Applications of momentary Fourier transform in frequency shift keying signal demodulation

Power line communication is attracting increasing attention for its characteristics of easy availability and low cost. Frequency shift keying with the virtue of antijamming capability is suitable for the low signal‐to‐noise ratio power line environment. The principle of the frequency shift keying communication system based on momentary Fourier transform (MFT) algorithm is provided first in this paper. Theoretically, the twiddle factor in the MFT algorithm is a decimal with infinite length. However, when it is applied on digital signal processor (DSP), the registers with finite length in DSP lead to the inaccuracy of the twiddle factor. Thus, the MFT algorithm with recursive form will cause accumulative error as a result. We analyze the impact of the inaccurate twiddle factor on the MFT algorithm from a novel point of view, by which the MFT algorithm is considered as a linear time invariant system. We prove that the accumulative error would not disperse when the modulus of the twiddle factor is smaller than 1. Simulation results show that the truncation approach is more effective than both carrying and rounding methods in a practical system. In addition, the number of significant digits of truncated twiddle factor must be chosen larger than 4 to overcome the problem of accumulative error.Copyright © 2012 John Wiley & Sons, Ltd.