Evolution of the radar target tracking algorithms: a move towards knowledge based multi-sensor adaptive processing

Though there are a number of methods for target tracking described in literature like Kalman filtering, extended Kalman filtering, Bayesian approach, IMM-PDA, ML-PDA, particle filters, random set theory, covariance intersection, neuro-fuzzy methods, tracking through genetic algorithms and so on, the goal has always been to bring adaptivity to tackle the changing situations. Since, no one sensor can perform well in all the conditions, Multi-sensor adaptive processing has been the inherent focus. This work presents a brief account of the target tracking algorithms developed till date and to be developed in future and brings out the main development trends. As a novel way of presentation, a Boston Consulting Group (BCG) matrix analysis has been performed and the algorithms have been classified in four classes i.e. Question marks, stars, cash cows and dogs. It has been applied to the radar target tracking algorithms. The evolution and further discussion about future trends clearly show a shift towards knowledge based adaptivity and sensor fusion. Though a number of papers have come out bringing complete account of target tracking algorithms but their presentation format does not provide a way of their practical utilization in the system development. The mathematical formulations are complex and mixing is too much for a non-expert or even a system manager to take decisions. Thus a need was felt to provide a suitable format to the decision makers and provide the non-expert a balanced simple account of the algorithms. Further, a knowledge based perspective has been brought out well in This work. Knowledge based theme though shown in target tracking here is not limited but applies to other areas of radar, ATR, air traffic control & collision avoidance, network centric warfare etc. also. Latest knowledge based research has been incorporated in a broader sense to cover ANNs, CI, fuzzy etc. also.