An efficient, microprocessor-based FFT implementation for FDM demultiplexing

The author presents an application of the FFT (fast Fourier transform) to the communications signal processing operations involved in demultiplexing a frequency-division multiplexed (FDM) signal. The inherent parallelism of the FFT algorithm is exploited to provide simultaneous down-conversion, filtering, and baseband time-domain outputs for each channel of a FDM structure. High radix and Winograd Fourier transforms are utilized to provide an efficient FFT implementation and an optimum mapping of the algorithm onto the architecture of the TMS320 series of DSP (digital signal processing) microprocessors. The completed design represents a single chip implementation that is applicable to real-time demultiplexing of small-capacity (12/24 channels) FDM structures. It is also shown that this algorithm can be easily extended to a multiprocessor configuration which can handle the increased computational requirements of demultiplexing larger-capacity FDM structures (60/120)