Advanced Quantum-enhanced Recurrent Neural Network for Efficient Spectrum Sensing in Cognitive 5g Networks

Abstract Spectrum sensing plays an important role in cognitive 5G network for efficient spectrum utilization. The existing spectrum sensing techniques has limitations in terms of low detection accuracy and high computational complexity. To overcome this limitation, a novel model is presented utilizing the benefits of deep learning and optimization algorithms. The proposed model incorporates a novel Quantum Inspired Recurrent Neural Network (QI-RNN) optimized using Hybrid Elephant Herding Bat algorithm (HEH-BA). The proposed QI-RNN utilize quantum principles like superposition and entanglement to improve the learning efficiency and capture complex patterns in spectrum data. Further to improve the performance, the hybrid optimization model combines the global search abilities of elephant herding optimization with local search refinement of bat algorithm. This ensures that the proposed model provides better balance between exploration and exploitation in the hyperparameter space. The experimental results of the proposed model demonstrates that QI-RNN with HEH-BA performs better than traditional Adaptive Wideband Spectrum Sensing (AWBSS), Multi-task Compressive Sensing (MCS-FAM), Convolutional Neural Networks (CNN), and Deep Belief Networks with Sunflower Optimization (DBN-SFO) models. The proposed model high detection probability of 97% with reduced false alarm of 0.01 and energy efficiency of 97% provides a significant improvement in performance and reliability in spectrum sensing.