Improved design and evaluation of adaptive antennas using interactive computing

The design of adaptive antennas is considered from the point of view of topology, index of performance, and algorithms to be used. First, the topology of a generalized linear adaptive network is shown to reduce to the network topology now in popular use in the adaptive antenna field, and the advantages and disadvantages of it as well as others are reviewed. The indicies of perfomance most widely used which include Least Mean Squared Error (LMSE) and Signal to Noise ratio (S/N), are reviewed and compared. More effective indicies are recommended. These include an index which maximizes S/N subject to a constraint on beam bandwidth, one which uses a weighting of interference (jammer) power as well as angle (which enables more efficient elimination of interferences near the signal although having lower power); an index which limits the magnitude of the gains as well as maximizes S/N; and one more suitable for tracking time varying signals. Algorithms are then compared for use with the above indicies and topology. The broad classes include gradient algorithms and stochastic approximation algorithms, and are compared with respect to complexity, speed, and implementation. The integration of the above aspects results in a complete adaptive antenna system. Following a brief summary of the various techniques used (e.g. null steering, side lobe cancellers, etc.) in terms of these three points of view, the evaluation of adaptive antenna systems is considered using interactive computer graphics. A description of the minicomputer based system used in the study is given, and dynamic results which demonstrate the effect of algorithms and performance index choice are shown as developed on the system. The usefulness of interactive computer graphics in the design and evaluation phase is emphasized.