Research On Resource Optimization Algorithms For Dynamic Spectrum Sharing Systems

Compared to traditional wireless systems,thedynamic spectrum sharing (DSS)systems have the following two characteristics: available resources of secondary users(SUs) depend on the spectrum usage of primary users (PUs) and SUs can access to thespectrum holes in the premise of not causing harmful interference to PUs. So,in ordertoguarantee the quality of service (QoS) of all users and maximize the system spectrumefficiency, reasonable resource optimization algorithms are needed.To research the resource optimization problemsof DSS systems, two scenariosshould be considered: one is with prior knowledge of the primary system (PS), thatis,considering the access behavior of PUs; the other is without any prior knowledge ofthe PS.Considering the access behavior of PUs, the following three aspectsare included:(1) Based on the analytical performance model with considering the state of everychannel of the DSS system, strategies for processing collisions between PUsand SUs are studied, and the average interference ratio is proposed to measureinterference to the PS. Theperformance increase of the DSS systemandimpacts of the scheduling period areverified by OPNET simulation.(2) By considering the parameter perturbation, impacts of imperfect parameterestimation are studied from the perspective of the parameter estimation errorand imperfect spectrum sensing. Two parameters: relative change and bias aregiven to measure impacts of parameter perturbation to the prediction algorithmand the whole system, respectively. Finally, the availability and effectivenessof the parameter: relative change as an indirect measure of performance erroris verified by MATLAB simulation. And the parameter perturbationbounds forthe stability ofsystem performance are given by OPNET simulation.(3) Based on the research of existing scheduling schemes, a fair schedulingstrategy based on collision statistics is designed for the SUs in the samepriority. And the fairness index is given to quantify the fair degree ofscheduling strategies. OPNET simulation results show that the proposed scheme can significantly improve the fairness across all SUs by evenlydistributing collisions between PUs and SUs over a large part of SUs or overthe whole procedure of SUs’ transmission.Without considering the access behavior of PUs, performance optimizations basedon channel reservation are shown below:By reserving channels for PUs, three optimizationproblems are solved, that is,maximize the achievable arrival rate of SUs, minimize theblocking probability of SUs,and minimize the number of channels required by the system. Moreover, anapproximate approach is presented to estimate the steady-state probabilities, and theclosed-form expressions forthe blocking probability of SUs and the droppingprobability of SUs are derived. Simulation results show that using theapproximateapproach in above algorithms can greatly reduce the computationalcomplexity.