Collaborative Spectrum Sensing Based on Distributed Adaptive Filtering

Spectrum sharing technology can reduce the problem of limited spectrum resources failing to meet the demands of multiple users, and it is one of the important topics in effectively integrating existing communication technologies with the sixth-generation mobile communication (6G). Spectrum sensing is generally a crucial prerequisite for achieving spectrum sharing, and designing stable and effective spectrum sensing algorithms is a necessary condition for enabling efficient spectrum sharing among multiple users. However, most existing spectrum sensing methods were designed for stationary signal scenarios, which are clearly unsuitable for non-stationary signal situations that arise in some practical 6G applications. Therefore, we transform the spectrum sensing problem into a distributed filtering problem and propose an adaptive collaborative spectrum sensing algorithm based on the distributed normalized least mean squares (NLMS) algorithm under the assumption of non-stationary signals. Secondary users (SUs) are treated as independent agents, and the communication network is modeled as an undirected graph. Each SU receives the primary user (PU)’s signal and information from neighboring nodes, using a collaborative algorithm to estimate the PU’s signal in order to determine whether the target spectrum can be shared. The main innovation of this paper is to propose a diffusion adaptive cooperative spectrum sensing algorithm based on NLMS (DNLMS-CSS), which can track signals under the assumption of non-stationary signals. The stability of the algorithm is derived, and an upper bound for its estimation error is established. Furthermore, the simulation results indicate that the sensing performance of this algorithm is superior to several other commonly used spectrum sensing methods.