Data Dissemination in Cognitive Radio Networks

Do ctor of Philosoph y – Mid Thesis Activit y Rep ort Pierre & Marie Curie Univ ersit y Sp ecialization C OMPUTER S CIENCE presen ted b y Mr Mubashir Husain Rehmani Submitted as Mid-Thesis Activity Rep ort for the partial requireme nt of Do ctor of Philosoph y from Pierr e & Marie Curie Univ ersity Data Dissemination in Cognitiv e Radio Net w orks 24 th Marc h 2010 Commitee: Dr. Mart in May Examiner Scien t ist and Thomson F ellow, Thomson Lab oratory Paris - F rance Dr. Nguyen T hi Mai T rang Examiner Asstt. Professor, Pierre & Mari e Curie University – P ar is - F rance Dr. Aline Carneir o Viana Co-Advisor Research Scientist, ASAP , INRIA – Saclay - F rance Dr. Hicham K halife Collab orator Asstt. Professor, LaBR I/EN SEIRB – B ordeaux - F rance Prof. Serge Fdida Main Advisor Professor, Pierre & Mar ie Curie Unive r sit y – P ar is - F rance Abstract In this rep ort, w e first describe the pr oblem that w e are dea ling with i.e. data dis s emination in multi-hop cognitive ra dio netw orks. T o address this problem, we pro pose a channel selection strategy named ‘SURF’. W e ev aluate the prop osed c hannel selection strategy in b o th single-hop and multi-hop s cenarios and compa r ed it with relev ant approaches. So far, one technical rep ort and a p oster is published as par t of this w o rk, while tw o publications are under review; one is in IE EE Communications Letters and the s econd one is in IE EE W oWMo M conference. In on-going works sections, we first mention some p ossible directions in the context of SURF. In addition to that, we men tion differen t resea rc h pro blems that w e are planning to deal during the course of this P hD disser tation. Publications P ap ers Under Review • Mubashir Husain Rehmani, Aline Carneir o Viana, Hich a m Khalife, and Serge Fdida, Chan- nel Assortment Str ate gy for R eliable Commu nic ation in Multi-hop Co gnitive R adio Net- works , Submitted to: IEEE Communications Letters . • Mubashir Husain Rehmani, Aline Car neiro Via na , Hicham Khalife, a nd Serge Fdida, T o- war d R eliable Contention-awar e Data D issemi n atio n in Multi-hop Co gnitive Ra dio A d Ho c Networks , Submitted to: IE EE W oWMoM 2010 . T ec hnical Rep ort • Mubashir Husain Rehmani, Aline Car neiro Via na , Hicham Khalife, a nd Serge Fdida, T o- war d R eliable Contention-awar e Data D issemi n atio n in Multi-hop Co gnitive Ra dio A d Ho c Networks , INRIA RR-037 5, 2009, F rance. h ttp:// hal.inria.fr/inria- 004418 92/en/. P oster • Mubashir H usa in Rehmani, Aline Carneiro Via na, Hic ha m Khalife, and Serge Fdida , A daptive and Oc cup ancy-b ase d Channel S ele ction for un r eliable Co gnitive R adio Networks , Rencontres F r ancophones sur les Asp ects Algorithmiques des T elec o mm unications (AL- GOTEL) 2009 , du 1 6 au 19 juin 2 009, Car ry Le Rouet, F r ance. T alks • Mubashir H usa in Rehmani, Aline Carneiro Via na, Hic ha m Khalife, and Serge Fdida , T owar d R eliable Contention-awar e Data Dissemination in Multi-hop Co gnitive R adio A d Ho c Networks , Pr esen ted in the workshop ”Group e de T rav ail” for the students of Ma s- ters in Computer Science , Sp e c ialit y Net works (RES), University P ierre and Mar ie Cur ie (UPMC), Jussieu, Paris, F rance, F ebruary 2010 . Data Dissemi nation in Cognitiv e Radio Net w orks 1.1 In tro duction Radio sp ectrum is a precio us natural reso urce and it has b een utilized for co mmunication since many decades. In this p ersp ectiv e, several tec hniques hav e b een pro posed to utilize the radio sp ectrum e fficie n tly . These techniques rang es from geo graphical reutilization of radio sp e ctrum and multiplexing techniques, just to name a few. Nev er theless, F edera l Co mm unication Com- m unica tions (FCC) rep orts that the ra dio sp ectrum is still under-utilized and the o ccupancy of radio sp e c trum v arie s from 15% − 85% [1]. According to F CC, this is due to fixed spectrum assignment p olicy and g e ographical and temp oral utilization of the r adio sp ectrum. Thus, keep- ing this notion in mind, the idea of Cognitive Radio Ad-Ho c Netw orks (CRN) is first coined by J. Mitola et