2000. [33] N. Jain, S. R. Das, A. Nasipuri, A Multichannel CSMA MAC Protocol with Receiver-Based Channel Selection for Multihop Wireless Networks, in Proc. of the 9 International Conference on Computer Communications and Networks, 2001. [34] J. S. Pathmasuntrharam, A. Das, A. K. Gupta, Primary Channel Assignment based MAC (PCAM) A Multi-Channel MAC Protocol for Multi-Hop Wireless Networks, IEEE WCNC, 2004. [35] P. Porwal, M. Papadopouli, On-Demand channel switching for multi-channel wireless MAC protocols, International Symposium on Parallel Architectures, Algorithms, and Protocols, August 2000. [36] YuanYuan Wang Medium Access Control Protocol for Cognitive Radio Network, MSc thesis, TU Delft 2005. [37] IETF RFC 3626: http://www.ietf.org/rfc/rfc3626.txt . [38] D. Niculescu, B. Nath, Ad Hoc Positioning System (APS), IEEE GLOBECOM, 2001.
th

AAF/D4.11

23

[39] A. Savvides, C.-C. Han, M. B. Strivastava, "Dynamic Fine-Grained Localization in Ad-Hoc Networks of Sensors", ACM MOBICOM 2001. [40] S. Capkun, M. Hamdi, J.P. Hubaux, "GPS free positioning in mobile ad hoc networks", Proceedings of the 34th Annual Hawaii International Conference on System Sciences, 2001. [41] D. Cabric, S. Mubaraq Mishra, R. W. Brodersen, Implementation issues in Spectrum Sensing for Cognitive Radio, IEEE MILCOM, 2001. [42] H. Nikookar, Advanced Topics in Digital Wireless Communications, Lecture Notes, TU Delft, 2005. [43] Ultra-Wideband/A disruptive Technology?, Intel white paper, 2005. http://www.intel.com/technology/ultrawideband/downloads/ultra-wideband_technology.pdf. [44] M. de Graaf, F. Brouwer, Scenarios and requirements, AAF project deliverable, 31 March 2005.

24

AAF/D4.11

You might also like