Distributed Estimation in Sensor Networks
Ram Rajagopal and Martin Wainwright
Institute for Transportation Studies
We study decentralized estimation problems under communication constraints. We show practical algorithms for estimating several quantities, including general M-estimators, and analyze statistical performance metrics of these algorithms. We show that our proposed decentralized strategies are optimal estimators under mild assumptions.
- R. Rajagopal and M. Wainwright, "Universal Quantile Estimation with Feedback in the Communication-Constrained Setting," IEEE Transactions in Information Theory (submitted).
- R. Rajagopal and M. Wainwright, "Stochastic Approximation Analysis of Distributed Estimation Algorithms," 45th Allerton Conference on Communications and Control, (invited paper) 2007.