Data-Driven Security–Reliability Analytics for Cognitive Wireless Networks: Modeling Outage and Intercept Risks in Multi-Hop Relay Systems
Main article
Abstract
Cognitive Radio (CR) networks operating in the underlay spectrum-sharing mode are increasingly deployed to relieve congestion in licensed bands, yet their open-air broadcast nature simultaneously elevates outage risk and exposure to passive eavesdropping. This paper develops a data-driven security–reliability analytics framework for multi-hop CR relay systems that operate over generalized α–μ fading channels and apply joint Transmit Antenna Selection and Selection Combining (TAS/SC) at every hop. Two end-to-end risk metrics are placed at the center of the analysis: Outage Probability (OP) as a measure of reliability risk, and Interception Probability (IP) as a measure of confidentiality risk. We model the secondary transmit power as an adaptive variable that is calibrated against the licensed user’s Quality-Of-Service (QoS) target rather than instantaneous channel state, decoupling our framework from the perfect-CSI assumption that limits many earlier designs. Two protocols are compared throughout: a conventional direct multi-hop transmission (DirecT) scheme, and an incremental cooperative multi-hop scheme (CoopC) in which an external relay is invoked only when the direct link fails. Closed-form OP and IP expressions are decomposed into hop-level components, and a Monte-Carlo simulation campaign of 10⁶ channel realizations is used to validate the model and to populate empirical risk surfaces over the primary transmit power, antenna count, target rate and hop count. Across the operating points considered, CoopC reduces the end-to-end OP by 1–3 orders of magnitude relative to DirecT, while its IP penalty is bounded below 6 % at the same operating point. The hop-count study reveals an interior optimum (M*=4 in our setting) where reliability gains balance against per-hop bandwidth division, suggesting that route-length planning is itself a risk-management lever. The paper closes with practical implications for spectrum policy and CR network engineering. This framing follows the spectrum-sharing foundation established in wireless cognitive networking research. The antenna-selection assumption is consistent with MIMO diversity research. This point further connects the paper with management analytics and industrial information integration research.
