2010 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media,

Transcription

1 200 IEEE. Personal use of this material is ermitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including rerinting/reublishing this material for advertising or romotional uroses, creating new collective works, for resale or redistribution to servers or lists, or reuse of any coyrighted comonent of this work in other works.

2 200 24th IEEE International Conference on Advanced Information Networking and Alications owards rust Establishment for Sectrum Selection in Cognitive Radio Networks Sazia Parvin, Song Han, Li Gao, Farookh Hussain and Elizabeth Chang Digital Ecosystems & Business Intelligence Institute Curtin University of echnology, Australia Perth, Australia {sazia.arvin, {Song. Han, Farookh.Hussain, Abstract Cognitive Radio (CR) has been considered as a romising concet for imroving the utilization of limited radio sectrum resources for future wireless communications and mobile comuting. As cognitive radio network (CRN) is a general wireless heterogeneous network, it is very essential for detecting the misbehaving or false nodes in the network. So in this aer we roose a trust aware model which rovides a reliable aroach to establish trust for CRN. his aroach combines all kinds of trust values together, including the direct trust and indirect trust value for the secondary users. Deending on this trust value, it is decided that whether the secondary user can user the rimary user s sectrum band or not. he mathematical results show that our trust model can efficiently take decision for assigning sectrums to the users. Keywords-cognitive radio networks; sectrum; trust; rimary user; secondary user I. INRODUCION Among different kinds of wireless technology suorting Internet access and other services, a very effective idea is to merge different wireless networks and to use one of them aroriately deending on the communication environments and the alication requirements. Cognitive radio ioneered by J.Mitola iii [] from software defined radio (SDR) was originally considered to imrove sectrum utilization. here is an ever increasing demand of sectrum for emerging wireless alications and there is a shortage of sectrum for the wireless alications. Considering these things Federal Communications Commission (FCC) has considered to make the licensed sectrum available to the unlicensed users. So the unlicensed users can use the fallow sectrum rovided they cause no interference to the licensed users. Most of the radio systems today are now aware of the radio sectrum. Cognitive radio is a aradigm for wireless communication in which either a network or a wireless node changes its transmission or recetion arameters to communicate efficiently avoiding interference with licensed or unlicensed users. A cognitive radio senses the available sectrum, occuies it and can vacate the sectrum on sensing the return of the rimary user. We can call future wireless networks as cognitive radio networks (CRN), which is retty much consistent of Haykins s definition of cognitive radio [2]: Cognitive radio is an intelligent wireless communication system that is aware of its surrounding environment (i.e., outside world), and uses the methodology of understandingsby-building to learn from the environment and adat its internal states to statistical variations in the incoming RF stimuli by making corresonding changes in certain oerating arameters (e.g., transmit ower, carriesfrequency, and modulation strategy) in real time, with two rimary objectives in mind: highly reliable communication whenever and wherever needed, efficient utilization of the radio sectrum. As referenced by [3, 4] once cognitive radios can find the oortunities using the sectrum holes for communications, cognitive radio networking to transort ackets on to of cognitive radio links is a must to successfully facilitate useful alications and services. A mobile terminal with cognitive radio caabilities can sense the communication environments (e.g. sectrum holes, geograhic location, available wire/wireless communication system or networks, available services), analyse and learn information from the environments with user s references and demands, and reconfigure itself by adjusting system arameters on forming to certain olicies and regulations. For examle, when a cognitive radio mobile terminal sensed that there WIFI and GSM systems nearby while sectrum holes exist in the frequency band of digital V, it may decide to download files from a certain WiFi A, make a hone call through GSM system and communicate with other cognitive radio users using those sectrum holes. Cognitive radio technology could also facilitate interoerability among different communication systems in which frequency bands and/or formats differ [4]. Cognitive radio, on the other hand, sits above the SDR (Software Defined Radio) and is the intelligence that lets an SDR determine which mode of oeration and arameters to use. Actually an SDR is simly a radio that uts most of the Radio Frequency (RF) and Intermediate frequency (IF) functionality, including waveform synthesis, into the digital (rather than the analog) domain, allowing great flexibility in the modes of radio oeration (called ersonalities ) []. A cognitive radio network is thus not just another network to interconnect cognitive radios. he CNRs are comosed of various kinds of communication systems and networks, and can be views as a sort of heterogeneous networks. he X/0 $ IEEE DOI 0.09/AINA

