Font Size: a A A

Research On Cooperative Spectrum Sensing Algorithms For Cognitive Radios

Posted on:2012-02-02Degree:DoctorType:Dissertation
Country:ChinaCandidate:K T CaoFull Text:PDF
GTID:1118330368988047Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
With the rapid development of wireless communications technologies and the applications, the last decade has witnessed the growing demand for wireless radio spectrum. This has resulted in that wireless spectrum has been increasingly scarce, which has led to the obstacle to the improvement of wireless communications technologies. Cognitive Radio (CR) is the optimal choice to solve the above problem. Spectrum sensing enables the capability of CR to detect, learn and be aware of the wireless electromagnetic parameters, and it determines whether CR can be applied to the practical system or not, is one of the key techniques for cognitive radio.In this dissertation, the centralized cooperative spectrum sensing algorithms are mainly focused on, and the disadvantages of the traditional spectrum sensing methods are deeply investigated. In order to overcome the disadvantages of the conventional spectrum sensing algorithms, this dissertation has explored some novel cooperative spectrum sensing algorithms based on cooperative game theory and random matrix theory, and has acquired some research fruits as follows.(1) Based on the distribution of the maximum eigenvalue of random matrix in asymptotic spectrum theory, a DET (Double Eigenvalue Threshold) cooperative spectrum sensing algorithm is proposed.DET can find the two maximum eigenvalues for hypotheses H0 (signal does not exist) and H1 (signal exists) by analysis of the sample covariance matrix of the received signals using Asymptotic Spectrum Theory (AST). The two maximum eigenvalues are regarded as two thresholds to decide whether the transmitted primary user signal is present or not.(2) In order to overcome the disadvantages that the maximum eigenvalue based algorithms are sensitive to the noise uncertainty, a ME-ED (Maximum Eigenvalue-Energy Detection) is presented in this dissertation.The proposed algorithm exploits the ratio of the Maximum Eigenvalue to Energy Detection (ME-ED) to determine whether the Primary User (PU) is absent or not. Through the theoretical analyses, ME-ED scheme can work well without the knowledge of the PU priori and the noise power. In addition, ME-ED algorithm is not sensitive to noise uncertainty at all, and can further improve the sensing performance and robustness.(3) A cooperative spectrum sensing game model has been constructed, based on which a novel cooperative spectrum sensing approach using General Nash Bargaining Solution (GNBS) is presented.In the centralized cooperative spectrum sensing scenario, different cognitive users (Secondary Users, SUs) have different average SNRs and different decision thresholds due to the different spatial position and wireless electromagnetic environment for each SU. Therefore, each SU contributes differently to the final sensing result, and has different weight at the fusion center. According to the aforementioned circumstances, the proposed algorithm exploits GNBS in Cooperative Game Theory (CGT) to construct a two-user cooperative sensing Nash bargaining solution problem. In the proposed scheme, Sensing creditability degree is used for characterizing effects of the distance and channel parameters on the sensing creditability, and the sensing performance for two-user case is derived by using the optimization method. For multi-user case, all SUs are grouped into pairs called coalitions with the assignment method, and for each pair, the sensing performance is obtained based on the two-user method. Finally, the sensing results for each pair are weighted at the fusion center to acquire the final sensing performance.(4) A novel asynchronous cooperative spectrum sensing (ACSS) algorithm is presented in this dissertation, which is based on the occupancy model of licensed band and Bayesian decision rule.In practice, each secondary user (SU) can not synchronously obtain the sensing information and make the decision, and the sensing data from all SUs can not be transmitted to the data fusion center at the same time. What's more, primary users'behaviors which can be modeled as the On-Off model also affect the detection performance of CR. The proposed algorithm ACSS takes all these factors into account. In ACSS, the asynchronous soft decision results from secondary users are transmitted to the fusion center, and these soft decisions are weighted based on the likelihood ratio from Bayesian decision rule and combined at the fusion center. In addition, the fact that each SU has different SNR is considered in ACSS. Compared with the traditional synchronous cooperative spectrum sensing (SCSS) algorithms, the proposed algorithm is theoretically more reasonable, more credible and more suitable to the practical situation.
Keywords/Search Tags:Cognitive Radio, Cooperative Spectrum Sensing, Random Matrix Theory, Sample Covariance Matrix, Maximum Eigenvalue, Cooperative Game Theory, Sensing Creditability Degree, Bayesian Decision Rule
PDF Full Text Request
Related items