Loading...
Multi-UAV-Enabled Mobile Edge Computing IoT Systems: Joint Association and Resource Allocation Framework
Abu Farha, Yazan Mahmoud
Abu Farha, Yazan Mahmoud
Date
2024-11
Author
Advisor
Type
Thesis
Degree
Citations
Altmetric:
Files
35.232-2024.60a Yazan Mahmoud Abu Farha.pdf
Adobe PDF, 1.6 MB
- Embargoed until 2026-10-10
Description
A Master of Science thesis in Electrical Engineering by Yazan Mahmoud Abu Farha entitled, “Multi-UAV-Enabled Mobile Edge Computing IoT Systems: Joint Association and Resource Allocation Framework”, submitted in November 2024. Thesis advisor is Dr. Mahmoud Ibrahim. Soft copy is available (Thesis, Completion Certificate, Approval Signatures, and AUS Archives Consent Form).
Abstract
Unmanned aerial vehicle (UAV)-enabled mobile edge computing (MEC) in IoT systems has recently gained prominence as a solution to accommodate the increasing quality-of-service (QoS) demands for emerging 5G-and-beyond IoT applications. In this work, we propose a multi-UAV-enabled MEC IoT network architecture, where the UAV-MECs cooperatively provide edge computing services to a large number of terrestrial IoT devices (IoTDs) across a wide rural area for environmental monitoring purposes. We study 2 separate scopes to gain insight into how some of the different design approaches and promising novel schemes in the literature can be utilized to bring about optimal QoS to the IoTDs. In our first scenario, we formulate an optimization problem with the goal of maximizing the total bits offloaded from all IoTDs in a finite service period through jointly optimizing UAV-IoTD associations, IoTD transmit powers, and UAV trajectories for a mobile UAVs case, under given energy budgets and QoS criteria, incorporating uplink Non-orthogonal Multiple Access (NOMA). In our second scenario, we adopt a partial offloading scheme for flexibly partitioning device task data between local and offloaded computation, and formulate a separate optimization problem with the goal of minimizing the maximum task completion latency in the system by jointly optimizing offloading associations, data partitioning, bandwidth allocations, computation resource allocations, and UAV hovering positions in space for a static UAVs case. This is done under energy budget constraints as well as UAV computing capacity constraints. In both scenarios, the formulated problems are highly complex and intractable. A solution approach based on a specialized penalty block coordinate descent (P-BCD) algorithm is proposed, where the original problem is decomposed into subproblems that are solved alternately and iteratively. In addition, the binary offloading decisions are incorporated into the objective as a penalty term. Successive convex approximation (SCA) techniques are also utilized to tackle non-convex expressions. To demonstrate the effectiveness of our proposed schemes for the 2 distinct scenarios, we carry out numerical simulations for each where we study their performance under varying system parameters and with direct comparison to a number of other relevant benchmark schemes, and show superiority of our proposed schemes in terms of sum bits offloaded and energy efficiency in the first scenario, and maximum task latency in our second scenario.