Performance Analysis of Reward Based Learning Technique Cooperative Spectrum Sensing (RL-CSS) in Cognitive Radio Networks

Spectrum allocation for secondary users (SUs) is a prominent task for improving communication and optimal channel utilization in cognitive radio networks (CRNs). Due to autonomous nature of CRNs, spectrum counterfeits in data sensing is a common threat experienced by the secondary users. In this manuscript, a reward based learning (RL) is presented for mitigating the influence of data falsification in cooperative spectrum sensing (CSS) process. The legitimate secondary users exploit the neighboring information to access and process data within the allocated spectrum. The decision of the fusion center (FC) helps to improve the reliability of other secondary users by sequentially evaluating the reward of the SUs based on detection probability. The process of SU reward estimation is based on the previous communication factors with the consideration of variance and normalization errors at the time of channel detection. The two different process of channel detection rate and reward assignment improves the rate of detection with a controlled delay. Experimental results verify the consistency of the proposed RL-CSS method by improving network throughput and detection rate and minimizing detection delay and channel disconnection ratio.