Research On Cooperative Spectrum Sensing Technology Based On Distributed Compressed Sensing

As the wireless communication technique develops so rapidly, that wireless communication standards are emerging in an endless stream, including WIFI、 Bluetooth ZigBee and 3G/LTE etc, which cause the scarcity of wireless spectrum resource simultaneously. According to the Federal Communications Committee (FCC), the licensed spectrum utilization is very low, under the normal circumstance, the utilization of the current communication system ranges from fifteen percent to eighty-five percent. Cognitive radios have emerged as a potential technology to revolutionize spectrum utilization, which can resolve the perceived bandwidth scarcity versus under-utilization dilemma.So it has aroused extensive attention by researchers around the world.Cognitive radios are defined as radio systems that continuously perform spectrum sensing, dynamically identify unused spectrum, and then operate in those spectrum hole where the licensed (primary) radio systems are idle. Spectrum sensing, as the key enabling functionality in cognitive radio sensor networks, needs to reliably detect signals from licensed primary radios to avoid harmful interference. This paper first studies the principle and method of single user energy detection in wide band, and then according to the scarcity in wideband spectrum, the performance of wideband spectrum sensing based on compressed sensing has been tested from different points. Compare with single user spectrum sensing, cooperative spectrum sensing model, based on compressed sensing are proposed, the performance of wideband cooperative spectrum sensing based on distributed compressed sensing are analyzed, the simulation result verify the detection performance of spectrum sensing algorithm under the sub nyquist sampling rate. The study found, in CR network, information of the nodes has the existence of a large number of correlations. If the correlation among nodes can be used in a rational manner, it can not only improve the detection performance of spectrum sensing algorithm, but also can reduce the sampling rate.According to the characteristic of distributed CR network, this paper is based on the correlation among the information which nodes deal with. In this algorithm, First, CR nodes sample the wideband signal in a distributed manner, and then broadcast the samplers to their neighbor nodes. Second, CR nodes jointly recover the signal by the information from neighbors, this method take advantages of correlation of within and between sensed data, can reduce the reconstruction error and power consumption; Third, a diffusion strategy based on distributed estimation is proposed to update the sensing data, the strategy acquires the data from the better neighbor nodes and allows nodes to converge towards the desired solution with acceptable mean-square error bounds, which takes the diffusion matrix as the weighted factor to update spectrum information. Lastly, each node compares the estimated spectrum location information with default threshold, and then makes the final decision individually. Simulation results show that the method not only has a better cooperative gain, but also has a better performance of spectrum sensing than average consensus algorithm. Due to its strong real time and high reliability, it can be well adjust to real network.