Abstract:

There is a great deal of issues in the traditional symbol detection algorithm such as high computational com-plexity and large engineering implementation difficulty. Considering these problems, an improved algorithm based on decision feedback was proposed. To reduce complexity, the approach in which ruling element information was employed to filter waveforms during length of observing bits. Besides, the theoretical performance was approved by a composition of upper bound and lower bound. Specifically, the form is a union bound which consists of being tight at high SNR and being tight at low SNR, the latter can be obtained by grid minimum Euclidean distance. Theoretical analysis and si-mulation results show that compared with the traditional algorithm, the improved algorithm could keep no loss of detec-tion performance, but the complexity is reduced at negative exponent power way with the increase of number of feedback element.

Key words: CPFSK, noncoherent multiple symbol detection, decision feedback, average matched filter