Security Assessment of Wideband Spectrum Sensors

Abstract

We investigate security vulnerabilities of wideband spectrum sensors to denial of service (DoS) attacks, launched by an adversary with limited power budget. We survey traditional spectrum analysis methods and compressed-sensing (CS) spectrum sensors in terms of their operation principles and system performance metrics. We develop and simulate end-to-end system models of the wideband spectrum sensors to evaluate their detection probabilities and false alarm probabilities in both non-adversarial and adversarial environments. We show that sweeping spectrum scanners are inherently secure against DoS attacks due to their high dynamic range and small instantaneous bandwidth (BW) equal to their resolution bandwidth. Next, we evaluate Nyquist-rate FFT-based spectrum sensors and show that they are only vulnerable to high-power DoS attacks due to their wide instantaneous BW equal to their Span. These traditional spectrum sensors, however, have high energy consumption for wideband RF spectrum sensing either due to their long scan time or high power. Thus, CS spectrum sensors have recently been proposed as an alternative for RF spectrum sensing thanks to their low energy consumption and fast scan time. A major contribution of this paper is to show that CS spectrum sensors are vulnerable to stealthy DoS attacks (i.e., the attacks are hard to detect). For the same attacker power budget, we further show that the attacks become more potent if the adversary uses multiple attack signals with low power rather than a single attack signal with high power. Finally, we discuss possible countermeasures against the attacks.