01whole.pdf (5.92 MB)
Download file

Power control in wireless body area networks for interference mitigation

Download (5.92 MB)
posted on 28.03.2022, 23:19 by Ramtin Kazemi
Wireless Body Area Network (WBAN) is a new advanced technology of wireless networking inside and around the human body which has the potential to provide ubiquitous and continuous health monitoring, and reduce the cost of health-care services. This technology benefits from the recent advances in electronics and telecommunications brought about tiny sensor devices which can be implanted inside or attached on the human body. A WBAN is composed of a number of these miniature devices sampling signals of the body and sending them to a coordinator node for real-time monitoring or other medical purposes. Energy is the scarcest resource in WBANs and it is therefore highly desirable to minimize energy dissipation in WBAN devices. One of the major sources of energy waste in WBANs originates from the interference between co-located WBANs working in the same frequency band. To mitigate this inter-network co-channel interference, transmission power control can be employed. A power controller adjusts the transmission power levels in order to maximize some utilities, such as throughput, with the least power. In this thesis, we aim to address the inter-network interference issue in WBANs by proposing practical power control mechanisms to reduce energy consumption and increase throughput as much as possible in WBANs. We design a fuzzy-logic-based power controller which makes decisions on the transmission power level based on the SINR and interference power level. The proposed fuzzy power controller is then optimized o_-line using genetic algorithms to increase throughput and reduce power consumption. Simulation results reveal that the proposed fuzzy power controller strongly outperforms a well-known power controller in the literature, called ADP, in terms of energy consumption per bit and also convergence. We also propose a power controller based on the game theory where players of a non-cooperative game struggle to maximize their throughput with as low power for transmission as possible. We show that a pure unique Nash equilibrium exists in the game. Having found the best response of the players, we evaluate the performance of the proposed approach using simulation and compare its performance with the fuzzy power controller proposed earlier. Simulation results indicate that although the proposed power control game is outperformed by the fuzzy power controller in terms of energy consumption per bit, it is superior in terms of convergence. More importantly, the game power controller enables us to adjust the tradeoff between power and throughput easily and even adaptively, whereas in the fuzzy power controller this adjustment has to be carried out off-line sat the design stage by time-consuming genetic algorithms. We also propose adaptive methods to adjust pricing mechanism, taking into account the power budget and channel conditions of WBANs, which allows them to make the best use of their good conditions to achieve a higher throughput. In an effort to enhance the adaptability and flexibility of the power controller, we employ learning algorithms and put forward a power controller which learns from experience to improve its performance. The proposed controller relies on the reinforcement learning to explore the environment and exploit the knowledge acquired from experience. We use approximation methods to tackle the curse of dimensionality issue and investigate a broad range of reinforcement learning algorithms. We scrutinize the performance of all the proposed approaches by extensive simulations and compare their performances in terms of throughput, power levels, energy consumption per bit and convergence. Simulation results illustrate that although the reinforcement-learning-based power controller suffers from a slower convergence compared to the fuzzy and game power controllers, it provides a better performance in terms of energy consumption per bit. Moreover, the reinforcement learning based power controller enjoys simplicity in design and high level of adaption to environment. Moreover, for applications where meeting QoS requirements is more important than saving energy, we formulate the power control problem as an optimization problem which minimizes the total power consumption and meets the individual target rate of each WBAN. Having attained the optimal solution by using the Lagrange multipliers method, we present a distributed approach based on the Jacobi method for fixed-point calculations, which approximates the optimal solution and is suitable for practical WBANs without the need of any central arbiter. The simulation results indicate that the distributed approach is able to provide good performance which is reasonably close to that of the global optimum solution. Additionally, for cases where the optimization problem is not feasible, the proposed distributed approach provides a better QoS provisioning.


Table of Contents

1. Introduction -- 2. Related work and literature review -- 3. Rate-power tradeoff - genetic-fuzzy approach -- 4. Rate-power tradeoff - game theory approach -- 5. Rate-power tradeoff - reinforcement learning approach -- 6. Minimizing power for target rate -- 7. Conclusions and future work.


A thesis submitted to Macquarie University for the degree of Doctor of Philosophy Department of Engineering May 2013 Includes bibliographical references

Awarding Institution

Macquarie University

Degree Type

Thesis PhD


PhD, Macquarie University, Faculty of Science, Department of Engineering

Department, Centre or School

Department of Engineering

Year of Award


Principal Supervisor

Rein Vesilo


Copyright Ramtin Kazemi 2013 Copyright disclaimer: http://mq.edu.au/library/copyright




1 online resource (xx, 141 pages) illustrations

Former Identifiers

mq:71839 http://hdl.handle.net/1959.14/1278634