Identifier

etd-11112005-123648

Degree

Master of Science in Computer Science (MSCS)

Department

Computer Science

Document Type

Thesis

Abstract

Wireless Sensor Network has emerged as an important technology of the future due to its potential for application across a wide array of domains. The collaborative power of numerous autonomousremote sensing nodes self configured into a multi hop network permits in-depth accurate observation of any physical phenomenon. A stringent set of computational and resource constraints make the design and implementation of sensor networks an arduous task. The issue of optimizing the limited and often non-renewable energy of sensor nodes due to its direct impact on network lifetime dominates every aspect of wireless sensor networks. Existing techniques for optimizing energy consumption are based on exploiting node redundancy, adaptive radio transmission power and topology control. Topology control protocols significantly impact network lifetime, routing algorithms and connectivity. We classify sensor nodes as strong and weak nodes based on their residual energy and propose a novel topology control protocol (NEC) which extends network lifetime while guarantying minimum connectivity. Extensive simulations in Network-Simulator (ns-2) show that our protocol outperforms the existing protocols in terms of various performance metrics. We further explore the effectiveness of data aggregation paradigm as a solution to the dominant problem of maximizing energy utilization and increasing network bandwidth utilization in sensor networks. We propose a novel energy efficient data aggregation protocol based on the well-known k-Means algorithm. Our protocol achieves energy efficiency by reduced number of data transmissions at each level of a hierarchical sensor network. Our protocol exploits the spatial and temporal coherence between the data sensed by neighboring sensor nodes in a cluster to reduce the number of packet transmissions. Sensor nodes apply k-Means algorithm to the raw data to generate a reduced set of mean values and forward this modified data set to cluster-head nodes. We further prove the effectiveness of our protocol in providing increased energy conservation in the network by extensive simulation results.

Date

2005

Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Rajgopal Kannan

DOI

10.31390/gradschool_theses.3870

Share

COinS