al. [2] to utilize the r adio sp ectrum efficiently . In fa ct, Cognitive Radio Netw or k s are comp o sed of intel ligent wireless no des a.k.a . Cognitive Radio (CR) no de s . These CR node s scan the spectr um and utilizes the sp ectrum as so on as they found the sp ectrum is free by the legacy users a.k.a. Prima r y Radio (PR) no des. Cognitive radio ad- hoc netw or ks hav e b een widely used in Military and E mergency Net works due to their mission critical nature [3], [4]. But with the adv ancement in tec hnolog y and the av aila bilit y of cheap consumer dev ices, the applications of cognitive r adio netw orks fo r consumer - based applications is no more longer a dream. In this p erspec tive, [5], [6] and [7] highlights some of the consumer- based applications of cog nitive radio netw orks. These types of consumer-bas ed cognitive radio netw ork a pplications can b e en vis aged in nea r future [5] and can be deplo yed in Railwa y stations, mar k ets, airp orts, public sp ots, and re s tauran ts. One pr o spective application could be data dissemination in these types of netw orks which enables a CR user to conv ey emergency alar ms and a le rts or to deliver lo w pr iorit y data such as advertisement messag es in a cognitive radio multi-hop co n text. Data diss emination is a classica l and a fundamental function in an y kind of netw or k . In wireless netw orks , the character is tics and problems intrinsic to the wir eless links bring several challenges in data dissemination in the shap e of mes sage loss es, collisio ns, and broadcas t s to rm problem, just to name a few. How ever, in the co n text of Cognitive Radio Ad-Ho c Net work (CRN) [1], reliable data dissemination is muc h more challenging than traditional wireless net- works. First, in addition to the already known iss ues of wirele s s e nvironments, the diversity in the nu mber of channels each cognitive no de can use adds another challenge by limiting no de’s accessibility to its neigh b ors. Seco nd, cognitive ra dio nodes hav e to compete with the Primar y 1 Radio (PR) nodes for the res idual r esources o n ma ny channels and use them opp ortunistically . Besides, during communication CR no des should co mmunicate in such a w ay that it should not degrade the reception quality of P R no des by ca using CR-to-P R in terferenc e . In a ddition, CR nodes should immediately in terr upt its tra ns mission whenever a neighbor ing PR activity is detected [15]. 0 10 20 30 40 50 60 70 80 90 100 5 10 15 20 25 30 35 40 Avg. Success Delivery Ratio (%) No. of CR Competing Nodes Channel # 1, 10% PR occupancy Channel # 2, 30% PR occupancy Figure 1.1: CR no des co mpeting for the same channel. Recently , a lo t of work has been car ried out for dynamic channel ma nagemen t in cognitive radio net works [9, 11, 16, 17]. How ever, all these appr oac hes focuses on single- ho p cognitive radio net works and either r equires the presence of a n y central en tity or co o rdination with primary radio no des in their channel selectio n decisio n. F or instance, [9] prop osed an efficien t sp ectrum allo cation architecture that adapts to dynamic traffic demands but they co nsidered a sing le -hop scenario of Access Poin ts (APs) in Wi-Fi netw ork s . [18] propos ed a c hannel s e le ction strategy based on the primar y use r’s o ccupancy but sp ecifically designed for single-ho p architecture. In multi-hop co gnitiv e ra dio ad ho c netw o rks, wher e co ordination b et ween CRs is hard to achiev e and no central entit y for r e g ulating the acces s ov er channels is to b e envisaged, re lia ble data dissemination is even mor e co mplex . In this persp ectiv e, the fir st step in having efficient data dis s emination is to know how to sele ct b est channels . Thus, differently from most w or k s in the literature dealing with single-hop comm unication [9], [10], we go a step further here and build up a c hannel selec tion strateg y for multi-hop comm unication in CRN. The ob jectiv e of every co gnitiv e radio no de is to select the b est c hannel ensuring a maxim um connectivity and consequently , allo wing the la rgest data dissemination in net work. This c o rresp onds to the use of channels having not only low primary radio no des (PRs) activities, nevertheless the reliability of the dissemina tion pr ocess is achiev ed by limiting the conten tion of cognitive ra tio nodes (CRs) acceding selected channels. 