Performance comparison of constant coefficient and variable coefficient tracking filters in radar detection

In this paper comparison between constant coefficient and variable coefficient tracking filters for radar has been presented. αβγ filter and Kalman filter are discussed as instances of constant coefficient and variable coefficient filters respectively. Both of the filters are sophisticated smoothing and prediction tracking filters which are used in sampled data target trackers. As polynomial predictor linear recursive filters they can construct future position based on present position measurements. For the same target model the mean residual error from both of the filters are presented to compare their relative performance. Noise variance was varied to observe its effect on both filters. Performance of αβγ filter becomes deteriorated with increased noise whereas Kalman filter can track a target with high accuracy even with increased noise variance. These clearly illustrate the advantage of using a variable coefficient Kalman filter over a fixed coefficient αβγ filter.