Boosted Trees Algorithm as Reliable Spectrum Sensing Scheme in the Presence of Malicious Users

摘要: Cooperative spectrum sensing (CSS) has the ability to accurately identify activities of primary users (PUs). As secondary users’ (SUs) performance is disturbed in fading and shadowing environment, therefore CSS a suitable choice achieve better results compared individual sensing. One problems occurs due participation malicious (MUs) that report false data fusion center (FC) misguide FC’s decision about PUs’ activity. Out different categories MUs, Always Yes (AY), No (AN), Opposite (AO) Random (RO) are high interest these days literature. Recently, for can be achieved using machine learning techniques. In this paper, boosted trees algorithm (BTA) been proposed obtaining reliable identification PU channel, where SUs access channel opportunistically with minimum disturbances licensee. The BTA mitigates falsification (SSDF) effects AY, AN, AO RO MUs. an ensemble method solving classifiers. It boosts some weak classifiers combination by giving higher weights classifiers’ decisions. Simulation verify improvement existing techniques such as genetic soft (GASDF), particle swarm optimization (PSOSDF), maximum gain (MGCSDF) count hard (CHDF). experimental setup conducted at levels signal-to-noise ratios (SNRs), total number cooperative samples show error probability scheme.