Research On Cooperative Spectrum Sensing Algorithms For Cognitive Radio Systems

With the rapid development of wireless communication technology and mobile hardware and other industries,the demand for spectrum resources from wireless terminal devices is increasing day by day,which forms an irreconcilable "supply and demand" contradiction with the traditional authorized spectrum allocation scheme,the increasing shortage of spectrum resources has become an urgent scientific problem in wireless communication.The solution to this problem corresponds to its supply and demand: on the one hand,searching for the unexploited spectrum resources,such as THz and visible light and other higher frequency bands,expected to be used in the next generation of mobile communication;on the other hand,many surveys show that the traditional spectrum allocation scheme generally has low efficiency in the utilization of spectrum resources,some idle authorized spectrum needs to be fully utilized.Cognitive radio technology can effectively sense the unoccupied spectrum and allow secondary users to access it dynamically,which greatly improves the efficiency of spectrum resource usage.As the most critical and fundamental part of cognitive radio,spectrum sensing has become the focus of research in wireless communication.Traditional single-user spectrum sensing method is affected by shadow effects,multipath fading,and hidden terminals,making it difficult to achieve ideal sensing performance.Collaborative spectrum sensing effectively overcomes the drawbacks of single-user sensing,and greatly improves the system detection performance by integrating information from multiple cognitive node.Based on this,this paper investigates the collaborative spectrum sensing algorithm under the conditions of multiple collaborative users and multiple frequency bands.The following are the details of the research in this paper.(1)According to the problems of collaborative spectrum sensing of conventional MED,a collaborative MED algorithm based on the power method and polynomial fitting is proposed.In this paper,the computational challenges of the classical collaborative MED algorithm are analyzed for the first time from the perspective of numerical computation,based on which a low-complexity collaborative MED algorithm is proposed by introducing the power method and the least squares-based polynomial fitting method.Specifically,the algorithm uses the power method to calculate the maximum eigenvalues of the fusion center sampling covariance matrix,then uses the least squares-based polynomial fitting method to calculate the threshold,in which the least uncertainty method is used to calculate the best-fit order and the best-fit coefficient.Simulations show that the least-squares-based polynomial fitting method has high computational efficiency and fitting accuracy,and the proposed collaborative MED algorithm can maintain excellent detection performance while reducing computational complexity.(2)The collaborative MED algorithm based on the power method and polynomial fitting has the problems: slow computational convergence and limited computational range of the threshold in large-scale sensing scenarios,a collaborative MED algorithm based on the Rayleigh quotient accelerated power method and cubic spline interpolation is further proposed.By comparing the convergence speed of the Rayleigh quotient accelerated power method and the conventional power method under different conditions,and analyzing the error size of the Rayleigh quotient accelerated power method,the superiority of the Rayleigh quotient accelerated power method in low signal-to-noise ratio and high-dimensional conditions are demonstrated.At the same time,the absolute and relative errors brought by using the cubic spline interpolation method are analyzed,and it is proved that the errors caused by the cubic spline interpolation on the threshold calculation are negligible.The simulations show that the proposed collaborative MED algorithm based on the Rayleigh quotient accelerated power method and cubic spline interpolation can greatly reduce the computational complexity while maintaining excellent detection performance under different signal-to-noise ratios and collaborative user numbers.(3)Considering the shortcomings of traditional convex optimization methods in solving multi-band collaborative spectrum sensing,which is analytically complex and easily falls into local optimality,an multi-band collaborative spectrum sensing method based on particle swarm algorithm with constraints is proposed.The method transforms the optimization problem with constraints into an unconstrained optimization problem by introducing a penalty function.Compared with the traditional convex optimization solution,the proposed method can efficiently search for the global optimal solution and effectively avoid getting trapped in the local optimum,thus improving the performance of the collaborative spectrum sensing algorithm.Simulations show that the proposed method has a very fast convergence speed and can quickly solve the weight coefficients and thresholds of each collaborative sensing node,which effectively improves the total throughput of the cognitive system.At the same time,the simulation analyzes the effects of the number of iterations,the number of populations,and the number of sampling points on the system performance,which provides a reference basis for the further rational setting of relevant parameters in practice.