Examination Committee

Prof Ilias DIMITRAKOPOULOS, CIVL/HKUST (Chairperson)
Prof Khaled BEN LETAIEF, ECE/HKUST (Thesis Supervisor)
Prof Yahong Rosa ZHENG, Department of Electrical and Computer Engineering, Missouri University of Science and Technology (External Examiner)
Prof Ross MURCH, ECE/HKUST
Prof Chin-Tau LEA, ECE/HKUST
Prof Qian ZHANG, CSE/HKUST

Abstract

The unprecedented wireless traffic growth and fast evolution of mobile applications shall significantly escalate the energy consumption and carbon emission in future wireless networks. Hence, developing green communications and computing systems is of strong needs, and energy harvesting (EH) provides a promising solution. However, powering these systems purely by intermittent renewable energy sources may degrade the quality of service (QoS). Fortunately, hybrid energy supply (HES) where EH and electric grid coexist, and mobile-edge computing (MEC) emerge as effective remedies for green communications and computing systems, respectively. In this thesis, we investigate novel design methodologies for HES wireless systems and green MEC systems in 5G networks.

We first investigate the grid energy consumption and QoS tradeoff in HES wireless networks using a single-user system. The total service cost is proposed to investigate the tradeoff, and base station assignment and power control (BAPC) is adopted to optimize the system. Optimal and low-complexity sub-optimal algorithms are proposed. Simulation results demonstrate their effectiveness and the fundamental tradeoff in HES networks. Next, we study multi-user HES networks and develop a low-complexity online BAPC algorithm. To determine the network operations, we only need to solve a deterministic problem at each time slot by an efficient inner-outer optimization algorithm. The proposed online algorithm is shown to be asymptotically optimal both theoretically and numerically.

To provide satisfactory and sustained computation performance as well as to achieve green computing, we propose an innovative MEC system with EH-powered mobile devices, and develop a dynamic computation offloading policy which greatly reduces task failure with minor latency performance degradation. We further consider general multi-user MEC systems with delay-tolerant applications. A low-complexity online joint radio and computational resource management algorithm is proposed to minimize the system-wise power consumption, which has the worst-case performance guarantee and characterizes the power-delay tradeoff in multi-user MEC systems