An approach to understand network challenges of wireless sensor network in real-world environments

Lim, Cheng Leong (2019) An approach to understand network challenges of wireless sensor network in real-world environments. PhD thesis, University of Glasgow.

Full text available as:
[thumbnail of 2019LimCLPhD.pdf] PDF
Download (10MB)
Printed Thesis Information: http://eleanor.lib.gla.ac.uk/record=b3367719

Abstract

The demand for large-scale sensing capabilities and scalable communication networks to monitor and control entities within smart buildings have fuelled the exponential growth in Wireless Sensor Network (WSN). WSN proves to be an attractive enabler because of its accurate sensing, low installation cost and flexibility in sensor placement. While WSN offers numerous benefits, it has yet to realise its full potential due to its susceptibility to network challenges in the environment that it is deployed. Particularly, spatial challenges in the indoor environment are known to degrade WSN communication reliability and have led to poor estimations of link quality. Existing WSN solutions often generalise all link failures and tackle them as a single entity. However, under the persistent influence of spatial challenges, failing to provide precise solutions may cause further link failures and higher energy consumption of battery-powered devices. Therefore, it is crucial to identify the causes of spatial- related link failures in order to improve WSN communication reliability.
This thesis investigates WSN link failures under the influence of spatial challenges in real-world indoor environments. Novel and effective strategies are developed to evaluate the WSN communication reliability. By distinguishing between spatial challenges such as a poorly deployed environment and human movements, solutions are devised to reduce link failures and improve the lifespans of energy constraint WSN nodes.
In this thesis, WSN test beds using proprietary wireless sensor nodes are developed and deployed in both controlled and uncontrolled office environments. These test beds provide diverse platforms for investigation into WSN link quality. In addition, a new data extraction feature called Network Instrumentation (NI) is developed and implemented onto the communication stacks of wireless sensor nodes to collect ZigBee PRO parameters that are under the influence of environmental dynamics.
To understand the relationships between WSN and Wi-Fi devices communications, an investigation on frequency spectrum sharing is conducted between IEEE 802.15.4 and IEEE 802.11 bgn standards. It is discovered that the transmission failure of WSN nodes under persistent Wi-Fi interference is largely due to channel access failure rather than corrupted packets. The findings conclude that both technologies can co- exist as long as there is sufficient frequency spacing between Wi-Fi and WSN communication and adequate operating distance between the WSN nodes, and between the WSN nodes and the Wi-Fi interference source.
Adaptive Network-based Fuzzy Inference System (ANFIS) models are developed to predict spatial challenges in an indoor environment. These challenges are namely, “no failure”, “failure due to poorly deployed environment” and “failure due to human movement”. A comparison of models has found that the best-produced model represents the properties of signal strength, channel fluctuations, and communication success rates. It is recognised that the interpretability of ANFIS models have reduced due to the “curse of dimensionality”. Hence, Non-Dominated Sorting Genetic Algorithm (NSGA-II) technique is implemented to reduce the complexity of these ANFIS models. This is followed by a Fuzzy rule sensitivity analysis, where the impacts of Fuzzy rules on model accuracy are found to be dependent on factors such as communication range and controlled or uncontrolled environment.
Long-term WSN routing stability is measured, taking into account the adaptability and robustness of routing paths in the real-world environments. It is found that routing stability is subjected to the implemented routing protocol, deployed environment and routing options available. More importantly, the probability of link failures can be as high as 29.9% when a next hop’s usage rate falls less than 10%. This suggests that a less dominant next hop is subjected to more link failures and is short-lived.
Overall, this thesis brings together diverse WSN test beds in real-world indoor environments and a new data extraction platform to extract link quality parameters from ZigBee PRO stack for a representative assessment of WSN link quality. This produces realistic perspectives of the interactions between WSN communication reliability and the environmental dynamics, particularly spatial challenges. The outcomes of this work include an in-depth system level understanding of real-world deployed applications and an insightful measure of large-scale WSN communication performance. These findings can be used as building blocks for a reliable and sustainable network architecture built on top of resource–constrained WSN.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Additional Information: This research is supported by EDB Singapore, under grant (OY-15-IPP-130006) to NXP Semiconductor Pte Ltd.
Keywords: Wireless sensor network, network reliability, Zigbee.
Colleges/Schools: College of Science and Engineering > School of Engineering
Supervisor's Name: Goh, Dr. Cindy S.F.
Date of Award: 2019
Depositing User: Cheng Leong Lim
Unique ID: glathesis:2019-74340
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 13 Sep 2019 08:43
Last Modified: 05 Mar 2020 21:52
Thesis DOI: 10.5525/gla.thesis.74340
URI: https://theses.gla.ac.uk/id/eprint/74340
Related URLs:

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year