Study On Optimization Problems For Cooperative Spectrum Sensing In Cognitive Relay Networks

Spectrum sensing is the premise and key of realizing cognitive radio (CR)technologies, and cooperative spectrum sensing (CSS) with relays can improvesignificantly the accuracy of spectrum sensing results, which is becoming popular in recentyears. This paper regards amplify-and-forward relay protocol based CSS as the mainresearch object, and investigates the tradeoffs between sensing performance and sensingenergy consumption, and between primary network benefit and secondary network benefit,by combining green communication ideas and optimizing the node transmission powers.This paper studies the optimization of single radio relay CSS, the joint optimization ofsingle radio relay CSS and cooperative data transmission, and the optimization of dualradio relay CSS. The single radio relay refers to the relay equipped with only one radiointerface, and the dual radio relay refers to the relay equipped with two independent radiointerfaces. The main contributions are listed as follows:In the optimization of single radio relay CSS, we study two cases: the single relaycase and the multiple relays case. We optimize over the relay amplifying gain and thenumber of samples so that the spectrum sensing consumption is minimized while satisfyingthe sensing performance constraint. In the single relay case, by our analysis, the aboveconstrained optimization problem with two variables can be transformed into anunconstrained optimization problem with only one variable, of which the numericalsolution can be easily attained by one-dimensional searching method. In the multiple relayscase, we can obtain the approximate closed-form solution to the above optimizationproblem when all the relay amplifying gains are assumed to be identical, where thecomputational complexity does not increase with the number of relays. The derived resultsare validated by simulations.In the joint optimization of single radio relay CSS and cooperative data transmission,we propose a new scheme in which a relay assists secondary users to simultaneouslyperform spectrum sensing and data transmission. We optimize over the total powerallocation of the secondary user network so that the secondary network benefit can bemaximized while guaranteeing a certain primary network benefit. Specifically, we choosethe optimal secondary source node transmission power and relay node amplifying gain, sothat the secondary network throughput can be maximized while guaranteeing that thedetection probability is no less than a threshold. By our analysis, the above constrainedoptimization problem with two variables can be reduced to an unconstrained optimization problem with only one variable. Simulation results show that, our proposed optimalnetwork power allocation scheme always outperforms the equal power allocation schemein terms of the secondary network throughput, and the performance gap between these twoschemes increases with total power budget.In the optimization of dual radio relay CSS, we propose a new scheme where onerelay simultaneously helps two secondary users perform spectrum sensing by its twoindependent radio interfaces. We optimize over the relay power allocation on its two radiointerfaces, so that the secondary network benefit is maximized while guaranteeing a certainprimary network benefit. Specifically, we seek two optimal amplifying gains on the relay’stwo radio interfaces, so that the sum of two false alarm probabilities measured by twosecondary users respectively is minimized (i.e., the opportunities that secondary usersaccess the primary frequency bands are maximized) while satisfying the individualdetection probability constraints. By our derivation and analysis, the above constrainedoptimization problem with two variables can be reduced to an unconstrained optimizationproblem with only one variable. Simulation results show that, when the relay power budgetis large, compared with equal power allocation scheme, the proposed optimal relay powerallocation scheme can decrease the sum false alarm probability by about10%.