Identifying Distinct Features based on Received Samples for Interference Detection in Wireless Sensor Network Edge Devices

Wireless Sensor Network (WSN) technologies have developed considerably over the past decade or so and, now, feasible solutions exist for various applications, both critical and otherwise. Often these solutions are achieved by using commercial off the shelf components combined with standardized open-access protocols. As deployments diverge into safety-critical areas, attack incentives intensify, leading to persistent malicious intrusion challenges, which are ever-changing as interference techniques evolve and dynamic hardware becomes increasingly accessible. Unique WSN security vulnerabilities, a fluctuating radio frequency (RF) spectrum and physical environment and spectrum co-existence escalate the problem. Thus, securing WSNs is a critical and demanding requirement, heightened by the burden of protecting sensitive transmitted information. This paper, by utilizing ZigBee and Monte Carlo simulations, aims to develop an initial framework for interference detection in WSNs. Initially, bit error location analysis motivates a feature-based detection strategy, relating to both subtle and crude forms of interference. The work expands to analyze Matlab simulated error-free and erroneous transmissions to investigate whether feature useful differences exist. A feature set, including the measured probability density function of, and statistics on, the in-phase and quadrature-phase samples is demonstrated and initially validated/feasibility tested using a designed support vector machine.