Neural Network Aided Enhanced Spectrum Sensing in Ognitive Radio

AbstractWireless communication applications are increasing day-by-day. As a consequence efficient spectrum utilization becomes a key task. Cognitive radio is a booming technique for efficient spectrum utilization. Spectrum sensing is a key technology of cognitive radio and it is the first step in cognitive cycle to find out the spectrum availability. The three spectrum sensing methods are matched filter detection, energy detection and cyclostationary. Matched filter method should have the prior knowledge about the primary users signal and it will not be an optimal choice. But no prior information is needed for cyclostationary method and it can extract information about the primary signal waveform. But this method is complex to implement and it is under research. Energy detection is the most common spectrum sensing techniques because this method does not require any prior knowledge about the unknown signal .It is less complex and it takes less sensing time but at the same time it is susceptible to uncertainty in noise power and it cannot differentiate between primary user and secondary user signal. This paper focuses on all three methods and in order to improve the performance of energy detection under heavy noise scenario double threshold technique is also proposed. The presence of primary is determined by three criteria, i.e., probability of detection, probability of miss-detection and probability of false alarm. Simulation results prove that the double threshold method is better than the single threshold