2 The effect o f CR conten tions on dissemina tion is hig hligh ted in Fig. 1.1 that shows the evolution of the average success delivery ratio at receivers o f a sing le so ur ce, with the n umber of comp e ting CRs. It is clea r that the p erformance of a channel with low PR activity decreases with the num b er of CR comp eting fo r the av aila ble resource. Nevertheless, a channel with higher PR a c tivit y ca n be a go o d choice if CR conten tion is low. The challenge here is then how to find a go o d tra deoff b et ween connectivity and conten tion. In this rep ort, we present our work i.e. a channel selection s tr ategy , named SURF. The goal of SURF is to ensure reliable conten tion-aware data diss e mination and is sp ecifically des igned for m ulti-ho p cognitiv e radio ad ho c net works. Usually channel selection strateg ies provide a way to no des to select channels for transmission. On the contrary , SURF endue CR no des to select bes t channels not o nly for trans mission but also for o verhearing. As a result, b oth s ender a nd receiver tuned to the right channel for effective and r eliable data dissemination. Additionally , b y dynamically explor ing residual reso urces on channels and by monitoring the num b er o f CRs on a particular c hannel, SURF a llows building a co nnected netw ork with limited conten tion where reliable communication c a n take pla c e. The r emainder o f this rep ort is o r ganized as follo ws : we discuss co nnectivit y vs . conten tio n trade-off in Section 1.2. W e give gener al ov ervie w of our c hannel selection strategy SURF in Section 1.3. Section 1.4 deals with per formance ev aluation of SURF. Finally , sec tion 1.5 briefly describ e the o n-going works and section 1.6 concludes the re port. 1.2 Cognitiv e radio ad ho c net w orks: connectivit y VS. con ten tion trade-off In a highly dynamic/o pp ortunistic cognitive r adio network, co gnitiv e users comp ete for residual resource s (a .k.a spec tr um holes) left by the activity of the legacy users more formally ca lle d primary radio users . Every cog nitiv e no de, using an intelligen t selection strategy , selects the appropria te channel for trans mittin g with the ma jor constra in t o f not degr a ding the serv ice of ongoing primary radio communications. Indeed, pr imary r adios ha ve the absolute prio rit y ov er the communication channels. In a n opp ortunistic multi-hop cognitive radio netw o r k where co ordination b etw een CRs is hard to achiev e and no central entit y for r egulating the acc e s s over channels is to b e envisaged, the o b jectiv e of every co gnitiv e ra dio is to select the channel ensuring a ma xim um connectivity . Suc h sp ectrum band has the highes t num b er of active cognitive radios hence allows quick and e ffective data disseminatio n in the net work. Int uitively , one may think that the b est stra tegy for al l CRs is to dynamically switch to the less occupied channel (by PRs). Thus, sa tisfying the ob jectiv e of verifying prior it y cons tr ain ts impo sed b y PRs. Nev ertheles s , such s trategy leads to many classical problems alr eady well known in wireless netw o rking. Firs t, forcing all CRs in a geographic ar ea to be active over the 3 same channel makes all nodes co mpete for the same resource th us gene r ating co n tention and collision problems. Sec o nd, suc h approa c h wastes the v aluable additional capacity on different channels that the cog nitive radio concept offer s. Indeed, it was already shown in traditional wireless netw orking that netw orks with high conten tion, where rep etitiv e co llisions ar e frequent, suffer from close to zero throughput [13]. A t ypica l exa mple is desc r ibed in Fig. 1.2. Initially , channel 1 has more primary radio activities and should be avoided b y the CR tra nsmitters. How ever, if enough CRs switch to channel 2 to comm unicate, channel 1 quickly b ecomes less o