On Radar Privacy in Shared Spectrum Scenarios

Abstract

To satisfy the increasing demand for additional bandwidth from the wireless sector, regulatory bodies are considering to allow commercial wireless systems to operate on spectrum bands that until recently were reserved exclusively for military radar. Such co-existence would require mechanisms for controlling interference. One such mechanism is to assign a precoder to the communication system, which is designed to minimize the communication system’s interference to the radar. This paper looks into whether the implicit radar information contained in such a precoder can be exploited by an adversary to infer the radar’s location. For two specific precoder schemes, we simulate a machine learning based location inference attack. We show that the system information leaked through the precoder can indeed pose various degrees of risk to the radar’s privacy, and further confirm this by computing the mutual information between the respective precoder and the radar location.