Delayed decision-feedback sequence estimation

Abstract

Delayed decision-feedback sequence estimation (DDFSE) is a detection algorithm that prpvides a direct tradeoff between complexity and performance in digital co~munications over intersymbol interference channels. The complexity of the algorithm is controlled by a pqrameter p and can be varied from zero tfl the lpemory of the channel (which can be infinite). The algorithq is based on a trellis with the number of states exponential in p. Whey p = Q, QDFSE requces to the decision-feedback detector. When the qemory of the channel is finite, PDFSE with maximal complexity i w equivaleat to the Viterbi slgoritbm. Of course, if the channel has infinite mepory, the Viterbi algorithm cannot be implemented. Far the intermediate values of p, the algorithm can be described as a reduced-state Viterbi algorithm with feedback incorporated into the structure of path metric computations.We first consider DDFSE for uncoded PAM signals. Estimates on the performance of the algorithm are given, and simulation results are provided for several examples. A more general form of DDFSE applicable to coded modulation systems is also presented. As an example, detection of trellis coded QPSK signals over intersymbol interference channels is discussed., m f -, , xi-into u, = ( m f -, , mj-,) and u, = x,-I . The dranch metric is ( y , -t, 1 where t, = x, + G,-,, and from (10) and (1 I), the partial state estimator is given by v^(u,, m ! , m f ) = x,(x, = x ( m ! , m2, u,) J J J is determined by (36), and w,- = -pu(u,)). Fig. 10 shows results of computer simulations for this channel ( p = 1/2). It gives error rates of the DDFSE and the

Sensitivity of the mean-square DDFSD to a noisy estimate of the noise variance

“…For channels with large delay spread, the optimum equalization algorithm, that is, the maximum likelihood sequence detector (MLSD) [1], is often too complex. For practical implementation, a suboptimum that allows a favorable trade off between complexity and performance is the delayed decision feedback sequence detector (DDFSD) [2].…”

“…In [2] the noise whitening prefilter of [1] was considered before the DDFSD (W-DDFSD). In [3] has been shown that performance, especially for highly frequency selective channels, can be improved by adopting the mean-square prefilter (MS-DDFSD).…”

Equalization of Time-Varying Channels

“…Two of the more well known techniques are reduced state sequence estimation [EQ88] and delayed decision feedback estimation [DHH89]. Tree-search algorithms can be partitioned into optimal and suboptimal approaches.…”

Maximum likelihood sequence estimation from the lattice viewpoint

“…Duel-Hallen and Heegrad [17] developed the method of the delayed decisionfeedback sequence estimation (DDFSE) . Their main ideas are to define the states of trellis by considering the first few terms of ISI, and to include an estimate of the tail of ISI in the definition of branch metrics.…”