A hybrid spectrum opportunities extraction scheme for cognitive wireless communication

Abstract

A realistic and credible model capable of forecasting spectrum occupancy patterns of primary users (PU) needs to be developed to improve the functioning of cognitive radio networks (CRNs) in terms of the spectrum decision process. This will result in more efficient utilization of spectrum resources and the creation of a cognitive radio network. The goal of this study is to design an adaptive spectrum extraction decision mechanism for CRNs to increase the efficiency of spectrum usage. This was accomplished by predicting channel vacancy durations (CVDs) which were generated from measured spectrum consumption data for reverse links of global system for mobile communication (GSM) 900/1800 bands using the long short-term memory (LSTM) approach. Subsequently, an analysis of spatial opportunities of co-existing primary and secondary users within the GSM network was carried out. The potential for secondary users’ communication in the vicinity of the primary user without violating the set carrier-to-interference ratio was equally evaluated. Finally, an algorithm for the spectrum extraction decision process by a secondary user (SU) was developed. The algorithm's performance was measured in terms of primary user interference probability and spectrum utilization increase. MATLAB2018b and EXCEL 2016 were used to do the simulation. The root mean-square error (RMSE) of the CVD prediction for GSM 900 RL and 1800 RL per 60-minute observation period was 2.964 and 2.791 minutes, respectively. The estimated spatial opportunities for primary and secondary systems range from 14.29% to 28.57% for defined areas and around 9% for random places. Thus, primary user interference probability (PUIP) decreased by more than four fold and spectrum utilization increased by up to 50%.