Quantifying Trust for Robust Fusion While Spectrum Sharing in Distributed DSA Networks

In this paper, we quantify the trustworthiness of secondary nodes that share spectrum sensing reports in a distributed dynamic spectrum access network. We propose a spatio-spectral anomaly monitoring technique that effectively captures anomalies in the spectrum sensing reports shared by individual cognitive radio nodes. Based on this, we propose an optimistic trust model for a system with a normal risk attitude and using approximation to the Beta distribution. For a more conservative and risk averse system, we propose a multinomial Dirichlet distribution-based conservative trust framework. Using a machine learning approach, we classify malicious nodes with a high degree of certainty regardless of their aggressiveness of attacks or variations introduced by the wireless environment. Subsequently, we propose two instantaneous fusion models: 1) optimistic trust-based fusion and 2) conservative trust-based fusion, which exclude untrustworthy sensing reports from participating nodes during spectrum data fusion. Our work considers random, deterministic, and preferential (ON-OFF) attack models to demonstrate the utility of our proposed model under varied attack scenarios. Through extensive simulation experiments, we show that the trust values help identify malicious nodes with a high degree of certainty.