Abstract
Fast advancement in physiological sensors, low-power circuits, and short-range wireless communications has enabled the development and deployment of ubiquitous wearable networks called Wireless Body Area Networks (WBANs) that consist of a few or tens of miniaturised body-mounted or implanted sensors. WBAN is an appealing technology for a wide range of applications including health monitoring, athlete training, and entertainment. In most applications, the sensor devices need to report the measurements to a body central gateway (e.g., a smartphone) through wireless communication, and the gateway can further send the measurements to a remote monitoring centre through the Internet or cellular networks. As a particular type of wireless sensor networks (WSNs), WBANs inherit many characteristics of traditional WSNs such as restricted energy capacity, limited computation capability, and unreliable wireless communication. Radio links in WBANs commonly experience highly time-varying attenuation due to dynamic network topology and frequent occlusions caused by body movements, making it challenging to design a real-time reliable, energy-efficient, and interference-aware communication protocol for WBANs.
Since communication protocols with fixed transmission power are not able to perform well, this thesis proposes communication protocols to adaptively adjust the transmission power level of the sensor nodes through which they can save energy and reduce potential interference while still guaranteeing high communication reliability. Firstly, a new channel prediction model for power- adaptive communication in WBANs is proposed. To this end, the channel behaviour in WBANs is precisely analysed, which demonstrates that the existing channel models either cannot accurately predict channel burstiness or have high prediction complexity. Then, a new, accurate lightweight Markov model is presented to allow the dynamic power adjustment at a per-transmission level. Motivated by the accuracy of the proposed channel model in terms of channel estimation, a channel-aware deadline-constrained policy for scheduling packet transmissions is proposed. The experimental results demonstrate that the proposed scheme can self-learn the channel burstiness patterns, and choose the best transmission power to reduce energy consumption and improve communication reliability. With retransmissions, this scheme can achieve almost 100% on communication reliability but significantly reduces the number of packets transmitted using high power levels. Since the proposed channel model requires a continuous channel measurement, it only supports applications with periodic constant packet rate.
Secondly, Chimp, a learning-based power-aware communication protocol is proposed in which each sending node can self-learn the channel quality and choose the best transmission power level to reduce energy consumption and potential interference but achieve good communication reliability. Chimp is designed to support event-driven applications as well as the applications with periodic constant packet rate. It is designed based on learning automata that use only the acknowledgment packets and motion data from a local gyroscope sensor to inform the real-time channel status. A new learning function is developed that can guarantee to select the optimal transmission power level to minimise the cost function for a given channel quality. For highly dynamic postures such as walking and running, Chimp exploits the correlation between channel quality and motion data generated by a gyroscope sensor to quickly estimate channel quality, eliminating the need to use expensive channel sampling procedures. The performance of Chimp is evaluated through experiments using TelosB motes equipped with the MPU-9250 motion sensor chip and compared with the state-of-the-art protocols in different body postures. Experimental results demonstrate that Chimp outperforms existing schemes and can work efficiently in most common body postures. It achieves almost the same performance as the optimal power assignment scheme for high date rate scenarios.
Thirdly, a localisation algorithm is proposed to find the location of the sending node by combining the locally measured orientation information of the sending node with anatomical constraints of the body parts. The results of the experiments show the channel quality between the sending node and the gateway is highly dependent on the location of the sending node. Motivated by this, Tuatara, a location- and learning-based communication protocol for WBANs is presented to support event-driven applications as well as applications with periodic constant packet rate, even at extremely low packet rate scenarios. Tuatara benefits from the location information provided by the localisation algorithm and prior knowledge about the optimal power level in different locations of the sending node around the body. The experimental results show Tuatara can perform almost as well as the Optimal scheme.