Cross-layer Analysis And Optimization For Opportunistic Spectrum Access In Cognitive Radio Networks

The contradiction between the dramatically increasing demand for spectrum resources of wireless communications and the under-utilization of spectrum resources of primary users(PUs)has been prevailing recently.Cognitive radio networks(CRNs)enable sec-ondary users(SUs)to opportunistically use the licensed spectrum(LS)when detecting the absence of PUs.Owing to the capability of avoiding harmful interference to PUs while improving the spectrum efficiency(SE),CRNs have been envisioned as one of promiss-ing technologies to relieve the contradiction and achieve dynamic spectrum access.On one hand,since both imperfect spectrum sensing on physical(PHY)layer and access contention on medium access control(MAC)layer impact the interference to PUs and SUs' throughput performance,it is imperative to analyze SUs' throughput and interfer-ence probability to PU considering imperfect spectrum sensing and contention access.On the other hand,duo to the shared licensed spectrum between PU and SUs,SUs' throughput and the interference probability to PU are two conflicting metrics,thus it is imperative to optimize the performance of CRNs considering the tradeoff between these two conflict-ing metrics.The above cross-layer analysis and optimization are imperative for theoretical research and practical deployment of CRNs.Aiming at analyzing and optimizing the per-formance of CRNs from a cross-layer perspective,this thesis conducts numerous research on typical scenarios of CRNs,including complex wireless environment,such as practical PUs' traffic distribution and heterogeneous fading channels,and typical access protocols,such as cooperative access and multi-channel CRNs.The main work and contributions are summarized as follows.Considering the tradeoff between the interference to PU and SUs' throughput in CRNs,the joint impact of imperfect spectrum sensing on PHY and access contention on MAC is analyzed and then a cross-layer analytical model and optimization scheme are proposed.This research first theoretically analyzes SUs' conditional transmission prob-ability,which provides the basis for the cross-layer analysis and optimization.First,we classify the collisions exposed by SUs' packets into two categories,i.e.,the collisions with PUs due to missed detections and the collisions with other SUs due to access contention,and compute the overall collision probability.Then,we model the backoff procedure as a two-dimensional Markov chain and obtain SUs' transmission probability when de-tecting the channel as idle.Second,we consider two widely-used MAC protocols,i.e.,slotted Aloha and DCF,and compute the closed-form expressions of interference proba-bility to PU and SUs' throughput.The tradeoff problem is thus formulated to maximize SUs' throughput satisfying interference probability to PUs constraint rather than detecting probability constraint.Finally,we prove that SUs' throughput is maximized when the in-terference probability constraint is transformed into an equality constraint.Binary search algorithm on the argument sensing duration is designed to solve the optimization prob-lem.Simulation results validate the outperformance of the proposed cross-layer scheme compared against conventional scheme ignoring the access contention on MAC.They also indicate that the optimal sensing duration and detection threshold vary with PU's signal-to-noise ratio(SNR)and SUs' frame duration and thus should be carefully designed in practical CRNs.Considering the impact of complex distribution of PUs' traffic on the joint optimiza-tion of spectrum sensing parameters and frame duration,the channel status transitions are modeled as alternating renewal process according to stochastic process theorem.Then,a joint optimization algorithm on both sensing parameters and SUs;'frame duration is proposed.The proposal adapts the optimal solution with the variation of statistics of the channel and thus is suitable to the practical PUs' traffic distribution.First,considering random departure and arrival of PUs' traffic within SUs' sensing period and data trans-mission period,energy detection is modeled as a quadruple-hypothesis testing problem.We use typical model for PUs' traffic distribution according to which the average idle du-ration and busy duration follow independent Poisson distribution.Thus,the channel status transitions are modeled as alternating renewal process and the status transition probabil-ity are further computed.Second,we analyze the impact of PUs' traffic distribution on SUs' spectrum sensing and data transmission.From a cross-layer perspective,we obtain the closed-form expression of interference probability to PU and SUs' throughput and formulate the tradeoff problem taking PUs' traffic distribution into account.Finally,a