On security vulnerabilities and mitigation in the cognition cycle of Distributed Cognitive Radio Networks

Abstract

Cognitive Radio (CR) enables unlicensed or Secondary Users (SUs) to operate in underutilized spectrum (or white spaces) owned by licensed or Primary Users (PUs) conditional upon PU encountering acceptable interference levels. A Distributed Cognitive Radio Network (DCRN) is a distributed wireless network established by a number of SUs in the absence of fixed network infrastructure such as a base station. Cognition Cycle (CC) is the key concept of CR to provide intelligence to SUs so that they are able to sense for white spaces and carry out an optimal or near-optimal joint action on its operating environment for network-wide performance enhancement. The CC can be applied in various applications in DCRNs such as dynamic channel selection, topology management, congestion control, and scheduling. Popular artificial intelligence approaches such as Reinforcement Learning (RL) have been widely applied to realize the conceptual CC. As with other computer networks, there are security aspects in DCRN but the research into them is still in its infancy. To the best of our knowledge, no attempt has been made to date to investigate security vulnerabilities in DCRN as a result of the application of RL. This paper aims to spark new research interest in this area. We provide our initial investigation on security vulnerabilities, as well as their mitigation, in RL-based applications in DCRNs.