3 heterogeneity exists in wireless access technologies, networks, user terminals, alications and service roviders [5]. So free sectrum sensing is a key characteristic used in CRNs. hrough this sensing rocess, unlicensed user can determine whether the radio can be used or not. But if the unlicensed user is not a trust worthy node, then it can break down the normal activities of the CRNs by injecting some malicious attacks. hat s why establishing trust for CRN is an oen and challenging issue. In this aer, we roose a trust aware model which rovides a reliable aroach to establish trust for the sectrum selection in CRNs. he organization of this aer is as follows: In section 2, related works is reviewed. In section 3, system architecture of our roosed model is described. In section 4 and 5, we show how trust is calculated and sectrum allocation decision is based on the result. We conclude the aer in section 6 including future remarks. II. RELAED WORKS Now-a-days trust in the human society has become the most imortant thing for human being s communications, work and lives. However trust can be regarded as criteria for making a judgment under comlex social conditions and can be used to guide further actions [6]. rust and security are very closely interrelated and indeendent that is difficult to searate each other. But, nowadays, establishing trust for CRN is an oen and challenging issue. rust has been widely mentioned in literatures regarding trusted comuting and web comuting, ad hoc networks and even social science [7-0].However, trust for CRN is comletely different from all of these scenarios. rust is critical in CRN oeration and beyond security design, as security usually needs communication overhead in advance. he authors [] describes the trust in CRN as follows: A cognitive radio senses a sectrum hole and to dynamically access the sectrum for transmission requires trust from originally existing system (i.e. rimary system) and regulator, even without creating interference to PS. A cognitive radio may want to leverage another existing cognitive radio to route its ackets, even though another CR is not the targeted recient terminal. It requires trust from another CR. A cognitive radio can even leverage PS to forward its ackets to realize the goal of acket switching networks. It needs trust from the PS, not only at network level but also in service rovider. Because of all these reasons, the idea of alying trust and reutation model in a CRN has recently attracted research interest. he imact of trust model on CRN is discussed briefly in [2]. In this aer, the authors suggested otential ways for incororating trust modeling to CRNs including identity management, the trust building rocess and ossible mechanisms for disseminating the trust information. But no exerimental results were established for these discussions. he authors in [3] integrated trust and reutation for the threat mitigation of Sectrum Sensing Data Falsification (SSDF) attack on CRNs. But trust modeling was not the meeting oint of their aer. In this aer, we integrated direct trust and indirect trust for using the rimary user s sectrum band if the sectrum is available to use. III. SYSEM ARCHIECURE Cognitive radio (CR) is a novel aroach for imroving the utilization for making it ossible for a grou of secondary (unlicensed) users (SUs) to access the sectrum band which is not being by the rimary (licensed) users (PUs) in some geograhical location. A. A CRN Architecture As like Wireless Networks, CRNs can be deloyed in various kinds of network configurations such as Centralized, Ad-hoc and Mesh Architecture. In our aer we imlement an infrastructure-based CRN with centralized network entities. Figure shows a general architecture of CRNs as deicted in [3]. Figure.A general cognitive radio network architecture [4] Figure 2. System Architecture of Proosed Model 580

