Scheduling for In-Network Sensor Query Processing

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


Title: "Scheduling for In-Network Sensor Query Processing"

By

Mr. Hejun Wu


Abstract

In-network query processing for sensor networks is a cross-layer paradigm,
in which database-style queries are injected into networked sensor nodes
and query results over online sensory data are produced by these nodes. In
this thesis work, we take a scheduling approach to improve the performance
and to provide quality guarantees for in-network sensor query processing.
Specifically, we propose the following three scheduling schemes: (1) DCS
(Distributed, Cross-layer Scheduling), (2) AHS (Adaptive, Holistic
Scheduling), and (3) QAS (Quality-Aware Scheduling).  Each of the three
schemes addresses a distinct aspect of the problem; collectively, they
serve a wide arrange of applications. In DCS, each node utilizes
cross-layer information to determine time slots for query processing and
arranges its own sleep and communication timings in coordination with its
neighbors for reliable query result reporting.  AHS further adapts the
schedules of both query operators and wireless communication to runtime
dynamics for continual efficiency.  QAS maximizes the total quality profit
of a multi-query workload through prioritizing individual queries and
modeling quality status.  Our results on both real-world and simulated
sensor networks demonstrate the efficiency of the proposed schemes.


Date:			Wednesday, 30 January 2008

Time:			10:30a.m.-12:30p.m.

Venue:			Room 3501
			Lifts 25-26

Chairman:		Prof. Francis Lui (ECON)

Committee Members:	Prof. Qiong Luo (Supervisor)
			Prof. Dik Lun Lee
			Prof. Bo Li
			Prof. Danny Tsang (ECE)
			Prof. Jeffrey Yu (Sys. Engg. & Engg. Mgmt., CUHK)


**** ALL are Welcome ****