Differential Evolution Based Machine Learning Scheme for Secure Cooperative Spectrum Sensing System

The secondary users (SUs) in cognitive radio networks (CRNs) can obtain reliable spectrum sensing information of the primary user (PU) channel using cooperative (CSS). Multiple SUs share their observations CSS system to tackle fading and shadowing conditions. presence malicious (MUs) may pose threats performance due reporting falsified data fusion center (FC). Different categories MUs, such as always yes, no, opposite, random are widely investigated by researchers. To this end, paper proposes a hybrid boosted tree algorithm (HBTA)-based solution that combines differential evolution (DE) (BTA) mitigate effects MUs systems, leading results. An optimized threshold coefficient vector, determined against employing DE, is utilized train BTA. BTA robust ensembling machine learning (ML) technique gaining attention operations. show effectiveness proposed scheme, extensive simulations performed at different levels signal-to-noise-ratios (SNRs) with samples, iteration levels, population sizes. simulation results more decisions be achieved compared individual utilization DE schemes. Furthermore, obtained minimum error exhibited HBTA DE-based Additionally, scheme several other schemes simple BTA, maximum gain combination (MGC), particle swarm optimization (PSO), genetic (GA), K-nearest neighbor (KNN) algorithm-based soft decision (SDF) validate its effectiveness.