Procedure for transmission of information on a channel including a system for reception of data signals by sampling using clock signals

Claim 2

A procedure for information transmission utilizing a system for reception of data signals by sampling using clock signals, comprising: calculating a phase shift between said clock signals and said data signals using a formulation of the gradient: t.sub.n+1 =t.sub.n +p/q-αe.sub.n (t.sub.n)e'.sub.n (t.sub.n), wherein: t n+1 : a phase shift signal to be calculated for a rank n+1 data signal; t n : a phase shift signal calculated previously for a rank n data signal; p/q: a ratio of a clock period to a data period; α: a matching coefficient; e n (t n ): an error between a real sample and an assumed theoretical value; and e' n (t n ): a derivative of the value of e n (t n ) with respect to t n , the value e n (t n ) being given by: ##EQU8## wherein: z k : a theoretical value of the data signal; h: the pulse response from a channel and the value e' n (t n ) being given by: ##EQU9## where h' is a derivative of the channel pulse response with respect to time; Te: the period (1/f e ); Ts: the symbol period (1/f s ); and Yn: the sampled signal.

Claim 3

The procedure according to claim 1, wherein: N possible states exist for each channel input configuration, x k , each being connected by a number of paths equal to not more than N states x(i) corresponding to a previous input x k-1 ; calculating the phase t n and a metric term characterizing the path length for each existing state x(j), for each previous state x(i), for each sample y n at the channel output, if the time subscript n corresponds to the path connecting x k-1 to x k for the states considered x(i) and x(j); processing all previous states x(i) and all samples y n to find only one survivor previous state x(i) and storing said survivor in memory, said survivor is chosen to minimize a sum of its previous metric and the metric terms that have just been calculated, the sum forming a new metric for an existing state xj), and also being memorized together with a corresponding phase and a last value of the subscript n; and choosing survivors to select probable paths during the procedure and to eliminate others such that there is only one survivor left providing the decoded data.

Claim 4

The procedure according to claim 2, wherein the metric of the path connecting one state to a previous state is found by performing the following operation: ##EQU10## where p(t n ) values correspond to weights assigned to phases t n ; and storing in memory each estimated phase and each metric for each state, said procedure being performed for each data sample.

Claim 5

The procedure according to claim 3, wherein the calculation procedure is performed for several samples of a variable until the phase t n+1 exceeds a threshold value.

Claim 6

The procedure according to claim 4, wherein the value of the threshold is 0.5.

Claim 7

The procedure according to claim 1, further comprising: applying a maximum likelihood sequence estimation (MLSE) algorithm to a decoding procedure; and performing a phase shift calculation at a same time as a metric calculation in said maximum likelihood sequence estimation (MLSE).