Security strategies in wireless sensor networks

This thesis explores security issues in wireless sensor networks (WSNs), and network-layer countermeasures to threats involving routing metrics. Before WSNs can mature to the point of being integrated into daily infrastructure, it is vital that the sensor network technologies involved become sufficiently mature and robust against malicious attack to be trustworthy. Although cryptographic approaches and dedicated security modules are vital, it is important to employ defence in depth via a suite of approaches. A productive approach is to integrate security awareness into the network-layer delivery mechanisms, such as multihop routing or longer-range physical layer approaches. An ideal approach would be workable within realistic channel conditions, impose no complexity for additional control packets or sentry packets, while being fully distributed and scalable. A novel routing protocol is presented (disturbance-based routing) which attempts to avoid wormholes via their static and dynamic topology properties. Simulation results demonstrate its avoidance performance advantages in a variety of topologies. A reputation-based routing approach is introduced, drawing insights from reinforcement learning, which retains routing decisions from an earlier stabilisation phase. Results again demonstrate favourable avoidance properties at a reduced energy cost. Distributed beamforming is explored at the system level, with an architecture provided allowing it to support data delivery in a predominantly multihop routing topology. The vulnerability of beamforming data transmission to jamming attacks is considered analytically and via simulation, and contrasted with multihop routing. A cross-layer approach (physical reputation-based routing) which feeds physical-layer information into the reputation-based routing algorithm is presented, permitting candidate routes that make use of the best beamforming relays to be discovered. Finally, consideration is given to further work on how cognitive security can save energy by allowing nodes to develop a more efficient awareness of their threat environment.