A Fair Bandwidth Scanning Strategy to Detect an Adversary

Detecting malicious users or unauthorized activities is a critical challenge in dynamic spectrum access. Traditionally, in such problems, an intrusion detection system (IDS) aims to maximize detection probability. Meanwhile, in networks or radio spectrum problems with multiple nodes or bands, respectively, a protocol that maximizes detection probability might lead to focusing on scanning the most plausible nodes or bands for intrusion and neglecting to scan less plausible ones due to limited scanning resources. To address this challenge, we propose a protocol that maximizes the fairness of detection probabilities across all bands within the bandwidth. We consider α-fairness as a fairness criterion. Using a game-theoretical approach, we model the IDS, which has to decide which of the bands to scan and how long to do it when the IDS faces an adversary who endorses artificial intelligence (AI), enabling it not only to infiltrate the bandwidth without being detected but also to do so in a less predictable manner for the IDS. The equilibrium strategies of the IDS and adversary are derived. An advantage of the fairness detection probability protocol in comparison with the maximizing detection probability protocol is illustrated.