Time-decentralized DOA estimation for electronic surveillance

Abstract

Due to high data rates necessary in multi-channel Electronic Surveillance systems, the full multi-channel data collected on each radar pulse is data compressed to a pulse descriptor word (PDW), which conventionally contains a DOA estimate for the pulse. It is only in down stream processing that the PDWs are clustered into groups identified as originating from single radar. A combined DOA estimate for the radar is conventionally achieved by some form of averaging over the individual pulse DOAs. If one could retain all of the multi-channel IQ data for all of the pulses in a cluster, one could optimally estimate a DOA for the radar using, for example, a maximum likelihood (ML) estimate. In this paper we propose a modification to the initial data compression and subsequent processing which, while adding only a modest amount of data to each PDW, allows estimation of the radar's DOA with performance approaching that which could be achieved by ML estimation using the full multi-channel data records for all the pulses in the cluster. The method is computationally efficient and also allows the optimal beamforming of each pulse without the need to explicitly estimate its DOA.