Neurobeamformer II: Further Exploration of Adaptive Beamforming via Neural Networks
Abstract
This paper discussed neural network technology as a tool for signal processing. Test results show that the adaptive beamformer method, based on neural network technology, performs the desired function of directing a beam so as to enhance a target signal and reject noise and interference. Comparing test output values with a matched-correlation output shows that the plotted crossbar circuit energy minima follow the shape of an inverted match-filter output. The neurobeamformer has certain advantages of implementation and adaptability over other methods. in concept, it is implementable in analog circuitry with no control code required. Thus a compact, simple, low-cost processor component that is not sensitive to array grooming can be produced. A straightforward adaptive beamformer cannot match the interference-cancellation performance of more exotic methods, which include sidelobe cancellers. So, a neuroprocessor, that will include a neurobeamformer as a component, will be built. This neuroprocessor will provide for cancellation of sidelobes, enhance source discrimination and angle-estimation through interaction of beams. Plans for this extended network were influenced by studies of the literature in biological sensory processing, both peripheral and central. (JHD)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1989
- Accession Number
- ADA215118
Entities
People
- S. L. Speidel