A hybrid breadth/depth first technique to improve convolutionally encoded /spl pi/4-shifted DQPSK over asynchronous CDMA fading channels

Abstract

A direct sequence spread spectrum multiple access communication system based on differential /spl pi/4-shifted QPSK modulation and convolutional coding is analyzed. The channel model is frequency selective fading typical of mobile digital radio systems. An efficient decoding scheme is proposed to relieve performance losses at the receiver due to user interference. Sequential decoding is a valid alternative to Viterbi decoding, and particularly the backtracking algorithm when long constraints convolutional codes are imposed by system performance. Purely breadth first sequential decoding algorithms such as the M-algorithm present inherent parallelism (adapt to a parallel HW realization) but are, as pointed out by Anderson [1992], less efficient in terms of average number of paths searched with respect to the backtracking algorithm, while representing a search performed by the Viterbi algorithm. The authors propose a technique which is a modified version of a Fano-like algorithm in which the forward operations are performed by means of a breadth first algorithm: this modification presents many advantages. First of all, backtracking events are limited. In this way the decoder proceeds faster and with a lower delay. Second: the searches are made in parallel, allowing a smaller buffering and hence a simpler VLSI realization with respect to the traditional Fano decoder. An analysis is performed of the uncoded detection scheme and the results of the simulations of the encoded scheme are compared. Large coding gains seem possible when employing an adapt interleaving and an efficient power control system.<>