Computational Complexity of the Kalman Tracking Algorithm

Abstract

It is known that the efficiency of the parallel implementation of the Kalman tracking filter can be improved by reordering the elements of the state vector in a suitable manner. It is shown in the present work that the use of such a reordered state vector in place of the standard one can, by itself, result in a substantial reduction in the computational load, when the system matrices are partitioned into smaller dimension matrices, even in a uniprocessor implementation of the filter.

Indexing terms:

• Kalman filters
• Target tracking
• Matrix partitioning

Additional information

Notes on contributors

V Vaidehi

V Vaidehi, has received the BE degree in Electronics and Communications from College of Engineering, Guindy, Chennai, Madras University in 1978, ME degree in Applied Electronics from Madras Institute of Technology (MIT), Chennai, Madras University in 1980. Currently she is doing her research in the area of parallel processing. Before joining MIT in 1982 as a lecturer she was with School of Automation, Indian Institute of Science. Currently she is an assistant professor in School of Instrumentation and Electronics, MIT, Anna University. Her research interests are in parallel processing, scheduling algorithms, control algorithms, genetic algorithms and neural and fuzzy logic.

C N Krishnan

C N Krishnan, was born in Kerala in 1947 and did his BTech at IIT Madras, and MTech and PhD at IIT Kanpur in the Electrical Engineering Departments. He has been with the Madras Institute of Technology of the Anna University since 1977, and is a Professor in the School of Instrumentation and Electronics there. Prof Krishnan is presently involved in teaching and Research in the areas of spread spectrum techniques and the GPS signal processing and communications.