Signal detection based on pattern classification for use in wireless CPFSK receivers

Conventional methods for signal detection in wireless receivers use well-established methods of signal processing either in the analog or digital domain. In this contribution a method for digital signal detection is presented. Using this method the detection of the received data is performed by using only the zero crossings of the hard-limited IF-signal. It will be shown that a superior performance compared to conventional zero crossing detectors can be obtained by applying methods of pattern recognition and the respective learning algorithms for signal detection. The method can be made very robust with respect to the nonidealities of the receiver using appropriate methods of digital signal processing. Thus significant improvements with respect to performance and implementation costs are obtained compared to classical solutions by shifting implementation complexity from analogue to the digital domain.