Intrusion Detection System Focused on Deep Learning for Mobile IoT Networks

Detection mechanisms for intrusion play a key role in the identification of vulnerable behaviors that denigrate IoT networks. Mobile Adhoc Networks (MANETs) are wireless networks with the ability to transmit data without the need for infrastructure to run them. Recently, the IoT networking model has arisen, which is a superset of the above conceptualization. Its fragmented existence and minimal availability of resources pose one of the major issues to provide secure data communication. Intrusion Detection System (IDS) is indispensable for acclimatizing with difficulties. Researchers have traditionally used a heuristic approach focused on the heterogeneity of appropriately classified cases (CCIs), with a cumulative estimate of fluctuation (AMoF). It consists of 2 phases; phase-1 accumulates values by dedicated sniffers (DSs) to produces the CCI which is further transmitted to the supernode (SN). SN conducts the linear regression procedure over the CCIs acquired from DSs in phase-2 to separate the benevolent and vulnerable nodes. For various extreme network situations, the identification characterization is provided in this work using Gauss Markov (GM), and Random waypoint (RWP). In the enhanced velocity environment, detection rates surpass 98%, whereas, 90% is registered for limited speed.