An Intrinsic Mode Function based energy detector for spectrum sensing in cognitive radio

Abstract

In this paper, the filtering characteristics of Empirical Mode Decomposition (EMD) are used to create a blind and adaptive energy detector for single or multi-channel spectrum sensing. EMD is an adaptive tool that decomposes time-series signals into a set of modes called Intrinsic Mode Functions (IMF). Due to the EMD filtering behavior, the first IMF is mostly contaminated by noise from the received noisy signal. The proposed approach takes advantage of Cell Averaging Constant False Alarm Rate (CA-CFAR) as an optimal detector to enhance the probability of detection. Alternative to conventional CA-CFAR (which requires at least one nearby vacant channel for good noise estimation), the first IMF will be used as a training function for noise estimation purposes. Based on the first IMF characteristics in frequency domain, the noise floor of the received signal is estimated and a threshold is derived for a given false alarm rate. Simulations show the improvement of the proposed detector in comparison with other conventional detectors.