Online-Demodulation stark gestörter winkelmodulierter Signale mit dem Extended Kalman-Filter

Kalman filters offer a good opportunity for processing heavily disturbed signals. Applied to the demodulation problem of communications an Extended Kalman filter is used as a demodulator of angularly modulated signals. In the range of low signal to noise ratios (0 to 17 dB) a number of several filter models is examined in this work /3, 5, 16, 24, 34, 35, 40, 42, 49, 54, 69, 77/.
The two quadrature components, real and imaginary part of a received signal in the baseband are considered as components of a two dimensional observation vector from which phase, frequency and even amplitude and amplitude derivative are estimated.
Supplementary observations as amplitude and phase, adaptive elements and smoother elements have been tested in an online development environment. It has been built up with a DSP (TMS320 C44) to prove data fusion concepts and information absorption in Kalman filters based on examinations which have been theoretical determined in /5, 42/. It proves that even virtually created additional observations like phase and amplitude determined with the two quadrature components lead to an increase of the absorbed information, which leads to better filter results.
Furtheron stability, accuracy and bandwidth have been observed in the online simulation. They are important criteria for online processing, whereas in case batch processing only the accuracy is the important parameter.
Chapter 3 will be laying the basics in signal theory for the following chapter 4, where the different Kalman filter models developed for the demodulation problem will be described. Chapter 5 will show filtering results and Chapter 6 gives a rough description of the DSP development environment used for online-demodulation /2, 33, 76/. A summary in Chapter 7 concludes this work with results and remarks.