PROTEUS: A Secure Proactive Matrix Completion based Spectrum Allocation in Energy Constrained Cognitive Radio Networks

Various proactive prediction-based channel allocation protocols have been proposed for cognitive radio networks (CRNs) that are highly dependent on the availability of channel sensing data. To handle this, recent works that propose channel allocation techniques in CRNs, recommended security overlays for collaborative spectrum sharing to implement simpler prediction-based algorithms for channel allocation to save energy. However, such algorithms show deteriorated performance and restricted scalability in very large heterogeneous CRNs like Internet of Things (IoT) based Wireless Sensor Networks (WSNs) due to sparse availability of spectrum sensing data. We propose PROTEUS, a secure completion matrix-based channel allocation algorithm that provides better performance in large CRNs compared to recently proposed prediction-based channel allocation protocols. PROTEUS shows 6.2%, 7.5%, 8.2% and 31.28% improvement in channel allocation, backoff rate, sensing delay and energy consumption compared to recently proposed algorithms like ProMETHEUS [1] and ProBLeSS [2] when under Spectrum Sensing Data Falsification (SSDF) Attack. It also showed 66.4% reduction in memory utilization when compared with Nuclear Norm [3] matrix completion based channel allocation. PROTEUS security was analyzed by comparing the empirical and theoretical data to establish an upper-bound for tolerance to number of malicious SUs (14.25%) under SSDF attack thereby highlighting its possible practical usage scenarios.