Date of Award

5-2018

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

Committee Chair

Yao Liang

Committee Co-Chair

Yu C. Hu

Committee Member 1

Fengguang Song

Committee Member 2

Ninghui Li

Committee Member 3

He Wang

Abstract

This dissertation studies the problem of energy efficient downstream communication in Wireless Sensor Networks (WSNs). First, we present the Opportunistic Source Routing (OSR), a scalable, reliable, and energy-efficient downward routing protocol for individual node actuation in data collection WSNs. OSR introduces opportunistic routing into traditional source routing based on the parent set of a node’s upward routing in data collection, significantly addressing the drastic link dynamics in low-power and lossy WSNs. We devise a novel adaptive Bloom filter mechanism to effectively and efficiently encode a downward source-route in OSR, which enables a significant reduction of the length of source-route field in the packet header. OSR is scalable to very large-size WSN deployments, since each resource-constrained node in the network stores only the set of its direct children. The probabilistic nature of the Bloom filter passively explores opportunistic routing. Upon a delivery failure at any hop along the downward path, OSR actively performs opportunistic routing to bypass the obsolete/bad link. The evaluations in both simulations and real-world testbed experiments demonstrate that OSR significantly outperforms the existing approaches in scalability, reliability, and energy efficiency. Secondly, we propose a mobile code dissemination tool for heterogeneous WSN deployments operating on low power links. The evaluation in lab experiment and a real world WSN testbed shows how our tool reduces the laborious work to reprogram nodes for updating the application. Finally, we present an empirical study of the network dynamics of an out-door heterogeneous WSN deployment and devise a benchmark data suite. The network dynamics analysis includes link level characteristics, topological characteristics, and temporal characteristics. The unique features of the benchmark data suite include the full path information and our approach to fill the missing paths based on the principle of the routing protocol.

Download

Share

COinS