STIR-ing the wireless medium with self-tuned, inference-based, real-time jamming

Jamming broadcasting to intentionally interfere with wireless reception, has long been a problem for wireless systems. Recent research demonstrates numerous advances in jamming techniques that increase attack efficiency or reduce the probability an attack will be detected by choosing attack parameters based on a system's configuration. In this work, we extend the attacker's capabilities by modifying the attack parameters in response to the observed performance of the target system, effectively creating a feedback loop in our attack model. This framework allows for more intricate attack models that are tuned online allowing for closer to optimal attacks against legitimate systems. To show the feasibility of the listening and attacking framework we introduce an attack called Self-Tuned, Inference-based, Real-time jamming or STIR-jamming. This attack listens to legitimate communication traffic, infers the systems performance, and optimizes jamming parameters. We propose the two types of STIR-jamming, mSTIR-jamming and tSTIR-jamming, and implement these attacks against an IEEE 802.15.4 link as a case study. With the empirical results, we demonstrate the attack system adapting to various scenarios and finding stable solutions.