Optimal parameter values for GPS post-processing: An experiment

Abstract

Mode detection stage is a crucial step of post-processing of GPS raw data. In the procedure developed by Schüsssler and Axhausen (2008), a fuzzy logic based algorithm is used for mode detection of each stage of a trip. This fuzzy logic uses fuzzy variables with three or four membership functions. Each membership function has four key points or parameters. This paper deals with optimization of these parameters to give best possible match of the results from GPS data post processing. Calibration of these parameters was carried out with the help of SPSS and Microsoft-Excel. A program was developed which was able to measure the stage by stage modal mismatch between the results from GPS data post processing and a sample of well-coded GPS data. Final optimized parameter profile produces considerably less modal mismatch than the original profile that was used before.