ccupied and able to ca rry higher throughputs than c hannel 2. Therefor e, taking in to account conten tion issues due to CR transmissions is necessar y when selecting sp ectrum bands for CR communications. PR Use Channel Free Channel CR Use Channel Channel # 1 Channel # 2 Figure 1.2: PR and CR No des o ccupancy over channels An y prop osed strategy for channel selection in CRN has to o ptimize the co nnectivit y vs. conten tion trade-o ff. W e pr o pose hereafter a channel selectio n strategy that monitor s the num b er of active CR no des on a par ticular channel. As a result, we ar e able to build a wel l c onne cte d network while dynamic al ly exploiting r esidual r esour c es on many cha n nels . 1.3 SURF: Channel Selection Strategy SURF channel selection strategy is sp ecifically designed fo r ad ho c cognitive radio netw orks . The general goal of our str ategy is to e nsure a reliable da ta dissemination ov er a mu lti-ho p CRN. Such technique can b e used to con vey emerg ency ala r ms and alerts or to deliv er low prior it y data such as adv er tisemen t messag es in a cog nitiv e radio m ulti-hop con text. Recall that in order to achiev e o ur go al and ensure cov er age and r eliabilit y , the connectivity vs. conten tio n trade-off should be optimized. SURF str ategy is exclusiv ely implemen ted by every CR node and is used for transmiss io n and/or overhearing. Using the decentralized algo r ithm pro posed by SURF, every CR se nder judiciously s e le cts the b est frequency band for sending messages and every CR r eceiv er tunes to the right channel (selected by the sender ) to retrieve the sent data. With SURF, each CR no de lo oks first for the less PR-o ccupied channel to help deciding au- tonomously which channel to use. In a dditio n to P R o ccupancy , we also c onsider CR neighbors comp eting for the same c ha nnel resource . More precise ly , every CR no de classifies av ailable channels based on the observed PR-oc cupancy over these c ha nnels . This clas sification is then refined b y iden tifying the num b er of active CRs ov er each band. The best channel fo r trans - 4 mission is the c hannel that has the lowest PR activity a nd a re a sonable ongoing CR activity . Indeed, choosing a channel with few CRs yields to a disconnected net work. The ch a llenge in our strategy is in finding the n umber of active CRs on every c hannel that giv es the b est connectiv- it y with limited con tention. Practically , every CR after clas sifying av a ilable ch a nnels, switches dynamically to the b est one and bro a dcasts the stor ed message. Additionally , CRs with no message s to transmit implement the SURF stra tegy in order to tune to the b est channel for data receptio n. Clearly , using the sa me strategy implemented by the sender allo ws nodes in the close geog r aphic a reas to se lect the sa me channel as sender for ov er hearing with high probability . Intuitiv ely , it is likely that CRs in the s ender’s vicinit y hav e the same PR occupa ncy , hence channels a v ailable to a CR sender is also a v ailable to its neighbors with high probability [11]. Therefore, SURF controls the num ber of CR receivers, th us a connected topolo gy with lo w con tention is crea ted. Once a pac ket is rec eiv ed, every CR receiver undergo es again the same pro cedure to c ho ose the appropriate c hannel for conveying the message for its neig h b or. Through simulations 1 , we show that SURF builds, as exp ected, a highly connected netw ork suitable for relia ble diss emination. Moreover, SURF outper forms existing algorithms. In orde r to ev aluate the p erformance of SURF, w e co mpare it with an intuitiv e random strategy (RD) and the tw o v a r ian ts of sele ctiv e broadca sting proto col [12], i.e. selective broadcasting strateg y (SB) [12] without any centralized authority , and se le ctiv e broa dcasting with c en tralized autho r it y (CA) to b e served as an upper b ound. The simplicit y and decen traliz e d natur e of o ur solution makes it usable in a d ho c CRNs deployed to convey emergenc y messages a nd alerts. It can also be emplo yed in co mmercial a pplications to disse mina te short publicity messages. 