Research On Multiple Antennas Assisted Spectrum Sensing

Due to the increasing popularity of wireless devices in recent years, the radiospectrum has been an extremely scarce resource. However, the fixed frequencyregulation policy drastically conflicts with the high demand for frequency resource.Cognitive radio (CR) is one of the most promising technologies to deal with suchirrational frequency regulation policy and has received lots of attention. Spectrumsensing is aim to find the spectrum hole by detecting the primary user’s (PU) signal.Hence, it is a fundamental issue in CR networks.In practical use scenarios, the sensing period is usually required to be short enoughso that secondary (unlicensed) users (SU) can fully utilize the available spectrum. Inorder to meet the requirements of spectrum sensing, Chapter2proposes two novelmultiple antennas based spectrum sensing algorithms with single SU. The first one isgeneralized likelihood ratio test (GLRT) based spectrum sensing method withoversampling. For the proposed spectrum sensing scheme, enough samples can beobtained during a very short sensing period through oversampling technology. Moreover,the time coherence characteristic of the fading channel can be utilized to improve thesensing performance without any prior knowledge about the variance of noise andchannel information between primary and secondary users. Simulation results show thatthe proposed novel GLRT based detector performs much better than the existing GLRTbased detectors and the energy detector (ED), especially in the low signal-to-noise ratio(SNR) region. Meanwhile, an efficient ED based cooperative spectrum sensingalgorithm with multiple slots (CSSMS) is also proposed in Chapter2. In the proposedalgorithm, the SU softly combines multiple sensing results obtained in different sensingslots to make an optimal decision on the existence of spectrum access opportunity.Moreover, the theoretical performance of the CSSMS algorithm under additive whiteGaussian noise (AWGN) channel is derived. Numerical results show that the proposedCSSMS scheme performs better than the conventional single SU based ED with singleslot.In order to achieve desirable performance under shadowing and fading environments, cooperative spectrum sensing with multiple SUs has been proposed as apromising scheme by taking advantage of spatial and multiuser diversity gain. InChapter3, we investigate cooperative sensing with multiple SUs in a heterogeneous CRnetwork scenario, where each SU may be equipped with different number of receiveantennas and have different signal processing capacity, e.g., sampling rate. Byconsidering the discrepancy in sensing reliability of different SUs, we first extended theexisting researches to propose an optimal cooperative sensing (OCS) scheme based onnormalized energy detection with reporting errors. Meanwhile, the performance of theproposed OCS scheme, the well-known equal gain combination (EGC) method and themaximum normalized energy (MNE) detector with reporting errors is derivedanalytically. Furthermore, in order to avoid the prior information of each SU’s SNR andthe variance of the reporting errors at the fusion center (FC) and simplify thedecision-making process as well as threshold setting, a sub-optimal cooperativespectrum sensing (SOCS) is further proposed. Numerical results show that the proposedOCS and SOCS schemes all perform much better than the existing methods. Secondly,as ED is suffered by the well-known noise uncertainty problem, a novel GLRT basedcooperative spectrum sensing scheme in heterogeneous networks is also proposed inChapter3. By considering the discrepancy of different SUs’ sensing reliability due todifferent hardware configuration, we demonstrate that the GLRT based cooperativespectrum sensing scheme is a linear combination of each SU’s local test statistic. Wealso analysis the proposed algorithm’s robustness to noise uncertainty. Numerical resultsshow that the proposed GLRT based cooperative spectrum sensing scheme performsbetter than the EGC scheme, especially under the case where the sensing reliabilitydiscrepancy of each SU is significant.In order to quickly find the vacant PU’s channels with low overhead, Chapter4propose a database based channel selection algorithm for wideband spectrum sensing.For the proposed algorithm, the CR system prior senses the PU’s physical channel withmaximum vacant probability. Simulation results verify that the proposed database basedchannel selection algorithm for wideband spectrum sensing is more able to find thePU’s vacant physical channel than the traditional random channel selection method.Chapter5summarizes the research contributions of this paper, and shows some valuable research issues for further research in the related areas.