Research On Spectrum Sensing Based On Compressed Sensing And Exponential Entropy

As the rapid development of wireless communication business, great demand for wireless spectrum resource has been growing exponentially in recent years and the available spectrum resource becomes increasingly scarce. However, on the other hand, a large part of licensed radio spectrum is utilized in low efficiency. Directing at this problem, cognitive radio technology is proposed to improve the utilization of licensed spectrum by employing the way of dynamic spectrum allocation under the premise of no interference with the communication of authorized users. As the precondition, spectrum sensing is the key technology in cognitive radio system.Firstly, the background and current research of cognitive radio is introduced in this paper and then theoretical models of transmitter based and receiver based spectrum sensing algorithm is analyzed in detail respectively. The detection principles of matched filter detection, energy detection, covariance matrix detection and cyclostationary detection are analyzed detailedly, then applicable occasions, advantages and disadvantages of them are summarized. Aiming at the shortcomings of single user detection, cooperative detection of multiple users is introduced and the fusion principles, advantages and disadvantages of hard decision based and soft decision based cooperative detection are analyzed respectively.Furthermore, corresponding analysis of detection performance is shown.In order to save the channel resource and transmitting power, the original signal can be compressed by compressed sensing theory and then transmitted in the channel when it is wideband and compressive. However, in cognitive radio network, the secondary users may still not have a strong sampling ability and on the other hand the data quantity is still too large.To solve this problem, this paper proposes a spectrum sensing algorithm for compressed sampling transmitting signal based on matched filter detection and compressed sensing theory.By adopting the compressed sensing technology, the data quantity in the receiving terminal is reduced greatly and on the other hand, the execution speed is enhanced. Furthermore, the simulation results show that the detection performance of proposed algorithm is better than conventional energy detection algorithm. The parameters which effect to the detection performance are theoretically analyzed in this paper and then demonstrated in the simulation.In addition, directing at some fatal problems of current spectrum sensing methods in cognitive radio, such as poor detection performance in low signal to noise ratio (SNR)situation and poor robustness to the uncertainty of noise power, etc, an exponential entropy-based spectrum sensing algorithm is proposed. According to the difference of exponential entropy between the situation of H0 and H1, the value of exponential entropy of received signal is estimated and then compared with the threshold to judge whether the primary user's signal exists or not. Furthermore, on this basis, a cooperative sensing scheme based on soft decision is proposed to improve its detection performance. The simulation results show that the algorithm proposed also has a strong robustness to the noise power uncertainty and can achieve a good detection performance at low SNR, the cooperative sensing scheme proposed at last also has a better detection performance than traditional cooperative method.