1.4 P erformance Ev aluation T o as sess the p erformance of SURF with RD, SB, and CA in term of r eliable data dissemination, t wo p erformance metrics a re ev aluated with differ en t to ta l num b er o f channels: (i) the average delivery ratio , which is the ratio of pack et received b y a par ticular CR no de ov er total pa ckets sent in the netw o rk and (ii) the average num be r of accumulative CR rece ivers at ea c h transmissio n, un til TTL=0. Recall that higher num ber of channels yields to lower P R o ccupancy . In addition, it is w or th mentioning her e that even the cen tra lized appro ac h CA co uld not get a 100 % of data dissemination b ecause of the per formed randomly as signmen t of Acs set to CR no des. This may generate top ology disco nnections cause d b y physical close nodes b eing ass igned to disjoint channels. In this w ay , as previously stated, we c onsider the CA approach g ets the theoretica l upper bound results in terms of message dissemina tio n. Fig. 1 .3(a) c ompares the n umber of accumulative CR re ceiv ers at eac h ho p o f communication un til TTL=0, for the four strategies. When C h = 1 5, SURF allows the message dis s emination 1 F or more detailed simulations resul t s please refer our T echnical Rep ort [22 ] 5 0 10 20 30 40 50 60 70 1 2 3 4 5 6 Avg. nb. of Accumulative Receivers Hop Count RD: Ch = 5 RD: Ch = 15 SB: Radio=1, Ch = 5 SB: Radio=1, Ch = 15 CA: Ch = 5 CA: Ch = 15 SURF: Ch = 5, β =10 SURF: Ch = 15, β =18 (a) A cc umulativ e receivers 0 0.2 0.4 0.6 0.8 1.0 10 20 30 40 50 60 Average Delivery Ratio CR Nodes’ ID RD: Ch = 5 RD: Ch = 15 SB: Radio=1, Ch = 5 SB: Radio=1, Ch = 15 CA: Ch = 5 CA: Ch = 15 SURF: Ch = 5, β =10 SURF: Ch = 15, β =18 (b) Delivery ratio Figure 1.3 : Average n umber o f a ccum ulative r eceiv ers p er hop and a verage de livery ratio in a 70-no de CRN, for random (RD), selective broadca sting (SB), centralized approa c h (CA), and our strategy (SURF). to 55% of no des in the net work (i.e. 3 8 out o f 70 CR nodes ), while CA allows 78% (i.e. 5 4 ov er 70 CR no des). Additionally , due to its central control a nd multiple transmiss ions, the CA strategy r eac hes this upp er bo und of r eceiv ers per cen tage a t the TTL=4. It can b e cle arly seen that SURF outper forms RD and SB and compared to CA, only provides a decre a se of 25% in per formance. The gain ach ieved with CA is a t the pr ice o f mor e trans mis s ions, more energy consumption, and mor e ex p ensive and sophistica ted devices. Fig. 1.3(b) compares delivery ratio of RD, SB, CA and SURF, as a function o f the CR nodes’ ID. SURF outper forms RD and SB in terms of delivery ratio, when num b er of c hannels are high. Compared to the CA strategy , SURF has only 20% o f p erformance reduction. In par ticula r, for 6 Ch=5 a nd Ch=15 , SURF g uarant ees the de livery o f approximately 60% of mes s ages (with a single transmissio n), co ntrarily to less than 20% for the RD and SB stra tegies (with sing le a nd m ultiple transmissions, re s pectively) and 80% for the CA s trategy (with multiple transmissions ). 1.5 On-Going W orks In this section, we describ e the goals that we wan t to achiev e during the course of PhD Thesis. The on-go ing w or ks of our thesis enco mpasses the following themes: 1.5.1 Extension of SURF Our prop osed channel selection strategy ‘SURF’ ca n be ex tended in several direc tions. One po ssibilit y is to in vestigate and o ptimize data dissemination delay and compar ed it with other relev ant appro a c hes. This delay in da ta dissemina tion is due to mess a ge lo sses when sender and receiver ar e not on the sa me channel. In a ddition to tha t we can add a utilit y - based heur istic to the channel weigh t co mputation and in vestigates SURF performance under dynamic traffic by considera tion of da ta r ates and tra ffic volume generated by CR no des in the c hannel w eight formula. By co nsidering the data rates, a channel that suppo rts higher data ra tes can be selected. While, consider ing the CR tr a ffic volume avoid co ng estion in the net work. Moreover, prediction and history can also b e accounted in SURF to enhance the p erformance. In fact, if a pattern exists in the tra ffic utilization of CR nodes , then by the help of prediction techniques, one c a n estimate the future o ccupancy of CR no des over channels and av oid highly CR occ upied channels. 