4 As referred by [5], we assume that PUs coexist with SUs in some geograhical area and PUs are controlled by a fixed PU base station (PUBS). In this CRN, SUs are also distributed in the coverage area of a fixed SU base station (SUBS). SUs can only communicate with each other through the SUBS within the transmission range of the SUBS. he SUs and PUs are not able to communicate between themselves. But the communication between SUBS and PUBS is ossible. Figure 2 shows the architecture of our system model. If one SU wants to use the PU s sectrum band, then at first the SU needs to sense the sectrum. During the sensing rocess, each SU senses the PU sectrum bands individually and reorts the results to the SUBS. By integrating the sensing results reorted by the SUs with its own sensing result, the SUBS determines the activity states of each PU sectrum band and allocates resources to SUs within its range. In this aer, we strongly assume that PUs, PUBS and SUBS are trustworthy entities in the CRN. B. Architecture of rust Model In a CRN, secondary user needs the service rovided by rimary users. A secondary user can sense the sectrum band of rimary user, if it is free then the secondary user will send sensing information to the SUBS. Every user can be denoted by User <User ID, A, V> where ID denotes the identity of the user, A denotes the attribute set of user ID, V denotes the value set of the attributes. As soon as the cooeration rocess starts that means the secondary user senses the rimary user s sectrum band and sends the sensing result to the SUBS, the trust relationsh is set u among the SUs and PUBS. In our trust model, a user is denoted by User <ID, A, V, >, where denotes the set of the trust values associated with the attributes as every attribute has its own trust value. C. rust Model Our trust model is a trust a modification of the trust model [6]. Han s model comutes trust for wireless sensor network. In our new model, we build trust model for cognitive radio networks considering that the networks consisting of PUs, SUs, SUBS, and PUBS. he architecture of our trust model is shown in the following figure. As shown in the figure, when the secondary user (Ex. SU) tries to use one rimary user s sectrum band, at first SU needs to sense the PU s sectrum band to check whether it is free or is in using. After sensing the sectrum band, the SU will send the sensing result to the SUBS. hen the SUBS will get the direct trust value for that SU and indirect trust value from other SUs. When the SUBS gets the direct and indirect trust value, the integrated trust module combines both the trust value and the indirect trust value to calculate the integrated trust value. Figure 3. System Model of Proosed Scheme IV. CALCULAION OF RUS As referred by [2], in our trust model, trust is reresented by a real number ranging from 0 (comlete distrust) to (comlete trust). he trust exists in the user, thus the trust value among the users can also be divided into three categories: direct trust value, indirect trust value and integrated trust value. Direct trust value is the tye of trust value can be established between SU and SUBS. And this is denoted by Directtrust. Indirect trust is established when SU 2.SU n rovide its trust value of SU to SUBS. Integrated trust value can be calculated by the SUBS based on the direct trust value of SU and the indirect trust value of SU from other SUs, denoted by Integtrust. A. Direct rust Calculation he direct trust value of secondary user can be determined by its multi-attribute trust value. he conditions of a User are always changing, thus the SUBS always need to evaluate trust value of the SU based on its multi-attribute trust value. he information about the ast cooeration is assembled in a table of cooeration record among the Users, as shown in the table. As referred by [2], each attribute has three relevant values : the number of the successes (S i, i =,2,.n), the number of failures (F i, i =,2 n) and the number of the cooeration (C i, i =,2, n). In order to simly our analysis, we assume that the cooeration/non-cooeration behavior is of equal value during the communication rocess between the Users. 58

5 ABLE I. COOPERAION RECORD ABLE = int egtrustγ SUBS + ( int egtrust ) M R M τ Γ τ Where W realiable + W = and W, W [0,] strange realiable strange C = S + F, i =,2,.n. he trust value for In table, i i i attribute A i can be comuted based on the table as follows: i Ci hus, the overall trust value for the SU with n attributes A, i =, 2,... n (denoted by node ) can be comuted using i as follows: User = n n n Since all, i =,2,., n can be calculated using data in table, the trust value of Secondary User who wants to sense and use Primary User s sectrum band, can be easily calculated by the SUBS. B. Indirect rust Calculation When the SUBS sends the cooeration request to other secondary users, three kinds of trust values from reliable nodes, the strange nodes (the node never cooerate with the SUBS before) and the unreliable nodes, are returned to SUBS. he trust value of reliable nodes ( reliable ) and strange nodes ( Strange ) can be ket, and the trust value of the unreliable nodes must be discarded. he SUBS assign some weight value to reliable nodes, strange nodes. hus the indirect trust can be calculated as follows: = S + ( ) C. Integrated rust here are different kinds of cooeration among secondary users and SUBS. So the SUBS can automatically assigns the different weights value based on the requirements of a certain task in our trust model. he weight of the direct trust is denoted by W directtrust, the weight of the indirect trust is denoted by W indirecttrust. he integrated trust value can be calculated by the following equation: W = + W, int egtrust directtrust directtrust indirecttrust indirecttrust Where Wdirecttrust + Windirecttrust = and W, W [0,] directtrust indirecttrust V. DECISION CALCULAION FOR ASSIGNING SPECRUM When the SUBS calculates the integrated trust value, then it will send this trust value to PUBS. hen the PUBS will take decision whether SU is able to use Primary user s sectrum or not deending on the following equation: = Γ + ( ) int egtrust SUBS int egtrust M R M τ Γ Where R is the overall sensing result from PUBS for PU sectrum band ; int egtrust is the trust value for the SU from the SUBS; Γ SUBS is the sensing result rovided by the SUBS; τ is the trustworthiness of SU i in the context of PU sectrum band ; τ 582

