Dissertation Defenses

Doctoral Dissertation Announcement

---

Candidate: Zill-e-Huma Kamal

Degree of: Doctor of Philosophy

Department: Computer Science

Title: On Implementing a Small Scale Class 2 Opportunistic Network and Utilizing Its Resources with QoS Constraints

Committee:
Dr. Ajay Gupta, Chair
Dr. Ala Al-Fuqaha
Dr. Leszek Lilien
Dr. Ikhas Abdel-Qader
Dr. Matt Mutka

Date: Friday, November 9, 2007 11:00 a.m. – 1:00 p.m.
D-212 College of Engineering and Applied Sciences

Abstract:
The current direction of the internet is towards pervasive/ubiquitous computing, with abstraction of applications as services [e.g. in Service Oriented Computing (SOC), or Web Services], a form of pervasive service oriented computing (PSOC).
Class 2 Opportunistic Networks (oppnets) are the new paradigm that promotes PSOC. Oppnets propose the dynamic interconnection of heterogeneous devices, networks or systems and the integration of their resources, e.g. computation, communication, sensing, actuation, storage, etc., irrespective of the discrepancies in hardware and/or software. The research presents the design and implementation of a small scale oppnet, called MicroOppnet, which acts as a proof-of-concept for oppnets and can be extended as a testbed/prototype.
In any service-oriented network, with the service abstraction in mind, the goal is simple: to meet consumers request for services. The Service Location and Planning (SLP) problem presented is a novel mechanism for meeting consumer’s requests for services with quality of service (QoS) constraints of throughput and delay and underlying link layer constraints, by installing services at a cost, on nodes in the network. When a node in the oppnet is already equipped with a service (resources) then the cost of service installation is minimal.
The research shows that the SLP problem can be solved optimally for small scale networks, with an Integer Linear Programming (ILP) formulation that models the SLP problem and is relaxed for large scale networks using Lagrangean Relaxation technique.
The SLP problem is versatile and can be adapted into various pervasive computing environments, ranging from oppnets, to SOC computing, to Cisco’s AON, etc.