1.5.2 Study of the NS mo del for Cognitiv e R adio N et w orks W e are studying the Net work Simulator (NS) model fo r Cognitive Radio Netw or k s. In fact, w e found a NS-2 patch [1 4] av ailable for Cognitive Radio Netw or ks that supp ort many functionalities of Co gnitiv e Ra dio Netw ork. T hus, our goa l is to first study in deta il the existing patch of NS Sim ula tor and use this existing patc h for our future sim ulations . 1.5.3 Data Dissemination t hrough Channel Bonding in Multi-Hop Cog- nitiv e Radio N et w orks Based up on our channel sele c tion strategy SURF, we incorp orate a nother impor tan t aspect that can b e exploited in the co n text of Co gnitiv e Ra dio Net work for efficient and r eliable co mmuni- cation. In fact, we exploit the av ailability o f contiguous non-overlapping c hannels to create a bo nded channel and use it with our channel selection strategy . Through c ha nnel b onding [19 –21], mult iple frequency c ha nnels are bonded into a single broadband channel. Therefore, the agg regated ba ndwidth is la r ge due to the sum of multiple 7 frequency channels and as a consequence, the rate of pack et trans mis s ion increa se. This will als o reduce the pack et trans mis sion time. Another adv antage of channel bo nding is the lo w delay . Thu s , we prop ose a broadcasting strategy based on Channel Bonding, named ‘BOND’; em- powers CR no des with the abilit y to infer, ba sed o n informa tio n reg arding PR o ccupancy , the less PR-o ccupied channel to use. Once cla s sified the less PR- o ccupied channels, CR no des detect and group the contiguous non-ov erlapping channels to crea te a bo nded channel. Cognitive radio no des then use this bonded channel for broadcasting in order to improv e the perfor mance o f the net work. W e claim that if w e do not exploit the usage o f nonoverlapping contiguous channels, the sp ectrum will b e us ed inefficiently . In this way , how to find a go od tra nsmission oppor tunit y in terms of PR occupa ncy and selection of b onded channel, constitutes our main goal. BOND will provide to CR no des a strategy to selec t channels to cr eate a bonded channel accor ding to their av aila bilities, g iving to no des the p ossibility o f selecting the b est classified one for transmission and/or ov erhea r ing. In o ther words, the use of channel b onding in conjunction with our channel selection strateg y , allow CR users to efficien tly disseminate and sha r e information. Our stra tegy is adaptive in nature and w ell suited for cognitive radio netw o rks, where the av ailable channel set changes dynamically . In future, we intend to c har acterize the num b er of channel to be b onded and it’s impact ov er netw ork p erformance metrics. Through simulations, w e in tend to de mo nstrate that our solutio n is s calable as the num be r of channel incr eases. Note that this is on- going w or k and we are thinking over several dimensions. 1.5.4 In ternet Access F ramew ork for F uture Cognitiv e Radio Net- w orks W e hav e g ained so me insights through the work that we hav e done s o far on channel selection strategies, c halleng es of broadca sting, and da ta dis s emination in cognitive ra dio netw ork s . Thus, by combining featur es of our prop osed channel selection strategy SURF, channel b o nding, and data dissemination, we intend to use them to cr eate a ba sis to build up on a framework for future cog nitiv e radio net works. This framework is b oth for c hallenged environments and can be extended to consumer-bas ed cognitive ra dio net works applications. In this co n text, an Internet Access F ramework for future Cognitive Radio Netw orks (IAF CRN) in under in vestigation. Through this framework, cognitive radio no des a c hieves the goal of r o- bust co nnectivit y to connect to in ter net and disseminate data in c halleng ed en vir onmen ts. More- ov er, IAF CRN assures efficien t utilization of the sp ectrum usag e and