Research On Spectrum Sensing For Cognitive Radio

Cognitive radio is a smart wireless communication system which can automatically sense the surrounding radio environment and adjust operating parameters in real time to adapt to changes of the external environment. Without affecting licensed users'normal communication, wireless equipments which possess the cognitive function can access to licensed spectrum holes in an opportunistic way and use the resource dynamically. As an important means of exploring characteristics of vacant spectrum and improving resource utilization, cognitive radio technology is currently one of the hottest researches in the field of radio.Based on detecting the free bands, the research on this paper is focused on several algorithms and their improvements for spectrum sensing, which is a key technology in cognitive radio. After the introduction of basic approaches used in spectrum sensing, energy detection, cyclostationary feature detection and cooperative sensing are chosen to mainly research in the proposed algorithms.First, based on the Welch's periodogram method, a spectrum sensing algorithm for cognitive radio using energy detection in frequency domain is proposed. By further smoothing the received signal data and introducing the mechanism of sliding window function to deal with, an algorithm based on maximum-average energy ratio of spectral window is proposed to improve the previous one. Simulations analyze the sensing performance under different simulation parameters in the low SNR environment, and the improved algorithm enhances the performance.Secondly, a spectral correlation feature detection algorithm based on cyclostationary strength measurement is researched. The detection statistic and time-domain smoothing estimation are derived. Some specific couples of cyclic frequency and spectral frequency are selected to form single detection points, and an improvement adopting multiple points'weighted-combination is proposed to increase detection reliability. Simulation results show this feature detection method is effective and its sensing performance is improved when using the proposed algorithm.Then, based on the estimated covariance matrix, a spectrum sensing algorithm for cognitive radio is researched. According to the basic theory of statistical covariance, this matrix is estimated by using continuous sampling data of licensed user and noise which the cognitive user receives, and a detection statistic is built from the matrix. Simulation results demonstrate that the algorithm possesses high feasibility and low computational complexity, and it also has good performance in the low SNR environment.Finally, a multi-cognitive radio verified cooperative spectrum sensing algorithm based on delay information of received signal. Each detector respectively detects the data segments from the received signal in different delay accumulations, and cognitive network center makes fusion of the local detection information with different decision-fusion strategies. According to each cognitive user's contribution to the sensing performance, apply two kinds of weighted-updated improved schemes to give different dynamic weights to the users that take part in cooperation. Based on the decision-fusion rules, simulation analyzes the sensing performance under different SNR conditions. Results show that the proposed algorithm with its improvement is reliable and effective, which improves overall sensing performance of the cognitive network.