Predictive Kalman Filter Design for Situation Awareness Under Bandwidth Constaints

A situation awareness system must combine information from many different sensors and platforms and construct a composite air picture for the commander. Sensors transmit the target position information over data-links. Due to the large volume of data transmitted over these links, it is essential to conserve the bandwidth of such links by not sending position updates at all times. This proposal seeks to develop an algorithm for minimizing bandwidth demands by using predictive Kalman filters which are widely used in speech compression. With this technique, the sensor updates are estimated as a function of previous updates, with the predictor coefficients changing as the signal statistics change. The difference between the actual and estimated signal statistics change. The difference between the actual and estimated signal is known as the residual signal. This algorithm will minimize the energy or variance of this signal so that they can be quantized with fewer levels than the original input. Thepredictor coefficients are calculated by a Kalman filter algorithm. T he Kalman filter is used to identify a predictable component in the signal which can be described by a few parameters. Bandwidth is conserved by not sending updates if a predicted position is within position oerror requirements, and if maximum bandwidth allications are exceeded.