Maximum Likelihood Beamspace Processing for Improved Search and Track

 

Richard M. Davis, Ronald L. Fante, William J. Crosby and
Robert J. Balla*
The MITRE Corporation, MS-H143
202 Burlington Rd
Bedford, MA, 01730-1420
Email: rmdavis@mitre.org

* US Army Space and Missile Defense Command

 

Abstract Digitization at the element level in modern phased arrays is currently possible at frequencies below about L-band. Digitization at the subarray level is possible at higher frequencies, but at the cost of many receivers. The advantages which come with digital beamforming (DBF) are many. One of the most promising aspects of DBF is that it supports efficient multi-beam, maximum likelihood (ML) processing. In this paper the authors demonstrate that multiple highly overlapped receive beams (covering the 3 dB transmit contour) provide a practical application for maximum likelihood processing. The use of ML replaces monopulse processing and eliminates the need to form difference beams. The result is a substantial reduction in antenna weight and volume and an increase in performance in both clear and difficult (electronic attack, clutter, weather, etc) environments. The authors have developed a two stage adaptive maximum likelihood beamspace processor (MLBP). The first stage entails adaptively weighting M subarray signals to form N beams. The N beams are then themselves adaptively weighted to form a single composite output beam. ML processing is performed on the N beams to find the weights that cohere the beams in the direction of the target return. The process must be repeated for each range cell. The functional form of the weights shows that the second stage ML processing is equivalent to operating a fully adaptive array of beams. The ML search finds the steering vector to be used in the beam weights. The authors show that the MLBP supports a substantial increase in search coverage compared to traditional approaches for the same number of transmissions. The improvement is due to a reduction in beamshape loss on receive. The MLBP is aslo shown to support improved angle accuracy compared to monopulse. Generic ballistic trajectories are used to compare tracking errors in an extended Kalman filter that uses monopulse versus one that uses the MLBP to generate angle measurements. Trade studies will be presented showing performance of the MLBP as a function of the number of beams used in the second stage. Performance is shown in both the clear environment and in an environment consisting of multiple mainlobe jammers. The authors believe that the combination of adaptive DBF and ML processing will replace monopulse processing in future radars and support substantial improved performance in both clear and difficult environments.

Presentation in PDF format