Signal transmission method

A signal transmission method according to the principle of quadrature/amplitude modulation, in which digital data occurring at a time interval T are formed into a transmit signal with the aid of pulse shapers (3a, 3b) having a given unit pulse response r(t) in a transmitter (4), the transmit signal is transmitted by means of a carrier oscillation via a channel (5) having a channel unit pulse response h(t), when an additive white noise w(t) is superimposed on it, so that a receive signal y(t) is present in a receiver (10), and in which the receive signal y(t) is preprocessed by channel-matched filters (7 a, 7b) in the receiver (10), has channel-matched filters (7a, 7b) which are matched to a filter cascade H(t) formed by pulse shapers (3a, 3b) and channel (5) as a whole. The channel-matched filters f(7a, 7b) are preferably transverse filters the data points of which are at a time interval .DELTA.T which is small in relation to the time interval T of the digital data. An estimation of the channel unit pulse response h(t) is determined in that the unit pulse response of the filter cascade H(t) is convoluted with a filter r.sup.-1 (t) which is inverse to the pulse shaper.

What is claimed as new and desired to be secured by Letters Patent of the United States is:

1. Signal transmission method according to the principle of quadrature amplitude modulation, comprising the steps of:

in a transmitter

(a) sampling digital data in predetermined time intervals T;

(b) forming a transmit signal by feeding the digital data to a pulse shaper, said pulse shaper having a given unit pulse response r(t);

(c) amplitude-modulating a carrier oscillation by the transmit signal;

(d) transmitting said amplitude-modulated carrier oscillation via a channel, said channel having a unit pulse response h(t) and superimposing an additive white noise w(t);

in a receiver

(e) demodulating the transmitted carrier oscillation into a receive signal y(t);

(f) sampling said receive signal in time intervals .DELTA.T which are small in relation to the time interval T;

(g) preprocessing said sampled receive signal in a transverse channel-matched filter, said channel-matched filter being matched to a filter cascade H(t) formed by said pulse shaper and said channel as a whole,

H(t)=h*r(t),

where "*" stands for the convolution product; and

(h) determining original digital data from the preprocessed sampled receive signal in an analysis circuit.

2. Signal transmission method according to the principle of quadrature amplitude modulation, comprising the steps of:

in a transmitter

(a) sampling digital data in predetermined time intervals T, and based thereon generating by that data coefficients c.sub.k ;

(b) forming a transmit signal by feeding the data coefficients to a pulse shaper, said pulse shaper having a given unit pulse response r(t);

(c) amplitude-modulating a carrier oscillation by the transmit signal;

(d) transmitting said amplitude-modulated carrier oscillation via a channel, said channel having a unit pulse response h(t) and superimposing an additive white noise w(t);

in a receiver

(e) demodulating the transmitted carrier oscillation into a receive signal y(t);

(f) sampling said receive signal in time intervals .DELTA.T which are small in relation to the time interval T, and based thereon generating a sampled transmit signal y.sub.k ;

(g) preprocessing said sampled receive signal y.sub.k in a transverse channel-matched filter, said channel-matched filter being matched to a filter cascade H(t) formed by said pulse shaper and said channel as a whole, H(t)=h*r(t), where * stands for the convolution product, having coefficients H(i), i=0, . . . , L.sub.c, which are determined in accordance with the principle of least error squares between the transmit signal and the receive signal, fulfilling the equations ##EQU6## where c.sub.k =data coefficient,

y.sub.k =sampled transmit signal,

y.sub.k =complex conjugate of y.sub.k, and

c.sub.k-i =complex conjugate of c.sub.k-i ; and

(h) determining original data coefficients c.sub.k from the preprocessed sampled receive signal in an analysis circuit.

3. Signal transmission method as claimed in claim 2, using data coefficients for forming the transmit signal and for preprocessing the sampled receive signal, which coefficients correspond to a predetermined sequence of digital data known to the transmitter and the receiver.

4. Signal transmission method as claimed in claim 2, wherein the filter cascade used for preprocessing is periodically estimated.

5. Signal transmission method according to the principle of quadrature amplitude modulation, comprising the steps of:

in a transmitter

(a) sampling digital data in predetermined time intervals T, and based thereon generating data coefficients c.sub.k ;

(b) forming a transmit signal by feeding the data coefficients to a pulse shaper, said pulse shaper having a given unit pulse response r(t);

(c) amplitude-modulating a carrier oscillation by the transmit signal;

(d) transmitting said amplitude-modulated carrier oscillation via a channel, said channel having a unit pulse response h(t) and superimposing an additive white noise w(t);

in a receiver

(e) demodulating the transmitted carrier oscillation into a receive signal y(t);

(f) sampling said receive signal in time intervals .DELTA.T which are small in relation to the time interval T, generating by that a sampled transmit signal y.sub.k ;

(g) feeding said sampled receive signal into a channel estimating circuit controlling a transverse channel-matched filter by calculating coefficients H(i), i=0, . . . , Lc, which are determined in accordance with the principle of least error squarres between the transmit signal and the receive signal, said coefficients H(i) fulfilling the equations, ##EQU7## where c.sub.k =data coefficient,

y.sub.k =sampled transmit signal,

y.sub.k =complex conjugate of y.sub.k, and

c.sub.k-i =complex conjugate of c.sub.k-i ;

(h) preprocessing said sampled receive signal in the transverse channel-matched filter, said channel-matched filter being matched to a filter cascade H(t) formed by said pulse shaper and said channel as a whole,

H(t)=h*r(t), where * stands for the convolution product,

and using said coefficients H(i), i=0, . . . , Lc;

(i) determining original data coefficients c.sub.k from the preprocessed sampled receive signal in an analysis circuit; and

(k) feeding back the determined data coefficients c.sub.k to the channel estimating circuit.

6. Signal transmission method as claimed in claim 5, comprising the step of determining the channel unit pulse response h(t) by convoluting an estimation of the filter cascade H(t) with a unit pulse response r.sup.-1 (t) which is inverse to the unit pulse response r(t) of the pulse shaper.