Optimal Sensor Configuration and Survivable Processing with Correlated Noise
Abstract
The research proposed under this project is concerned with the problem of accurate target tracking in the SDI environment. The SDI scenario considered in this report is as follows: Multiple sensors that are possibly located at dispersed geographical locations are observing a single target. The measurement noises are correlated across the sensors. The local processors attached with the sensors possibly may not have the complete information of the target model. The tracking problem can be solved in two ways: first, the sensor measurements can be transmitted to a central node where the optimal state can be estimated conditional on all the measurement history-this is centralized processing. Alternatively, each sensor can process its own measurement locally and send (some function of) processed data to the central node. Here it is fused with other incoming data into the global estimate-this is decentralized processing. The latter is superior to the centralized processing from the system performance considerations in the face of node failures. There exists a good amount of literature in the area of decentralized estimation in presence of uncorrelated sensor noise. In this work, the available results have been extended to the case of correlated sensor noise under the assumption that the local central node only one vector of the dimension of the state.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1989
- Accession Number
- ADA210107
Entities
People
- Raman K. Mehra
- Shahjahan Mahmood