Migration and mutation (MeTa) hybrid trained ANN for dynamic spectrum access in wireless body area network

Conventional radio transmission for wireless patient monitoring systems (WPMS) experiences spectrum overcrowding and congestion. To address this issue, dynamic spectrum access systems are needed. Cognitive radio (CR) networks have emerged as a solution capable of dynamically accessing spectrum through efficient spectrum sensing. Artificial intelligence (AI) has become a valuable tool for detecting spectrum opportunities. However, conventional AI-based spectrum sensing methods for the precise prediction and sorting of unoccupied spectrum have limitations in terms of convergence to local optima. This limitation serves as the motivation for proposing the migrated and mutation (MeTa) hybrid optimization algorithm, which trains artificial neural networks (ANN) for efficient spectrum sensing. The novel MeTa hybrid optimization algorithm achieves a balance between global and local search abilities by optimizing the weights of the ANN. Numerical results demonstrate the superiority of the MeTa hybrid NN in terms of detection probability, false alarm probability, and convergence rate compared to existing conventional spectrum sensing schemes. Real-time outcomes based on the Universal Software Radio Peripheral (USRP) further validate the efficiency of the proposed method. Compared to particle swarm optimization (PSO)-ANN, classical ANN, and energy detection-based spectrum sensing techniques, the MeTa NN shows detection probability improvements of 9%, 16%, and 37% respectively. • Developed cognitive radio aware wireless patient monitoring systems with efficient spectrum sensing through novel migrated and mutation (MeTa) Artificial Neural Networks (ANN). • Novel MeTa optimization scheme is implemented for efficient neural network weight training. • Developed migration and mutation-based hybrid optimization scheme for next generation network. • Deployed a MeTa ANN for effective real-time spectrum sensing using USRPs.