Bluetooth Anomaly Based Intrusion Detection System

Abstract

Bluetooth is a wireless technology that is used to communicate over personal area networks (PAN). With the advent of Internet of Things (IOT), Bluetooth is the technology of choice for small and short range communication networks. For instance, most of the modern cars have the capability to connect to mobile devices using Bluetooth. This ubiquitous presence of Bluetooth makes it important that it is secure and its data is protected. Previous work has shown that Bluetooth is vulnerable to attacks like the man in the middle attack, Denial of Service (DoS) attack, etc. Moreover, all Bluetooth devices are mobile devices and thus power utilization is an import performance parameter. The attacker can easily increase power consumption of a mobile device by launching an attack vector against that device. As a part of this thesis we present an anomaly based intrusion detection system for Bluetooth network, Bluetooth IDS (BIDS). The BIDS uses Ngram based approach to characterize the normal behavior of the Bluetooth protocol. Machine learning algorithms were used to build the normal behavior models for the protocol during the training phase of the system, and thus allowing classification of observed Bluetooth events as normal or abnormal during the operational phase of the system. The experimental results showed that the models that were developed in this thesis had a high accuracy with precision of 99.2% and recall of 99.5%.