Real-time Link Quality Estimation and Holistic Transmission Power Control for Wireless Sensor Networks

Wireless sensor networks (WSNs) are becoming widely adopted across multiple industries to implement sensor and non-critical control applications. These networks of smart sensors and actuators require energy efficient and reliable operation to meet application requirements. Regulatory body restrictions, hardware resource constraints and an increasingly crowded network space makes realising these requirements a significant challenge. Transmission power control (TPC) protocols are poised for wide spread adoption in WSNs to address energy constraints and prolong the lifetime of the networked devices. The complex and dynamic nature of the transmission medium; the processing and memory hardware resource constraints and the low channel throughput makes identifying the optimum transmission power a significant challenge. TPC protocols for WSNs are not well developed and previously published works suffer from a number of common deficiencies such as; having poor tuning agility, not being practical to implement on the resource constrained hardware and not accounting for the energy consumed by packet retransmissions. This has resulted in several WSN standards featuring support for TPC but no formal definition being given for its implementation. Addressing the deficiencies associated with current works is required to increase the adoption of TPC protocols in WSNs. In this thesis a novel holistic TPC protocol with the primary objective of increasing the energy efficiency of communication activities in WSNs is proposed, implemented and evaluated. Firstly, the opportunities for TPC protocols in WSN applications were evaluated through developing a mathematical model that compares transmission power against communication reliability and energy consumption. Applying this model to state-of-the-art (SoA) radio hardware and parameter values from current WSN standards, the maximum energy savings were quantified at up to 80% for links that belong to the connected region and up to 66% for links that belong to the transitional and disconnected regions. Applying the results from this study, previous assumptions that protocols and mechanisms, such as TPC, not being able to achieve significant energy savings at short communications distances are contested. This study showed that the greatest energy savings are achieved at short communication distances and under ideal channel conditions. An empirical characterisation of wireless link quality in typical WSN environments was conducted to identify and quantify the spatial and temporal factors which affect radio and link dynamics. The study found that wireless link quality exhibits complex, unique and dynamic tendencies which cannot be captured by simplistic theoretical models. Link quality must therefore be estimated online, in real-time, using resources internal to the network. An empirical characterisation of raw link quality metrics for evaluating channel quality, packet delivery and channel stability properties of a communication link was conducted. Using the recommendations from this study, a novel holistic TPC protocol (HTPC) which operates on a per-packet basis and features a dynamic algorithm is proposed. The optimal TP is estimated through combining channel quality and packet delivery properties to provide a real-time estimation of the minimum channel gain, and using the channel stability properties to implement an adaptive fade margin. Practical evaluations show that HTPC is adaptive to link quality changes and outperforms current TPC protocols by achieving higher energy efficiency without detrimentally affecting the communication reliability. When subjected to several common temporal variations, links implemented with HTPC consumed 38% less than the current practise of using a fixed maximum TP and between 18-39% less than current SoA TPC protocols. Through offline computations, HTPC was found to closely match the performance of the optimal link performance, with links implemented with HTPC only consuming 7.8% more energy than when the optimal TP is considered. On top of this, real-world implementations of HTPC show that it is practical to implement on the resource constrained hardware as a result of implementing simplistic metric evaluation techniques and requiring minimal numbers of samples. Comparing the performance and characteristics of HTPC against previous works, HTPC addresses the common deficiencies associated with current solutions and therefore presents an incremental improvement on SoA TPC protocols