Strategic Placement of Intrusion Detection Systems in IoT Mesh Networks through Machine Learning
Samhitha Perala, Manaswitha Reddy, Sharvari Ravindran, Sasirekha GVK, Jyotsna L. Bapat · 2023
The usage of Bluetooth Low Energy (BLE) mesh networks in Internet of Things (IoT) applications, such as healthcare gives it several advantages such as resilience to network disrupts and better coverage. However, mesh networks are prone to complex attacks, comprising multiple individual attacks like Denial of Service (DoS), Man-in-the-Middle (MitM) and so on. These are referred to as multi-stage attacks that can drastically deteriorate the network’s performance. The need to detect such attacks has been well investigated and several Intrusion Detection Systems (IDS) for detecting and mitigating attacks in mesh networks have been developed. However, the deployment of IDS for monitoring intrusions is costly in terms of computations and the power required. Hence, achieving the maximum probability of intrusion detection through strategic placement for a given limited number of IDS is the challenge. In this paper, Machine Learning (ML)-based analysis is proposed to determine the suitable positions for the IDS on relay nodes of the BLE mesh network. In comparison to the graph theoretical methodology proposed in the prior art, the approach discussed in this paper considers practical aspects, such as the network topology, underlying routing protocol, routing path of the application and variation in the traffic metrics with and without attack.