increase the capacity of the net work by explo iting channel b onding and channel agg r egation tec hniques . The prop osed framework is completely decen tra lized in nature b y in tegr ating and exploiting the new pa radigm of Cog nitiv e Radio Sens or Net works. Thus, by deploying low-cost sens ors in the vicinit y of co g- nitive radio a ccess po in ts, a sp ectrum opp ortunit y map is cre ated to facilitate channel selectio n 8 decision. This feature makes the IAFCRN framework co ordination-indep enden t with the exist- ing infrastructure i.e. Primary Netw ork. Moreov er, deploying sensors instead of cognitive radio base station requir es less complexit y and cos t. W e then investigates the applicability of exis t- ing solutions a nd their shortcomings in the c o n text of IAF CRN framework. W e envisage that by integrating channel bonding , channel agg regation, and cognitive r a dio senso r netw or ks, new consumer-bas ed co gnitiv e r adio communication para dig ms c an b e rea lized. V alidation through simulations ar e left for future works. 1.5.5 Challenges of Broadcasting in Cognitive R adi o Netw orks A t the b eginning of the Thes is , we have done a comprehensive literature review on broadcas ting strategies. Now, based o n the e xperience that we hav e acquir ed, we are wr iting a survey pap er on br oadcasting strategies in Cognitive Radio Net works. In this s urv ey paper, we give an ov erview of v ar ious br oadcasting str ategies tha t hav e b e en prop osed so far for Cognitive Radio Net works. Moreover, we also in tend to g iv e a comprehensive surv e y of broa dcasting schemes for Cognitive Radio Netw or ks. W e identif y req uired key character istics of broadcasting strategies in Cognitive Radio Netw or k s. In addition to that we cla s sify these sc hemes and w e b eliev e tha t this classification would be useful for aca demic and industry based r esearchers who are engaged in the desig n of broadcasting str ategies for Cognitive Radio Net work. W e also give some insig h ts to improv e these s chemes, which may b e helpful to res e a rc her s who wan t to further improve them. 1.6 Conclusion F or efficient sp ectrum utilization, recently the Mutli-hop Co g nitiv e Radio Ad- ho c Netw orks has gained a lot of attention and emerg ed as a promising technology . In fact, there is a exp o nen tial bo ost in cognitive r a dio netw ork res e arc h due to the proliferatio n of consumer-based applicatio ns. Nevertheless, s eeing cognitive radio netw orks functiona l on-gro und yet requir ed consider able efforts. Thus, this thesis is o ne step further and we hop e that it will op en new horizon of resea rc h in realizing practical cognitive radio net work applications. Evidently , dealing the cutting edge cognitive radio netw ork r esearch pr oblems and working with eminent re s earc her s, i have gained techn ica l exp erience in terms o f r esearch. Besides this, working with SURF, i got deeper insight int o how to tackle a research pr oblem and iden tify the pro s a nd cons of the propo sed solution. F or instanc e , (1) b oth PR and CR traffic should b e consider ed b ecause a go od channel in terms of PR do es not mean the ch a nnel is appropriated, a nd (2) fo r a g oo d disseminatio n transmitter and receiver should b e tuned to the same channels. Moreov er , generally cognitive r adio netw ork should be handled in suc h a way that i meet the r equiremen ts of consumer-base d applications, while considering the tech no logical constraints. I hope that during the co urse of m y PhD Thesis, i will b e able to make a rema rk a ble contribution in the s cien tific communit y 9 Bib liogr aph y [1] I. F. Akyild iz, W.