Date of Award
Doctor of Philosophy
Electrical & Electronic Engineering
Prof. Dirk Pesch
Dr Ramona Marfievici
Our world is increasingly being instrumented with low-power wireless networks that deliver data for diverse application domains such as energy management, health and well-being, security, and other Internet of Things services. As these wireless networks operate in the unlicensed radio spectrum while co-existing with various other communication networks, their communication reliability and energy efficiency are affected by the radio environment. Because of this vulnerability to radio interference, low-power wireless networks may be unable to deliver their application requirements and be inefficiently consuming energy, functioning as a low-dependability network. The situation is exacerbated in indoor environments such as office and residential buildings, wherein collocated wireless devices and electrical appliances impair packet reception, further reducing communication reliability and energy efficiency. The performance of co-existing low-power wireless networks can be increased with a profound understanding of Radio Frequency (RF) noise in the operating environment and by designing more reliable and energy-efficient communication solutions. In this thesis, to better understand the sources of Cross-Technology Interference (CTI), a large set of noise traces were collected using real-world scenarios. Because Wi-Fi is the dominant source of indoor RF noise, initially Wi-Fi traces were collected in order to investigate the impact of Wi-Fi activities on the performance of low-power wireless networks. As low-power wireless devices perceive noise from sources other than Wi-Fi and a handful of work has been done in that regard, the focus of the thesis was moved toward the sensor node’s perspectives wherein how the sensor nodes perceive noise from unknown sources is studied. As the analysis of the traces demonstrated the existence of noise patterns, the traces were further exploited to model their statistical distributions. The accuracy of the noise model to capture wireless activities and the performance of the whitespace model to accurately predict transmissions opportunities, also known as white-spaces, are the key to the proposed approach. These models motivated the vii design of a packet-loss-aware proactive MAC protocol (LUCID) for low-power wireless networks. LUCID was designed for periodic data applications and was evaluated in realistic simulations with varying application data rates and network sizes. LUCID achieves slight performance improvements w.r.t. the state-of-the-art CRYSTAL technique, showing a 1.2% increase in packet delivery ratio, 0.02% decrease in duty-cycle, and 7.4% more energy efficient under bursty indoor radio frequency noise. All these promising results are achieved with a high energy budget required for collecting noise measurements and training the models preferably in a separate identical network. Despite the modest amelioration of the performance of low-power wireless networks, LUCID opens new research directions for further improving the performance of wireless communication networks in general. In summary, this thesis presents a mechanism to analyse patterns in noise traces, a mechanism to use the noise patterns to predict noise-free opportunities for transmission, and a protocol (LUCID) that uses the predicted transmission opportunities to identify rendezvous points for low-interference communication. Investigations presented in this thesis do help to enhance the performance of low-power wireless networks by LUCID, in which nodes utilise the predicted transmission opportunities in a model-based receiver-aware setting.
Abeywickrama Dhanapala, Indika Sanjeewa, "Model-based Cognitive Communications for Low-power Wireless Networks" (2022). Theses [online].
Available at: https://sword.cit.ie/allthe/39
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
info:eu-repo/grantAgreement/IRC/Enterprise Partnership Scheme Postgraduate Scholarship 2014/EPSPG/2014/66/IE/Cognitive Radio Communication Framework for Wireless Sensor Networks//