| The combination of cognitive radio(CR)and multicarrier modulation(MCM)technology can effectively improve the efficiency of spectrum utilization.As an effective implementation of the MCM scheme,the Orthogonal Frequency Division Multiplexing(OFDM)and Orthogonal Frequency Division Multiple Access(OFDMA)technologies have been widely used.The complexity problem of OFDMA-based cognitive radio network(CRN)resource allocation faces many challenges.This paper focuses on the joint optimization of sensing duration and resource allocation of OFDMA-based CRN and proposes some low complexity algorithms as follows: multicore-based modified Hoare's FIND(McMHF)based message passing(MP)algorithm(McMHF-MP);two algorithms for sensing duration and resource allocation joint optimization based on piecewise convex decomposion and concavity-proserved interpolation strategies,respectively;cross-layer joint optimization algorithm for sensing duration,resource allocation and routing selection based on variable decoupling,approximate convexification and band handover strategy.Moreover,the traditional polyblock algorithm for sensing duration and resource allocation bilevel optimization is theoretically analyzed,and an approximate upper bound of iteration times is accordingly obtained.The specific research work and major innovations in this dissertation can be summarized in four aspects as follow:(1)To meet the requirement of further speedup of message passing(MP)algorithm for OFDMA resource allocation,a multicore-based modified Hoare's FIND(McMHF)algorithm is proposed by exploiting the cooperation of idle users.Firstly,a Multicore Modified Hoare's FIND(McMHF)algorithm is proposed addressing the selection algorithm embedded within the MP algorithm for OFDMA resource allocation optimization,enabling adjustment of the trade-off between operation speed and signaling overhead.Secondly,a set of iterative formulae of probability generation functions(PGF)of McMHF operation delay and signaling overhead are derived,which can be used for theoretical analysis of McMHF and other related algorithms.Finally,a McMHF-based MP(McMHF-MP)algorithm is proposed for OFDMA resource allocation.Theoretical analysis and simulation results indicate that the proposed algorithm significantly outperforms the traditional MP-based OFDMA resource allocation optimization algorithms.(2)Based on the bilevel optimization model of OFDMA-based CRN sensing duration and resource allocation,the Lipschitz continuity and non-concaveness of the upper-level objective function is proved.An approximate upper bound of the iteration times of the corresponding polyblock algorithm is obtained.Firstly,Lipschitz continuity and convexity analyses on the upper level objective function are performed to compensate for the insufficiencies of existing literatures addressing the traditional bilevel optimization model.Secondly,the approximate upper bound of the number of iterations of the corresponding polyblock algorithm is derived,which provides a basis for estimation of the computation complexity of the bilevel optimization scheme based on the polyblock algorithm.Finally,the complexity of the polyblock algorithm under the bilevel optimization scheme is analyzed successively using the metrics of iteration times in the upper layer,the times of the upper layer's calling lower layer optimization subprocedure and the evaluation times of the lower level objective function.(3)Aiming at the problem that polyblock algorithm is too complex to fulfil joint optimization of sensing duration and resource allocation of OFDMA-based CRN,two suboptimal algorithms are proposed based on the piecewise convex decomposition and concavity-preserved interpolation strategies,respectively,which have significantly lower complexity and small loss of optimization accuracy.Firstly,a Piecewise Approximate Convex Decomposion(PACD)algorithm is proposed,which decomposes the original feasible region into multiple subregions in which the sensing duration and resource allocation optimizations become two decoupled convex subproblems enabling parallel procession.Secondly,based on concavity preservation strategy,another Concavity-Proserved Interpolation Formula(CPIF)algorithm is proposed for joint optimization of sensing duration and resource allocation,which tries to find a concave function as an interpolation function to replace the upper-level objective function,converting the original non-convex problem approximately into a convex problem.The theoretical analysis and simulation verification indicate that the above two algorithms show great advantages over the polyblock algorithm in terms of complexity with low loss of optimization precision.In addition,each has its own advantage over its peer,the former has higher accuracy and the latter lower complexity.(4)Aiming at high complexity of the sensing duration,resource allocation and routing selection cross-layer joint optimization problem for minimizing the packet delay of OFDMA-based CRN,an algorithm based on Variable Decoupling,Approximate Convexification and Band Handover(VDAC-BH)is proposed based on a sequential optimization strategy,which significantly reduces the computation complexity.Firstly,a strategy based on exponential spline interpolation to achieve the Optimal Sensing Duration(OSD)is proposed.Then,the Probability Density Function(PDF)of the OSD and CRN packet frame aligment delay under the Nakagami-m fading environment are derived.Finally,based on M/G/1/K queuing theory,the CRN packet delay minimization model under Nakagami-m fading environment is formulated.It minimizes the average delay of multiple streams of data packets as the objective,involving the opmizable variables of system sensing duration,each user's bandwidth and transmission power,buffer capacity and Device-to-Device(D2D)routing decision,meanwhile constrained by the queuing system stability.A suboptimal VDAC-BH fast algorithm is proposed to address such a model.Theoretical analysis and simulation results indicate that the proposed algorithm has good optimization performance and low complexity. |
PDF Full Text Request