Download files
Abstract
Wireless intrusion detection systems have recently attracted considerable attention, specifically in the context of location-spoofing attacks. In the civilian domain, location spoofing is of growing concern due to extensive applications for future Intelligent Transport Systems (ITS), and emergency ad hoc networks. In the military domain, location spoofing attacks are of prime concern for networked multi-robot systems such as those deployed for e.g. mine sweeping, border surveillance and identification.
In this thesis we will explore in detail the issue of location spoofing in emerging mobile networks. Specifically, we will investigate how to determine the time frames required for a user to be marked 'honest' or 'malicious' at required detection and false positive-rates. We also investigate the time frames needed to meet specific location verification accuracy. Having determined these time frames we will then investigate the additional time required for the mobile nodes to be moved to the required positions, necessary for the position verification, under distributed-control algorithms. Such distributed control algorithms, where each mobile node possesses no intrinsic accurate location information but only range estimates from its nearest neighbours, represent the likely emergency ad hoc deployment scenario. We now specify the contributions of the thesis in more detail.
The first main contribution in this thesis is the development of a novel framework for a wireless intrusion detection algorithm that identifies malicious nodes which are not at their appropriate locations. We explicitly determine how the performance of the algorithm, as measured by detection and false positive rates, is influenced by the amount of tracking information collected. Based on the implementations of particle filters and detection thresholds set by Cramer-Rao Lower Bounds, we show how our tracking verification algorithm is capable of verifying any reported positions within a specific time frame at the required performance rates.
The second main contribution in this thesis is the development of a new motion coordination scheme that can be employed in order to minimise the detection time for a specified location verification accuracy. We develop a greedy approach that achieves the optimum trajectory that minimises detection time in every sampling time. We also develop a sub-optimal approach to this problem that provides a good trade-off between performance and computational resources.
The third main contribution of this thesis is the development of novel distributed control algorithms for controlling the positions of the mobile nodes that participate in the location verification. In these algorithms the nodes participating in the verification, use ranging information from nearest neighbours in order to align themselves in the best positions for the required location verification task. Such studies allow us to estimate the additional time required for our verification systems. This additional time arises because any realistic range-based distributed-motion control incurs latency costs in moving a node to a specific position. As an aside, our studies of distributed-motion control are also used to answer some key open questions regarding detection coverage in the context of military robotic detection systems.