Research On Denoising And Completion Technology For High-Order Radio Map Based On Tensor Theory

With the rapid development of wireless communication technology,there is a growing demand for spectrum resources in various fields of society,and the shortage of spectrum resources is becoming more and more serious.However,due to the inefficient allocation strategy of spectrum resources,the utilization rate of spectrum resources is very low.The radio-map-assisted wireless network resource management system can realize the fine management of spectrum resources,and provide an effective way to improve the utilization of spectrum resources.Spectrum data has the characteristics of multiple dimensions such as time,space,frequency,and signal power.Tensor is a high-dimensional array.The radio map represented by tensor can describe the occupancy of spectrum resources from multiple dimensions at the same time,and has great advantages and potential in improving spectrum resource management.The real communication environment is complex and changeable,and the radio map is inevitably corrupted by gross noise.In addition,a portion of spectrum data may be unavailable during the acquisition process.To solve the above problems,this paper focuses on the denoising and completion of high-order radio map based on tensor theory.The main contents and innovations of this paper are as follows:Firstly,based on the characteristics of spectrum data and sparse noise,the problem of removing sparse noise under high-order radio map is modeled as tensor robust principal component analysis(TRPCA),and an improved TRPCA algorithm based on weighted nuclear norm minimization is proposed.By weighting the tensor nuclear norm and extracting the low-rank components from the core matrix whose entries are from the diagonal elements of the core tensor,the large singular values are shrunk less,and the low rank structures in multiway data are further exploited.The proposed TRPCA model is efficiently solved by alternating direction method of multipliers(ADMM),and simulation results show the superiority and effectiveness of the proposed TRPCA algorithm.Secondly,aiming at the problem that a portion of high-order radio map entries may be lost,a completion algorithm based on minimizing the sum of truncated nuclear norms is proposed.The missing data in radio map is completed by tensor completion algorithm.In this paper,the tensor completion algorithm is modeled as minimizing the sum of the truncated nuclear norms of each mode-n tensor,which improves the completion performance of the existing model based on tensor nuclear norm minimization,and avoids the degradation of the completion performance caused by the orientation dependence of the tensor nuclear norms.Finally,the solving process of the proposed completion model is derived,and the ADMM framework is used for iterative optimization.Simulation results show the good performance of the proposed algorithm.