Optimal value of past samples for decision making in cognitive radio networks

Context: The modeling and prediction of spectrum usage by PUs is an important aspect in reducing interference between SUs and Pus, and for improving overall spectral decision-performance. This process requires past spectral information that might allow the algorithm to model the behavior of the PU. Objective: To determine the optimal value of past samples and recalculation time of decision criteria for the decision-making algorithms in cognitive radio networks. Methodology: Several simulation experiments were carried out using the FFAHP algorithm. Two different approaches were used (real time, and better effort), with high and low traffic, in the GSM frequency band. A statistical analysis of the data obtained is performed, varying the time range parameters while the time criteria remained constant, and vice versa. Results: In high traffic conditions, it is enough to take 1800 previous samples to calculate the initial value of the parameters and update them every 10 minutes (1800). Whereas, in low traffic conditions, 5400 previous samples are needed in order to calculate the initial value of parameters and update them every 10 minutes (1800). Conclusions: A high number of previous samples is not necessary to determine the initial value of the decision parameters in order to obtain a good performance of the handoff rate, nor is it necessary to update those parameters to obtain traffic corresponding to the band of GSM frequency. Funding: The present work is a result of a research project financed by the Center of Research and Scientific Development of the District University Francisco José de Caldas.