Intelligent Medium Access Control Protocols for Wireless Sensor Networks

The main contribution of this thesis is to present the design and evaluation of intelligent MAC protocols for Wireless Sensor Networks (WSNs). The objective of this research is to improve the channel utilisation of WSNs while providing flexibility and simplicity in channel access. As WSNs become an efficient tool for recognising and collecting various types of information from the physical world, sensor nodes are expected to be deployed in diverse geographical environments including volcanoes, jungles, and even rivers. Consequently, the requirements for the flexibility of deployment, the simplicity of maintenance, and system self-organisation are put into a higher level. A recently developed reinforcement learning-based MAC scheme referred as ALOHA-Q is adopted as the baseline MAC scheme in this thesis due to its intelligent collision avoidance feature, on-demand transmission strategy and relatively simple operation mechanism. Previous studies have shown that the reinforcement learning technique can considerably improve the system throughput and significantly reduce the probability of packet collisions. However, the implementation of reinforcement learning is based on assumptions about a number of critical network parameters. That impedes the usability of ALOHA-Q. To overcome the challenges in realistic scenarios, this thesis proposes numerous novel schemes and techniques. Two types of frame size evaluation schemes are designed to deal with the uncertainty of node population in single-hop systems, and the unpredictability of radio interference and node distribution in multi-hop systems. A slot swapping techniques is developed to solve the hidden node issue of multi-hop networks. Moreover, an intelligent frame adaptation scheme is introduced to assist sensor nodes to achieve collision-free scheduling in cross chain networks. The combination of these individual contributions forms state of the art MAC protocols, which offers a simple, intelligent and distributed solution to improving the channel utilisation and extend the lifetime of WSNs.