Reserach On Resource Allocation And Performance Of Cooperative Relay For Cognitive Radios

With the explosive growth of wireless systems and services, the spectrum resourcehas become more and more scarce. However, fixed spectrum assignment policy causes alarge portion of the assigned spectrum to be used sporadically. This motivates thedevelopment of cognitive radio (CR), which is exploited as a new communicationparadigm to improve spectrum utilization. On the other hand, cooperative relayingtechnology has been shown to be effective to improve the reliability, coverage andthroughput of wireless systems. By integrating with cooperative relay, CR can achievehigher spectrum efficiency through exploiting more spectrum opportunities.In this dissertation, the resource allocations and system performance analysis inCR systems are respectively investigated. Specifically, with the theory of convexoptimization, the resource allocations improve CR system performance by jointlydesigning different system parameters. Meanwhile, by using statistics forsignal-to-noise rate (SNR) of received signal, the system performance analysis ofcognitive relay networks is derived under different power constraints. The maincontents of this dissertation are summarized as follows.In Chapter2, two optimization problems of cooperation spectrum sensing (CSS)are investigated. Firstly, the reporting delay is considered and the optimization problemof CSS with sensing user (SU) selection is formulated to maximize the averagethroughput of the cognitive radio network (CRN). The optimal algorithms arerespectively proposed in both additive white Gaussian noise environment and Rayleighfading environment. It is shown that selecting all the SUs within the CRN to cooperatemight not achieve the maximal average throughput. Secondly, two optimizationproblems of multi-channel CSS are respectively formulated to maximize the averagethroughput of the CRN under data fusion rule and decision fusion rule. The branch andbound algorithm and the greedy algorithm I are proposed to obtain the optimal solutions.Through the proposed algorithms, the following three key questions can be solved:⑴how to optimally assign SUs to cooperatively sense multiple channels;⑵how tooptimally set the sensing time in the CRN;⑶how to optimally set the sensing threshold for each channel.In Chapter3, the joint optimization of spectrum sensing and power allocation isinvestigated. Firstly, the optimization problem of jointly designing sensing time, sensingthreshold and power allocation is investigated under total transmit power constraint ofthe CRN and the average interference power constraint of primary network. An iterativealgorithm is proposed to obtain the locally optimal values for these parameters. Then,we consider a multi-channel CRN where each SU can only choose to sense a subset ofchannels. We formulate a joint optimization problem of sensing channel selection,sensing time and power allocation under the constraints of average transmit powerbudget and average interference power budget, which maximizes the CRN’s averagethroughput. The greedy algorithm II, which has much less computational complexity, isproposed to solve the joint optimization problem. Moreover, it is shown that the searchspace of the greedy algorithm II can be further pruned.In Chapter4, system performance of the cognitive relay network is analyzed inunderlay model. Specifically, an amplify-and-forward (AF) cognitive relay networkwith partial relay selection (PRS) is considered. Based on this, the exact close-formoutage probabilities of different PRS schemes are derived under interference powerconstraint and transmit power constraint. Meanwhile, based on the tight upper bound ofend-to-end SNR, the lower bound on symbol error rate and the upper bound on ergodiccapacity are derived over Rayleigh fading channels. Monte Carlo simulations are shownto validate the accuracy of the analytical results.In Chapter5, the resource allocations of underlay model and overlay model arerespectively formulated to improve cognitive relay networks’ performance. For onething, in underlay model, bandwidth and power allocations with AF ordecode-and-forward (DF) relaying protocol are presented to⑴maximize the sumnetwork throughput;⑵minimize the total transmit power of the cognitive relaynetwork with considering the fairness of power drain of relay SUs;⑶maximize theenergy efficiency of the cognitive relay network. It is shown that the correspondingresource allocation problems are equivalently reformulated as convex optimizationproblems and, therefore, can be solved efficiently. In particular, when considering thedecoding rate constraint in DF relaying protocol, we propose the hybrid relayingprotocol that combines AF and DF relaying protocols. The joint bandwidth and power allocation problem with hybrid relaying protocol is formulated to maximize the sumnetwork throughput. The greedy algorithm III is developed to solve the jointoptimization problem, which has much less computational complexity. It is shown thatthe greedy algorithm III has comparable performance to the exhaustive search algorithm.On the other hand, considering the “all participate” AF (AP-AF) and selective AF (S-AF)cooperative relay scheme respectively in overlay model, we formulated the problems ofdesigning the optimal sensing time and power allocation that maximize the cognitiverelay network’s average throughput and minimize the outage probability of secondarytransmission. The optimal algorithms are proposed to acquire the optimal sensing timeand power allocation. It has been shown that significant improvements in the averagethroughput and outage behavior are achieved when both the parameters for sensing andpower allocation are jointly optimized.