Multi-Layer Perceptron Based Spectrum Prediction in Cognitive Radio Network

The technology of cognitive radio network (CRN) appears to be an excellent alternative for the efficient allocation of the frequency spectrum between the authorized and unauthorized users, in which the spectrum sensing plays a very significant role. It is often used to identify the availability of the primary user (PU) channel in its spectrum band. The probability of PU channels in high traffic CRN is usually high. While, the secondary user senses only those channels, in the predictive manner, which remain idle after prediction. In this paper, the impact of the collision factor has been considered, and it has been observed that without prediction, the throughput decreases as the collision factor increases. Therefore, the network has been trained through the neural network, based on the multi-layer perceptron model, to reduce the prediction error. After the spectrum prediction, the obtained results have established the improvement in SU's throughput.

Automated summary

The paper discusses the impact of primary user channel availability on secondary users in CRN, and explores the use of neural network training to reduce prediction errors and improve SU throughput.