Method for the equalization of dispersive, linear or approximately linear channels for the transmission of digital signals and arrangement for carrying out the method

Claim 1

A method for the equalization of dispersive, linear or approximately linear channels, particularly radio channels, for transmitting digital signals by means of an matched filter and a following equalizer, comprising the steps of: providing a matched filter (1) which is matched to the entire transmission channel; filtering, in the matched filter (1), a received signal r(t) characterized by a data sequence a k to produce a time-discrete signal sequence mfo k at the output of the matched filter (1), which automatically determines the optimum sampling times for the time-discrete signal sequence mfo k ; and equalizing the time-discrete signal sequence mfo k in sections in the equalizer (2) by means of an iteration procedure with relaxation, with said procedure comprising the steps of: subdividing the signal sequence mfo k into shorter mutually overlapping sections and storing the sections individually in a first and second store (22, 23) of the equalizer (2); equalizing the shorter mutually overlapping sections of signal sequence mfo k stored in each case in an equalization filter (25) of the equalizer (2); feeding the data stored in the second store (23) to the input of a first shift register (251) which is connected in series with a second shift register (252); feeding, in each clock interval of one iteration cycle, the data stored in the individual register locations (2511-2514; 2521-2523) of the first and second shift registers (251; 252) via weighting sections (2541-1547) of a weighting stage (254) to a summation section (253), where they are summed together; feeding back the data δβ k summed together at the output of the summation section (253) with a time delay and additional weighting (relaxation) to the input of the second shift register (252); generating in the equalization filter (25), during each iteration cycle, new estimated values β k of the data sequence a k which are in each case output at the output of the second shift register (252) and are subsequently stored in the second store (23) and during this process overwrite the data hitherto stored; summing together in each iteration cycle the square of the magnitude formed from the processed data δβ k at the output of the summation section (253) and the individual magnitude squares and using the sum signal s to determine the relaxation factor α for the processed data δβ k fed back and as criterion for the termination of the iteration process; and after termination of the iteration process, outputting the estimated values β k , last stored in the second store (23), of the data sequence a k at the output (8) of the equalizer (2) and subsequently storing the next section of the output signal sequence mfo k of the matched filter (1) in the two stores (22, 23).

Claim 2

A method for the equalization of dispersive, linear or approximately linear channels, particularly radio channels, for transmitting digital signals by means of an matched filter and a following equalizer, comprising the steps of: providing a matched filter (1) which is matched to the entire transmission channel; filtering, in the matched filter (1), a received signal r(t) characterized by a data sequence a k to produce a time-discrete signal sequence mfo k at the output of the matched filter (1), which automatically determines the optimum sampling times for the time-discrete signal sequence mfo k ; and equalizing the time-discrete signal sequence mfo k in sections in the equalizer (2) by means of an iteration procedure with relaxation, with said procedure comprising the steps of: subdividing the signal sequence mfo k into shorter mutually overlapping sections and storing the sections individually in a first and second store (22, 23) of the equalizer (2); equalizing the shorter mutually overlapping sections of signal sequence mfo k stored in each case in an equalization filter (25) of the equalizer (2); feeding the data stored in the second store (23) to the input of a first shift register (251) which is connected in series with a second shift register (252); feeding, in each clock interval of one iteration cycle, the data stored in the individual register locations (2511-2514; 2521-2523) of the first and second shift registers (251; 252) via weighting sections (2541-1547) of a weighting stage (254) to a summation section (253), where they are summed together; feeding back the data δβ k summed together at the output of the summation section (253) with a time delay and additional weighting (relaxation) to the input of the second shift register (252); generating in the equalization filter (25), during each iteration cycle, new estimated values β k of the data sequence a k which are in each case output at the output of the second shift register (252) and are subsequently stored in the second store (23) and during this process overwrite the data hitherto stored; summing together in each iteration cycle the square of the magnitude formed from the processed data δβ k at the output of the summation section (253) and the individual magnitude squares and using the sum signal s to determine the relaxation factor α for the processed data δβ k fed back and as criterion for the termination of the iteration process; and after termination of the iteration process, outputting the estimated values β k , last stored in the second store (23), of the data sequence a k at the output (8) of the equalizer (2) and subsequently storing the next section of the output signal sequence mfo k of the matched filter (1) in the two stores (22, 23).