Dynamic Spectrum Allocation via Quantum-Driven Learning for Cognitive Radio Systems

The sharp increase in the number of wireless devices has increased the need of dynamic and smart spectrum management within Cognitive Radio Networks (CRNs). The traditional methods of static allocation are not efficient towards responding to real-time spectrum scarcity and dynamic channel environment. In this paper, a new Quantum-Driven Reinforcement Learning (QRL) paradigm is proposed as an efficient and safe method of allocating dynamically the spectrum in CRNs. The suggested model uses the principles of quantum amplitude encoding and Grover-inspired quantum search to allow the Secondary Users (SUs) to acquire the knowledge of spectrum availability at a higher rate of convergence and reduced collision risk with the Primary Users (PUs). The CRN simulation environment has real world limitations, that are multipath fading, user mobility and adversarial interference. The suggested QRL model was tested in comparison with the classical Q-learning, Deep Q-Networks (DQN), and Support Vector Machine (SVM) classifiers. Findings indicate that the QRL model attains 95.0 % efficiency in spectrum utilization,$\mathbf{0. 7 \%} \text{PU}$interference and a faster convergence rate ($\mathbf{3 0 \%}$) than DQN. In addition, the QRL architecture has less inference latency and scales to multiple SU agents, and it is also computationally efficient. These advancements confirm the possibilities of quantum-enhanced learning to real-time spectrum decision-making of next-generation 6 G and IoTdriven wireless networks. The paper draws a conclusion that quantum-based decision agents will be capable of offering adaptive, energy efficient and interference aware spectrum allocation even in highly dynamic settings. Further development of this project will be to execute quantum circuit optimization and to test the model on quantum processors hardware to determine the viability of real-world implementation in distributed spectrum systems.