Telecommunication systems such as PON-FITL-Systems have a TDMA frame structure comprising a main station and a number of substations. In order to achieve accurate fine ranging, an A/D-converter for acquiring samples from a ranging pulse transmitted from the substation is operated at a frequency which is high with respect to the master or bit clock, i.e. real time oversampling occurs. Such high speed A/D-converters are expensive, consume much power and have limited resolution. Furthermore, for systems with even higher bit rates the known oversampling becomes cumbersome. A quasi-oversampled coarse and/or fine ranging is proposed operating at relatively low sampling rates. Thus the substations send a repetitive ranging-bit-pattern which is sampled in the main station and thereafter reconstructed in a processing apparatus to a quasi-oversampled ranging-bit-pattern. In order to find the ranging error to be sent to the relevant substation, the quasi-oversampled ranging-bit-pattern is matched against a reference pattern, and the ranging error is found from the best match.