Letter: Application of a simplified fano metric to adaptive receivers for digital mobile radio systems

Abstract

The detection list of sequential detection algorithms like the Generalized‐Stack‐Algorithm normally includes paths with different search lengths. Therefore, a special metric, the socalled Fano metric, has to be used to allow a comparison between the path metrics. A drawback of the Fano metric is its great computational effort in the case of a time‐variant channel impulse response which occurs typically at data transmission over frequency‐selective mobile radio channels. In this paper the most expensive part of the metric is replaced by a bias depending on the instantaneous signal‐to‐noise ratio and the phase characteristic of the channel. Simulation results obtained for the GSM system show the good performance of this method in view of bit error rate and computational complexity relative to the optimum Viterbi‐Algorithm.

Soft-decision output at sequential detection algorithms in digital mobile radio systems

“…After the interleaving the transmission is carried out in bursts of 148 bits (114 information bits and 34 additional tail and midamble bits for synchronisation and channel impulse response estimation) with a rate of about 271 kbit/s [1, 5,9]. The adopted GMSK modulation scheme can be well approximated by Offset-QPSK with the trans- respectively.…”

“…The adopted GMSK modulation scheme can be well approximated by Offset-QPSK with the trans- respectively. Therefore, the size of the symbol constellation is given by Q = 2 (binary modulation) [5,9]. The modulator output signal g(t) is affected on the mobile radio channel by time-and frequency-selective fading.…”

“…The low complexity of SDA is based upon the fact that especially at high SNR only a small subset of all possible symbol sequences has to be controlled for optimum detection. Investigations have shown that SDA like the M-or the Stack-Algorithm (MA, SA) are applicable to the GSM system without any performance degradation against the VA [5].…”

“…The adaptation of the WMF which can be implemented by a modified Kalman filter[5] is done in such a manner that {a,,} is always minimum-phase, i.e. the energy is concentrated in the first samples.Each possible transmitted symbol sequence {$)} corresponds to a sequence of channel states (nodes) {gp)} with and can be represented by a path in the tree or trellis diagram [2-41.…”