-Y. Lee, M. C. V uran, and S. Mohant y , “Next gene ration/dynamic sp ectrum acc ess/cognitive radio wireless netw orks: a su r v ey ,” Computer Ne tworks: The International J ou rnal of Computer and T ele com- munic ations Networking , vol. 50 , Issue 13, p p. 2127 – 2159, 2006. [2] J. Mitola and G. Q. Maguire, “Cognitive rad io: Makin g softw are radios m ore pe rsonal,” IEEE Personal Comm u- nic ations , v ol. 6, no. 4, pp. 13–18, 1999. [3] R. Leschhorn and B. Buchin, “Mil itary softw are defi ned radios- rohde and sc hw arz status and p erspe ctiv e, ” in Pr o ce edi ngs of the SDR 04 T e chnic al Confer ence and Pr oduct Expositi on , 2004. [4] O. Y ounis, L. Kan t, K. Chang, and K. Y oung, “Cognitive m anet design for mission-critic al netw orks,” IEEE Communicati ons Magazine , p p. 64–71, Octob er 2009. [5] S . Ball, A. F erguson, and T. W. Ronde au, “Consumer applic ations of c ognitiv e radio d efined netw orks,” i n First IEEE International Sympo sium on New F r ontiers in Dynamic Sp e c trum A c cess Networks (DySP AN’ 05) , 8-11 Nov 2005, pp. 518–525. [6] M . M. Buddhikot, “Und erstanding dynamic spec trum ac cess: Mod els, taxon om y , and challenges,” in DySP AN , April 2007, p p. 649–663. [7] J. M. Chapin and W. H. Lehr, “The path t o market success for dynam ic sp ectrum access te c hnology ,” IEEE Communicati ons Magazine , vol. 45(5), pp. 96–103, 2007. [8] H. Khalife , S . Ahuja, N. Malouch, an d M. Kru nz, “Prob ab ilistic p ath selection in opportu nistic c ognitiv e rad io netw orks,” in pro c ee dings of the IEEE globe c om confer enc e , 30 Nov - 4 Dec 2008, pp. 1–5. [9] L. Y ang, L. Cao, H. Zhe n g, and E. Belding, “T raffic-a ware d ynamic sp ectrum acc ess,” in Pr oc e edi ngs of The F ourth International Wir eless Internet Conferenc e (WICON 2008) , Ha waii, USA, 17-19, Nov embe r 2008. [10] C. Cordeiro, K. Challap ali, D. Birru, and S. N. Shanka , “Ieee 802.22: An introd uction to the fi rst wirele ss standard based on cognitive radios,” Jou rna l of Communic ations , vol. 1, no. 1, pp. 38–47, Ap r i l 2006. [11] Q. Zhao, L. T ong, A. Sw ami, and Y. Che n, “Dece n traliz e d cogni tiv e m ac for opportu nistic sp ectrum access i n ad ho c net works: A p omdp framew ork,” IEEE Journal of Sele cted Ar e as in Comm unic ations. - Sp e cial Issue A daptive, Sp ectrum A gile C ognitive Wir ele ss Networks , vol. 25, n o. 3, pp. 589–600, Ap ril 2007. [12] Y. R. Kondare ddy and P . Agraw al, “Selec tiv e b roadcasting in multi-hop cognitive radio ne t works,” in IEEE Sarnoff Symp osium , Pr i nceton, Ne w Jersy , 28-30 April 2008, pp. 1–5. [13] J. Li, C. Blake, D. S . J. D. Couto, H. I. Le e, and R. Morris, “Capacity of ad hoc wirel e ss netw orks,” in ACM MobiCom , Rome, Italy , Ju l y 2001, p p. 61–69. [14] Jin g Zhong, and Jialiang Li, “Cognitive Radio Cognitive Netw ork Simulator,” URL: http://stu web.ee.mtu.edu/ ∼ l jialian/ [15] H. Khalife , S. Ahuja, N. Malouch, and M. Krunz . P robabilistic path selection in opportu nistic c ognitiv e rad io netw orks. In pr oc e edings of the IEEE g lob e com c onfer e nce , pages 1–5, 30 Nov - 4 Dec 2008. [16] D. Niy ato an d E. Hossain. Compe titiv e sp ectrum shari ng in cogniti ve radio netw orks: A dynamic game approach. IEEE T r ansactions on wir eless communic ations , 7, No. 7:2651–2660, Ju ly 2008. [17] H. Rah u l , N. Kushman, D. Katabi, C. So dini, and F. Edalat. Learnin g to share: Narrowband-friendly wideband wireless netw orks. In ACM SIGCOMM , volume 38 , Issue 4, pages 147–158, SEA TTLE, W A, USA, 17-22 A UGUST 2008. [18] Mu bashir Husain Re hmani, Aline Carneiro Viana, Hicham Khalife, and Se rge Fdida. Adaptive and o ccupan cy- based c hanne l selection for unrel iable cognitive radio netw orks. In R e nc ontr es F r ancophones sur les Asp e c ts A lg orit hm iques des T ele communic ations (ALGOTEL) , Carry Le Rouet, F rance, 16-19 Jun e 2009. [19] IEEE P202.22 (D0.1) Dr aft Standar d for Wir eless Re gional Ar e a Networks Part 22: Co gnitive Wir eless RAN Me dium A c cess C ontr ol (MAC) and Physic al Layer (PHY) spe cific ations: Policies and pr oc e dures for op er ation in the TV Bands . [20] IEEE P202.22 Dr aft Standar d for Wir ele ss Re gional Ar ea Networks : Channel Bonding versus Channel A gg r e- gation . [21] IEEE P202.22 Dr aft Standar d for Wir ele ss Re gional A re a Networks : Channel A ggr egation Summary . [22] M. H. Rehmani, A. C. Viana, H. Khalife, and S . Fdida, “T ow ard reliable contention-a wa re data dissemination in multi-hop cognitive radio ad ho c netw orks,” INRIA, F ranc e , T ech. Rep. RR-0375, De cember 2009. [ Online]. Av ailable: http://hal.inria.fr/inr i a- 00441892/en / 10