Research On Distributed Compressive Spectrum In Cooperative Cognitive Networks

Cognitive radio technique is considered as one of the solutions of currentspectrum resource scarcity.And spectrum sensing is one of the key technology ofcognitive radio.In the spectrum sensing process,single-node detection will producethe performance degradation because of the channel fading,shadow effects, noise andother factors.In order to improve the performance of spectrum sensing,cooperativespectrum sensing is proposed，which can recover the limits of spectrum sensing byonly one node.This thesis mainly investigates the cooperative spectrum sensing technique. Themain tasks are as follows：Introduce the fact that, when comparing with single user sensing, collaborativespectrum sensing can obtain a lot of benefits.And then we make a total comparison forthe existing algorithms of spectrum sensing in cooperative, compare the performancesof algorithm of hard merge, soft merge and algorithm based on multi-bit integration.Introduce the backgrounds of the innovation of the compressive sensingtechnology and its theory frameworks. And then detail the theory of the compressivesensing technology from the aspects of sparse representation、design of measurementmatrix and reconstruction algorithm.Developed a distributed cooperative spectrum sensing approach that permeatesthe benefit of compressive sensing and consensus optimization.The compressivesensing technique is exploited to reduce the sampling rate.The spectrum of primaryusers is recovered jointly by cooperative cognitive radio users, which utilize iterativeconsensus optimization to reach globally sensing outcomes via one-hop localcommunication only. In particular, a weighted consensus-averaging constraint isintroduced to reduce the number of the consensus constraints, which lowers thecomputation loads and expedites the convergence. The convergence of our proposeddistributed cooperative spectrum sensing scheme is proved analytically. Simulationsdemonstrate that the proposed scheme can effectively sense the spectrum fromcompressive sample.