6 Γ is the sensing result for PU sectrum band rovided by SU i ; M is the number of SUs whose trustworthiness with resect to PU sectrum band is above a redefined threshold η. In the case when the variance in trustworthiness of each SU is the context of a rimary sectrum band is not considered, τ are set to. hen the is set to 0 and all second term in this equation reduces to a simle average of all the sensing results obtained from the SUs. It is effectively a weighted sum of the SUBS sensing result and a majority voting from all the SUs who choose to articate in the distributed sensing oeration. he final decision D is made based on the sign of R, D, R < 0 = 0, R = 0, R > 0 Whenever D =, that time the SU will be able to use the rimary user s frequency sectrum band. VI. CONCLUSION AND FUURE WORKS In this aer, we roose a trust aware hybrid sectrum sensing scheme for CRN. In the CRN, some misbehaviouring SUs may want to access the PU s available sectrum band. Such malicious SUs can seriously imact on the whole network erformance. So in this aer, we roose the combination of all kinds of trust values together, including the direct trust and indirect trust value for the secondary users. Deending on this trust value, it is decided that whether the secondary user can user the rimary user s sectrum band or not. In the future work, we want to erform the simulation deending on the mathematical terms. Conference, Network and Parallel Comuting Workshos, NPC Workshos [7]. Ghosh, N. Pissinou, and K. Makki, owards designing a trusted routing solution in mobile ad hoc networks. Mobile Networks and Alications, (6): [8] P. Naldurg and R.H. Cambell. Dynamic Access Control: Preserving Safety and rust in Network Defense Oeration,. in Proceedings of the Eighth ACM Symosium in Access Control Models and echnologies (ACM SACMA 2003) [9] G. heodorakooulos and J.S. Baras, On rust Models and rust Evaluation Metrics for Ad-Hoc Networks, IEEE Journal on Selected Areas in Communications, (2): [0] A.A. Pirzada, C. McDonald, Establishing trust in ure ad-hoc networks,. in Proceedings of the 27th Australasian conference on Comuter science [] K.Chen,Y.Peng,N.Prasad,Y.Kiang,S.Sun, Cognitive radio network architecture: art II -- trusted network layer structure, in Conference On Ubiquitous Information Management And Communication 2008, ACM: Suwon, Korea [2].C.G. Clancy,, N. Goergen, Security in Cognitive Radio Networks: hreats and Mitigation, in Cognitive Radio Oriented Wireless Networks and Communications, [3] R. Chen, et al., oward secure distributed sectrum sensing in cognitive radio networks, in IEEE Communications Magazine Secial Issue on Cognitive Radio Communications [4] I.F. kyildiz, NeXt generation/dynamic sectrum access/cognitive radio wireless networks : A surve,. Comuter Networks, : [5]. Qin,, et al., owards a trust aware cognitive radio architecture., ACM SIGMOBILE Mobile Comuting and Communications Review (2): [6] G. Han,, et al. A Reliable Aroach of Establishing rust for Wireless Sensor Networks, in International Conference on Network and Parallel Comuting Workshos 2007: IEEE Comuter Society. REFERENCES [] J.Mitola, Cognitive Radio: An Integrated Agent Architecture for Software Defined Radio, PhD thesis,, in Royal Institute of echnology (KH) [2] S.Haykin, Cognitive radio: brain-emowered wireless communications,ieee Journal on Selected Areas in Communications, (2): [3] K.Chen,Y.Peng,N.Prasad,Y.Kiang,S.Sun, Cognitive radio network architecture: art I -- general structure, in Proceedings of the 2nd international conference on Ubiquitous information management and communication Suwon, Korea ACM. [4] FCC, E Docket No Notice of Proosed rule making and order. December, [5] X.Gao, G.Wu, and.miki, End-to-end quality of service rovisioning over heterogeneous networks. IEEE Wireless Communication, (3): [6] G.Han, D. Choi, and W. Lim. A Reliable Aroach of Establishing rust for Wireless Sensor Networks. in IFIP International 583