Desempenho do método CPSC em redes de rádios cognitivos sob canais rápidos e seletivos

Abstract

The recently proposed spectrum sensing using cooperative power spectral density cancellation (CPSC) has the promising advantages of being robust against noise uncertainty and having low computational complexity. However, to the best of the author’s knowledge, its performance over a channel that varies during the sensing period has not yet been taken into consideration. In this dissertation, the performance of the newly proposed CPSC method is investigated over both frequency-selective and fastfading Rayleigh channel models. Two approaches to calculate the decision variable in the CPSC algorithm were initially compared. Following that, the model is adjusted to vary its complex gain values over a single sensing interval. The fading model is and then also made frequency-selective by modeling additional paths at the receiver. Performance was measured by running simulations to find the Receiver Operating Characteristic (ROC) curves and the maximum achievable throughput curves of a cognitive network sensing with CPSC. The obtained results show that the probability of detection is improved as the number of collected samples or the number of cooperating cognitive users increase, when compared to the energy detection spectrum sensing. In the secondary user’s network, optimum sensing times were found in order to maximize the network’s throughput. Lastly, when the sensing time was time-variant, it was seen that performance was increased as the channel variability increased. In this last analysis, performance was also increased when operating through frequency-selective fading channels, though to a lesser extent. This behavior must be considered when designing a network that intends to use this technology.