Researches Of Spectrum Sensing Based On The Minimum Bayes Risk

With the rapid development of wireless communications technology, spectrum resource is increasingly in shortage. How to improve the spectrum efficiency under the situation of the limited spectrum resource is a serious problem. Cognitive radio provides a quite good solution. Spectrum sensing is the prerequisite and basis for cognitive radio applications. Cognitive users find the frequency bands that can be opportunitily used, through spectrum sensing in the wireless network. It improves spectrum efficiency, while avoids interference to the primary user. Therefore, researches on spectrum sensing are of great significance.This paper studies how to optimize the detection performance of spectrum sensing. A new evaluation standard is proposed, that can interpret and evaluate the detection performance of the system better than the existing ones. Under this standard, different spectrum sensing techniques are studied and optimized. Moreover, in cooperative spectrum sensing, the idea of multi-step optimization and adaptive fusion algorithm are proposed.The thesis first introduces the existing spectrum sensing technologies, and then analyzes the existing evaluation standards of the detection performance. Considering the shortcomings of the traditional evaluation standards, a new one is proposed, which called Bayes risk. Based on this standard, minimizing the Bayes risk is regarded as the optimization criteria of the detection performance, and then all of the parameters are analysed and studied.Secondly, under the new evaluation standard, the local spectrum sensing is researched. Taking energy detection as a model, an optimal decision threshold can be obtained by using the minimum Bayes risk criteria. The threshold makes the detection performance best, and simulation results verify the conclusion. Because the local spectrum sensing will appear a lower detection performance in the wireless environments, such as shadow, multi-path fading and hidden terminals, cooperative spectrum sensing based on data fusion technology is proposed. In multi-user cooperative spectrum sensing, where the detection probabilities and false alarm probabilities of cognitive users are the same, a two-step optimization scheme is presented, considering optimization individually in the local detection system and the fusion center system. The expressions of the optimal decision and the optimal decision threshold are given. In multi-user cooperative spectrum sensing, where the detection probabilities and false alarm probabilities of cognitive users are different, according to the minimization Bayesian fusion rule in multi-sensor data fusion, the optimal fusion rule is derived, that is the judgment of the weighted sum of the local results of each cognitive user. Comparing the above two optimization schemes with the other traditional criterion (such as, AND criteria, OR criteria), the proposed optimization schemes are feasible and effective.Finally, there may be some distrusted users in a multi-user system, so an adaptive fusion algorithm is proposed. The adaptive function is embedded in the existing criteria of data fusion, so that distrusted users are identified and removed effectively, thereby the detection performance is improved. In addition to removing distrusted users, cooperative spectrum sensing based on relay cooperation is also researched. By adding a slot for two users amplified forward mode, the selection range of the relay user is expanded, while the detection performances of the helped user and the whole system are both increased. Then, the proposed scheme is continuously studied in the multi-user system. Simulation results show that the detection performance using this scheme is much better than the one using the traditional criteria OR or AND.