Skip to main content
eScholarship
Open Access Publications from the University of California

Energy Efficient Distributed Data Fusion In Multihop Wireless Sensor Networks

  • Author(s): Huang, Yi
  • Advisor(s): Hua, Yingbo
  • et al.
Abstract

This thesis addresses a transmission energy problem for wireless sensor networks. There are two types of wireless sensor networks. One is single-hop sensor network where data from each sensor is directly transmitted to a fusion center, and the other is multihop sensor network where data is relayed through adjacent sensors. In the absence of a moving agent for data collection, multihop sensor network is typically much more energy efficient than single-hop sensor network since the former avoids long distance data transmission. Progressive data fusion is a distributed fusion method that fuses data as they hop through sensors, which is effective to further reduce the energy cost. With the knowledge of a routing tree and all channel state information, the transmission energy allocated for each sensor can be pre-determined to even further reduce the energy cost while satisfying a pre-determined performance. In this thesis, we develop several energy planning algorithms for the above purpose. Specifically we designed two energy planning algorithms for progressive estimation with digital transmissions between sensors and one energy planning algorithm for progressive estimation with analog transmission. We also show that digital transmission is more efficient in transmission energy than analog transmission if the available transmission time-bandwidth product for each link and each observation sample is not too limited.

We also study energy cost for consensus estimation which is a distributed fusion method for peer-to-peer multihop sensor networks. The impact of fusion weights and energy allocation for each sensor is also investigated. We demonstrate that to achieve an approximately same performance, the total energy cost for consensus estimation can be much higher than that for progressive estimation, but the peak energy for the former is less than

that for the latter.

Main Content
Current View