An optimised neural network-based spectrum prediction scheme for cognitive radio

A cognitive radio (CR) technology enables all the users to utilise spectrum without interference. There will be a spectrum sensing for all the non-authorised users to perceive the other possibilities of getting a channel. The traffic feature will be unknown to be a priori to design the spectrum predictor with the back propagation (BP) neural network (NN) model and the multi-layer perceptron (MLP).This work proposed an optimised neural network to obtain improved results. The BP algorithm will not require prior knowledge of the real world problems that are trapped within the local minima. This is used widely to solve the problems and found in literature as an evolutionary algorithm like the bacterial foraging optimisation algorithm (BFOA) used for the MLP NN for enhancing the process of learning and improving the rate of convergence as well as accuracy of classification. Performing this spectrum predictor will be analysed using some extensive simulations.