Sectorial direction finding antenna array with a MLP beamformer

Mobile terminal (MT) antennas used for satellite links require a tracking system to minimize the degradation caused by the vehicle's motion on the link performance with the satellite. In practice, azimuthal variations are more severe than elevation fluctuations and a one-axis only tracking system, working over a full 360-degree sector, is found to be appropriate. Typically, direction-finding (DF) can be performed with a monopulse system and beam-steering can be achieved mechanically or electronically. Due to the limited electrical size of MT antennas, the sum and difference beams of the standard monopulse system have poor directivity, which makes the tracking system prone to back-lobe locking. Our objective in this paper is to overcome this difficulty by proposing an enhanced monopulse system which is immune to back-lobe locking. This objective was achieved by the implementation of an artificial neural network (ANN) at the output of the antenna array. In this work, a fixed number of three array elements was used. One of the advantages of using an ANN is that the DF system can be trained to compensate for non-ideal behaviour or time degradations of RF circuit components, antenna elements, radomes etc. Such capabilities are demonstrated by the use of the fitting and regression properties of the multilayer neural feedforward with hyperbolic tangent decision functions.