binary-dimensional search algorithm on the arguments frame duration and sensing dura-tion is designed to solve the problem,including the optimal sensing duration,detection threshold and frame duration.Simulation results validate the feasibility of the binary-dimensional search algorithm in solving the optimization problem.They also prove that SUs' throughput degrades evidently with the increase of channel status transition frequen-cy.Considering heterogeneous fading channels from PU to multiple SUs in power lim-ited CRNs,the impact of free space propagation and power limited SUs is analyzed and a cluster-based particle swarm optimization(C-PSO)algorithm is designed to obtain the optimal spectrum sensing parameters and transmission power.The proposal can improve both the spectrum efficiency and energy efficiency by considering.First,since energy de-tection is mainly impacted by the amplitude of PUs' signal,the typical large-scale fading model free space propagation is used.Moreover,each SU's average power consumption for spectrum sensing and data transmission is restricted not exceeding a predefined thresh-old.And,SUs adapt the transmission rate according to PU' s SNR and the remaining transmission power.Second,by modeling the impact of heterogeneous fading channels and power consumption constraint,we obtain the closed-form expressions of the interfer-ence probability to PU,SUs' throughput and SUs' average power consumption.Then,we formulate the tradeoff problem to maximizing SUs' throughput under two constraints,i.e.,the interference probability to PU and SUs' average power consumption.Finally,to solve this mixed integer non-linear programming problem,C-PSO algorithm is proposed.By iteratively updating the position,fitness,velocity and known best particle,the globally optimal solution is achieved.Simulation results validate the feasibility and efficiency of C-PSO algorithm.Compared against related contributions which assume homogeneous fading channels,our proposal achieves better throughput performance while strictly sat-isfying interference probability and SUs' average power consumption constraints.Considering the fact that SUs randomly choose one from all sub-channels in multi-channel CRNs,the performance of multi-channel CRNs is analyzed and optimized in a cross-layer approach.This research studies the discrete and independent spectrum holes in terms of multiple channels in practical CRNs and provides theoretical basis for perfor-mance analysis and optimization in multi-channel CRNs.First,owing to SUs' hardware constraint,each SU randomly chooses to opportunistically access one from multiple sub-channels.The probability distribution of the number of SUs contending for the same sub-channel is computed.And then,SUs' transmission probability when detecting the absence of PU is calculated.Second,we compute SUs' throughput and the interference probability to PU under slotted Aloha and DCF for given number of SUs.Applying the probability distribution of the number of SUs contending for the same sub-channel,the overall performance of the global multi-channel CRNs is obtained.Finally,an exhaus-tive search algorithm on the argument sensing duration is proposed to solve the tradeoff problem.Simulation results show that SUs' throughput performance degrades with the increasing number of sub-channels for fixed total bandwidth and thus it is recommended to choose the spectrum occupied by less PUs for better performance.Considering the improved spectrum sensing reliability as well as the overhead re-sulting from cooperative sensing,a joint optimization of cooperation scheme and sensing parameters is proposed.This research analyzes the joint impact of cooperative sensing and access contention and is suitable to cooperative sensing based CRNs.First,it is assumed that SUs' spectrum sensing results are identically and independently distributed.Multiple SUs and a fusion center(FC)cooperatively sense the channel using mini-slot protocol.FC uses k-out-of-n scheme to determine the channel status and then SUs contend to access according to FC's determination.Second,the expressions of interference probability to PU and SUs' throughput are obtained taking into account the improved sensing reliability and overhead resulting from cooperative sensing.We optimize SUs' throughput under the constraint of interference probability to PU.Finally,the optimization problem is decom-posed into two sub-optimization problems.One is the optimization of spectrum sensing parameters for fixed fusion rule and the other one is the optimization of fusion rule.By exhausively searching the fusion rule and sensing duration,the optimal combination of fusion rule and spectrum sensing parameters is obtained.Simulation results indicate that the optimal solution achieves better throughput and varies with PU's SNR and the number of contending SUs.