Network Optimization for Dynamic Spectrum Allocation in Rural Cognitive Radio Networks Using Adaptive Genetic Algorithms in Television White Space Band.

Abstract

Abstract: The extension of reliable broadband into rural environments and an undertaking fundamentally limited by the severe free-space path loss of high-frequency architectures are requirements to close the identified digital divide in the globe. An optimized Dynamic Spectrum Allocation (DSA) model for Cognitive Radio Networks (CRNs) operating within the sub-1 GHz Ultra-High Frequency (UHF) Television White Space (TVWS) bands was developed in this study to address the limitations. Through the design and deployment of an Adaptive Genetic Algorithm (GA), the non-convex, NP-hard nature of multi-objective resource allocation was targeted. Initial localized allocation approaches revealed a 35% network failure rate caused by localized "spectral blindness" and unmitigated Secondary User (SU) co-channel interference and this challenge was resolved by engineering a Network Optimization model. The AI's evolutionary output was grounded in physical constraints. This was actualized due to the explicit incorporation of the deterministic Hata-Okumura open-area propagation model using the advanced multi-objective fitness scalarization. Robust and stable average network fitness of 0.0471 was achieved because of the successful elimination of the interference-induced failures states by Simulation. Absolute compliance with the IEEE 802.22 wireless standard was demonstrated by the global GA by maintaining a strict Primary User (PU) protection threshold of -114 dBm. It was revealed among others that GAs functioning under memory constraints is mathematically viable for scalable, low-power embedded hardware. Also, GAs successfully balanced maximum Shannon-Hartley throughput with aggressive energy minimization. Keywords: algorithm, cognitive, spectrum, optimization, television