Research On Key Techniques Of Wideband Spectrum Sensing For Cognitive Radio

Nowadays, due to the dramatical growth of the wireless communication networks andservices, the shortage of spectrum resource becomes more and more obvious, which hasresulted in the obstacle to the improvement of wireless communication technologies.Current wireless networks are regulated by a static spectrum allocation policy in whichparticular frequency band is restricted to the licensed primary users (PUs). Cognitiveradio (CR) has emerged as a promising technology for maximizing the utilization of thelimited bandwidth while accommodating the increasing amount of services andapplications in wireless networks.A fundamental requirement for CR is spectrum sensing and which is an essentialfunctionality that enables second users (SUs) to detect spectral holes and toopportunistically use underutilized subbands without causing harmful interference toPU networks. In practice, spectrum sensing must be fast and accurate in order tomaximize the opportunistic throughput without adding unacceptable level ofinterference to the PUs. To guarantee high spectrum efficiency while avoiding theinterference to PUs, the SUs should be able to adapt to the spectrum conditions flexibly.In this thesis, wideband spectrum sensing algorithms are focused on, and thedisadvantages of the traditional spectrum sensing methods are deeply investigated．Inorder to overcome the disadvantages of the conventional spectrum sensing algorithms,this thesis has explored some novel wideband spectrum sensing algorithms based onGerschgorin disk theorem and deflection coefficient theorem, and has acquired someresearch fruits as follows.①Gerschgorin disk theorem based wideband spectrum sensing methods,namely Gerschgorin Likelihood Estimate algorithm (GLE) and Gerschgorin DiskRadii estimate algorithm (GDE)Gerschgorin disk theorem is proposed to perform wideband spectrum sensing for CR.By means of exploring the possibility of utilizing Gerschgorin disk theorem inwideband spectrum sensing, the occupied and the idle licensed subbands weredistingwished by identifying the cardinalities of the corresponding subband sets.Simulation results show that GLE sensing performance remains consistent when theSNR is low, whereas GDE requires no a priori knowledge of the noise power and thePU signal and it overcomes the practical problem of small spectrum observation samples. With salient performance and practical feasibility, the proposed GLE and GDEmay serve as candidate wideband sensing technologies for CR.②Deflection coefficient based optimal cooperative wideband spectrum sensingschemes for CRThis thesis proposes multiple cooperative secondary users serve in the CR network toprovide space diversity for spectrum sensing. After the fusion center acquires theoptimal fusion weights, a global threshold is utilized to obtain the final global decision.Since the proposed optimal cooperative sensing scheme requires precise estimations ofPU signal strengths and the noise variances, an improved sensing algorithm is alsoproposed to facilitate the practical implementation. The proposed schemes significantlyimprove the spectrum sensing performance and outperform the conventional equal gaincombining scheme. The proposed sensing algorithm can effectively implement theoptimal cooperative wideband sensing scheme.