Spectrum sensing for 4G LTE OFDM signals in heterogeneous network using neural network

Abstract

Spectrum sensing is an important requirement in cognitive radios deployed in advanced 4G wireless communication systems. The cognitive radio has to classify the signal of interest from many such signals in its vicinity. There has been wide spread acceptance of multicarrier signal like orthogonal frequency division multiplexed (OFDM) due to its better protection characteristics against channel degradations. Many recent and upcoming wireless standards thus employ OFDM signal, and a CR radio has to classify the OFDM signal operating in a heterogeneous operating environment. In this paper, we have used artificial neural network-based classification of OFDM signal of third-generation partnership project long term evolution (3GPP LTE) signals that is used on 4G wireless networks. We used reference signal-induced cyclostationarity and cyclic prefix property as feature for classification. The 3GPP LTE OFDM signal classification is done in a heterogeneous network environment, in which other OFDM signal from IEEE WiMAX network and other single-carrier digital modulation signal presence are considered. Comparison of classification performance for multilayer perceptron and radial basis function neural network is presented. Effect of two training algorithms, Levenberg - Marquardt (LM) and Back Propagation with momentum, on the convergence rate for training the neural network is presented.