Hybrid genetic and shuffled frog‐leaping algorithm for neural network structure optimization and learning model to predict free spectrum in cognitive radio

Abstract

SummaryCognitive radio, which is called an intelligent radio, will dynamically access the available spectrum. This mechanism will bring revolt in wireless communication, which lightens the spectrum utilization problem. Machine learning plays a vital role in every technology. Here, in cognitive radio, it helps train the model to predict the free spectrum. The cognitive clients utilize range‐detecting procedures to detect the groups previously transmitting on them to stay away from impact with the authorized clients that prompts deferral and moderates more vitality. To decrease postponement and vitality utilization and to foresee the future use of channels, range expectation procedures are utilized. Spectrum prediction is used to anticipate the future channel status in light of gathered chronicled information; here, to take care of the issue, the neural system‐based spectrum prediction utilizes a backpropagation training model that has been proposed. To enhance the structure of the neural system and to decrease the forceful weight auxiliary pattern, genetic algorithm (GA) along with the hybrid combination of shuffled frog‐leaping algorithm (SFLA) is proposed. Here, GA has been utilized to abstain from catching nearby ideal solutions. The selection, crossover, and mutation functions were performed to build the haphazardness, which stretches out the populace unite to the set that contains the global ideal solution. SFLA has been proposed for structure improvement; the paired structure has been recommended to demonstrate the memes with the motivation behind building up a subaccumulation with lesser measurements than that of the original group where recognizing affectability and precision would be versatile with that of the primary status. Simulation results show the GA‐SFLA‐based hybrid algorithm that is used has increased the results of getting the best weights by improving the system; additionally, the proposed conspire results show high forecast accuracy.