Energy Detection Approach for Spectrum Sensing in Cognitive Radio Networks

Abstract

ABSTRACT In the last few years, there has been enormous improvement in mobile communication services. Due to this rapid growth in the field of communications, the demand for wireless spectrum has increased rapidly. In the early days in order to remove the interference problem, the spectrum was assigned statically. To assign the spectrum statically, the fixed spectrum assignment policy was used. Due to the fixed spectrum assignment policy, the spectrum was not efficiently utilized and remained vacant most of the time.Cognitive radio (CR) have been proposed as a possible solution to improve spectrum utilization by enabling opportunistic spectrum sharing. The main requirement for allowing CRs to use licensed spectrum on a secondary basis is not causing interference to primary users. Spectrum sensing allows cognitive users to autonomously identify unused portions of the radio spectrum, and thus avoid interference to primary users. The main objective of this search was to study and performance evaluation of energy detection technique. The process of threshold selection for energy detection is addressed by the Constant False Alarm Rate method and selection is carried out considering present conditions of noise levels. in this thesis simulated and compared Signal-to-Noise Ratio versus Probability of Detection with varying Sample Size, Signal-to-Noise Ratio versus Probability of Detection with varying Probability of False Alarm, ROC curves at different Signal-to-Noise Ratios with Energy Detector, Probability of False Alarm versus Probability of Misdetection with varying Signal-to-Noise Ratio. From the simulation results, it was observed that the energy detector can detect signals as low as -12 dB at the desired probability of detection of 0.9, and probability of false alarm at 0.1, at a sample size of 2048. With the same parameters, the lowest experimental real-time SNR that could be detected was observed to be -8 dB. In addition, the analysis of simulated sensor proved that the sample size affected the performance of the detector, thus implying a correlation between sample size and the energy detector’s performance. The simulated results proved that the lowest sample size that could be used optimally at the desired and values was 512. In addition, the results obtained showed that for the energy detector to achieve high Vprobability of detection, the probability of false alarm should be minimized. Also It is important to note that regardless of the choice of probability of false alarm, the energy detector performs optimally and can easily distinguish a primary user in the spectrum from the noise. In addition, a reduction in the signal strength greatly affects the performance of the detector, which is observed from the reduction in probability of detection, for all false alarms. Below -20 dB, the performance continues to deteriorate significantly, and it becomes challenging for the detector to distinguish the determinant (PU) signal from the noise signal. Finally we show that probability of false alarm versus probability of misdetection, with varying signal strength. the probability of misdetection decreases as the SNR increases. In addition, as the probability of false alarm increases, the probability of misdetection decreases. Unlike the ROC curves, as the area under the CROC curves reduces, the performance of the energy detector increases.