Practical design and implementation methods for Kalman filtering for mission critical applications

摘要

The SigmaRho Kalman filter is a new implementation useful for design and application in stressful real-time environments. This filter has been implemented in a satellite navigation receiver. This paper presents application of results of the new mechanization compared with the original matrix form of the Kalman filter and Bierman-Thornton UDUT-factored form. A sample problem from the signal processing area is presented with extreme time variation between samples. This problem, often referred to as software radio, requires estimation of signal phase changes in a repetitive signal (carrier) with additive noise where the digital sampling is at a rate well below the carrier repetition rate. The paper also demonstrates a computationally effective implementation of the SigmaRho filter using integer arithmetic. The implementation is successfully applied at sample rates approaching 100 MHz for a carrier at 1.5 GHz, demonstrating its potential as a signal processing tool for high-speed commercial applications.