11:00 AM on Monday, April 22, 2013
Location: 430 Durham Hall
Ph.D. Dissertation Defense for Dinesh Datla
Graduate Advisory Committee:
Dr. Jefrey H. Reed, Co-Chair
Dr. Tamal Bose, Co-Chair
Dr. Allen B. MacKenzie
Dr. Madhav V. Marathe
Dr. Jung-Min Park
Many applications hosted by modern radio networks require abundant computation and power resources in order to process complex computational tasks. Additional challenges are imposed by the need to meet stringent resource constraints and quality-of-service (QoS) requirements. It may be hard for a single radio to fulfill the requirements, mainly due to resource constraints or limited functionality. Traditionally, the computational tasks are processed locally in a single node. As an alternative to this approach, multiple resource-constrained radio nodes can form a mobile ad-hoc cloud computing network (MACNET) and employ wireless distributed computing (WDC) to collaborate on executing complex computation tasks in a distributed manner.
This dissertation proposes an architecture for MACNETs and presents details of a prototype of the proposed architecture. Second, this dissertation presents fundamental models of a typical MACNET environment which capture important WDC performance metrics, namely, power and energy efficiency, response time, and resource availability. Third, the dissertation proposes a workload allocation and scheduling algorithm which can be used for WDC resource allocation in MACNETs.