A spectrum sensing technique based on autocorrelation and Euclidean distance and its comparison with energy detection for cognitive radio networks

摘要

Spectrum sensing is an important aspect of cognitive radios. This paper describes a method for spectrum sensing based on the autocorrelation of the received samples. The proposed method was evaluated by means of experiments wherein the probabilities of detection and false alarm at different signal-to-noise ratios (SNRs) were observed. The platform used for the experiments was a set of Universal Software Radio Peripheral (TM) (USRP (TM)) devices acting as radio frequency front ends in combination with GNU Radio software. Since the signal processing was performed in the software domain, Gaussian noise of different levels was emulated by changing the standard deviation of a Python random number generator. In addition, the output power of a signal generator was varied to obtain different levels of SNR A metric called the Euclidean distance was derived to analyze the autocorrelation of the samples received by the USRP (TM) device in order to decide between two possible situations: only noise present or signal plus noise present. The proposed method was compared with two methods: one based on the value of the autocorrelation at the first lag and another one based on the power of the signal, known as energy detection spectrum sensing technique. (C) 2015 Elsevier B.V. All rights reserved.