We show how the well known least square (LS) algorithm can be used for combined channel estimation and frequency correction in packet oriented mobile communication systems. The method is derived by investigation of the effect of a frequency offset, e.g., due to the instability of the local oscillators, on the LS estimates. The algorithm corrects frequency offsets resulting in phase shifts per symbol that are in the range of several degrees. Imperfect correlation properties of the applied synchronisation sequence are automatically compensated by the algorithm.

Proposes a modification of generalized concatenated codes, which allows to construct some of the best known binary codes in a simple way. Furthermore, a large class of optimal linear unequal error protection codes (LUEP codes) can easily be generated. All constructed codes can be efficiently decoded by the Blokh-Zyablov-Zinov'ev algorithm if an appropriate metric is used.>

We propose an algorithm for bounded minimum distance decoding of (partial) unit memory codes up to half the "designed" extended row distance. It makes use of a reduced trellis with the nodes found by bounded minimum distance decoding of the block codes used in the unit memory code. The results can be extended to general multimemory codes. The complexity of this algorithm is upper bounded by 2(d~/sub 1sup r/-2d/sub /spl alpha) times the complexity of the bounded minimum distance decoder of the block codes in the unit memory code. Here d/sub /spl alpha is the linear increase of the designed extended row distance d~/sub isup r/.>