Title:Game theoretic path selection to support security in device-to-device communications

Abstract:Device-to-Device (D2D) communication is expected to be a key feature supported by 5G networks, especially due to the proliferation of Mobile Edge Computing (MEC), which has a prominent role in reducing network stress by shifting computational tasks from the Internet to the mobile edge. Apart from being part of MEC, D2D can extend cellular coverage allowing users to communicate directly when telecommunication infrastructure is highly congested or absent. This significant departure from the typical cellular paradigm imposes the need for decentralised network routing protocols. Moreover, enhanced capabilities of mobile devices and D2D networking will likely result in proliferation of new malware types and epidemics. Although the literature is rich in terms of D2D routing protocols that enhance quality-of-service and energy consumption, they provide only basic security support, e.g., in the form of encryption. Routing decisions can, however, contribute to collaborative detection of mobile malware by leveraging different kinds of anti-malware software installed on mobile devices. Benefiting from the cooperative nature of D2D communications, devices can rely on each other's contributions to detect malware. The impact of our work is geared towards having more malware-free D2D networks. To achieve this, we designed and implemented a novel routing protocol for D2D communications that optimises routing decisions for explicitly improving malware detection. The protocol identifies optimal network paths, in terms of malware mitigation and energy spent for malware detection, based on a game theoretic model. Diverse capabilities of network devices running different types of anti-malware software and their potential for inspecting messages relayed towards an intended destination device are leveraged using game theoretic tools. An optimality analysis of both Nash and Stackelberg security games is undertaken, including both zero and non-zero sum variants, and the Defender's equilibrium strategies. By undertaking network simulations, theoretical results obtained are illustrated through randomly generated network scenarios showing how our protocol outperforms conventional routing protocols, in terms of expected payoff, which consists of: security damage in inflicted by malware and malware detection cost.