Traffic-driven intrusion detection for massive MTC towards 5G networks

Massive machine type communications (MTC) have been proposed to be one of the major scenarios in 5G networks, which play the role of supporting future IoT towards ubiquitous information acquisition and exchange. Due to the requirement on low-cost devices, the requirement on anti-intrusion security calls for dedicated research efforts. To meet the MTC security requirements, some security enhancement techniques have been proposed, i.e., the evolved packet system-authentication and key agreement (EPS-AKA). However, with the growth of malicious devices's computing capability and attack ability, security protocol applied on high layer faces more threats and challenges, and even become much less effective. In this paper, we take the essential nature of traffic characteristics into consideration and establish the framework of intrusion detection for massive MTC networks. Then, we propose a traffic-driven intrusion detection scheme, the innovations of which contain two folds: 1) effective estimation of traffic arrival process based on Markovian chain and air-interface access statuses; 2) low-latency yet accurate decision criterion. We also conduct abundant simulations to evaluate our scheme's performances. Simulation results demonstrate that our scheme can well track the arrival process with high accuracy, outperforming the baseline schemes in terms of the not only the detection probability but also the detection time.