Digital Modem Having a Monitor for Signal-to-Noise Ratio

In view of the above considerations the objective of the present invention is to provide a device for direct monitoring of the signal-to-noise ratio in a digital modem by using some of the standard components of the modem but without interfering with the modem's normal operation.

Another objective is to provide a device which uses relatively simple, easily available components.

Other objectives and advantages of the invention shall become clear during the course of the preferred description of the invention.

The modem according to this invention is adapted to receive QAM-type signals from a digital communication network, said signals being represented by a signal constellation, the modem comprising means for determining the ideal signals corresponding to the received signals, and means for generating noise signals, said noise signals being the difference between said ideal and said received signals, means for generating the rms value of said noise signals and means for determining the signal-to-noise ratio. Since for signals located relatively far from the origin of said signal constellation the difference between the received and the ideal signals is due to noise, as well as phase and amplitude errors induced by the communication network, the monitoring means is adapted to respond only to the innermost signals, i.e. signals located relatively close to the origin, where the effects of said phase and amplitude errors are negligible.

FIG. 1 shows the first quadrant of the signal constellation associated with a typical 64 point QAM-type modem;

FIG. 2 is an enlarged portion of FIG. 1 showing the innermost point;

FIG. 3 shows the subject device incorporated in a phase correction loop of a digital modem; and

FIG. 4 shows a digital low pass filter used in the device.

The invention shall now be described in reference to a 14400 b/s 64-point QAM-type digital modem, it being understood that the device may be easily adapted to other digital modems by one skilled in the art. The typical constellation of the signals by the modem of the type described above over a standard communications line is shown in FIG. 1. Actually only the first quadrant is shown since the other three quadrants would obviously look similar. The ideal points for each possible signal have been marked by an X and the closed oval loops drawn around each X delineates the signal space of a typical signal received by a modem of the type described above for that particular ideal signal X. These loops are due to phase jitter, frequency jitter, amplitude modulation and noise. It should be noted that the area of the loop around each point X increases as the radial component of X, i.e. its distance increases from the origin 0. Furthermore the eccentricity of these loops also increases with the radial component due to the increased effect of the phase jitter. On the other hand points near the origin, and especially the innermost points such as point A are surrounded by almost circular shapes. This indicates that the errors in the received signals corresponding to the innermost points are predominantly due to noise. Therefore, a signal-to-noise determination for the digital modem may be accomplished by analyzing only the signals corresponding to the innermost points.

In other words if a signal Yo is received corresponding to an ideal signal Xo, where Yo is an innermost signal A, the error signals other than noise N may be ignored so that signals Xo, Yo and N are related by

One great advantage in modems using QAM or PSK-type modulation is that Xo and therefore its rms value is known. If, for example when a Yo is received which is in signal space A, it is inherently obvious that the ideal signal Xo has been sent. Therefore in a QAM or PSK modem S/N may be readily calculated. Generally the signal-to-noise ratio is defined as

where x is the rms value of the ideal signal X and n is the rms value of the noise N.

A further approximation may be made by assuming that for small variations of N with respect to Y, the rms value n of N taken over Y samples is ##EQU1##

One way of implementing the signal-to-noise ratio monitor is shown in FIG. 3. In this figure incoming signals are fed into an equalizer 10 which feeds the equalized signals into a coordinate rotation circuit 20. Circuit 20 rotates the signals by an angle .phi. provided by phase error detection circuit 30 which is designed to eliminate phase errors. Details of the error detection circuit are disclosed in the commonly assigned Patent Application Ser. No. 407,451 filed on Aug. 12, 1982 which is hereby incorporated by reference.

Briefly, the rotated signals Y are also fed into a determinator circuit 40 which determines the ideal signals X corresponding to said signals Y. The determinator generates a complex constant for each signal Y and sends it to the phase error detection circuit 30. The determinator circuit also outputs demodulated data in accordance with the ideal signals X.

At the same time equalizer error calculation circuit 50 calculates the difference D between the incoming signals Y and the ideal signals X. As it was pointed out earlier, for the innermost points, X=Xo, this difference is effectively a noise signal, i.e. D=N. The difference D is fed into an error rotation circuit 60 which is provided to eliminate the effect of the coordinate rotation circuit 20, by rotating the difference D by an angle .phi.. The output of circuit 60 is fed into equalizer tap update circuit 70 and is in phase with the output of equalizer 10. The equalizer tap update circuit 70 is used to adjust the taps of equalizer 10.

The signal-to-noise ratio monitor circuit 80 comprises an absolute value circuit 90 and a low-pass filter 100. The absolute value circuit 90 generates the absolute value of the output D of circuit 60. The low pass filter 100 is a standard digital filter with the components shown in FIG. 4.

The input of the filter 100 is connected to a first adder 110 and from there the signal is passed through a first multiplier 120, a second adder 130, and an integrator 140. The output of the integrator provides the output of the filter 100, and is also fed back to second adder 130 and via second multiplier 150 to first adder 110. The multipliers multiply their respective inputs by scaling factors. These factors are usually powers of 2 so that the multiplication can be easily accomplished by mere binary shifting of the inputs. The integrator 140 is normally a digital memory device and it is operational only when it gets an ENABLE signal. This low-pass filter 100 is usually referred to as a leaky integrator and its output does not overflow, nor does it return to zero with a zero input. Thus the filter may be used to obtain an output which is the average of the input signals taken over a period of time which is equal to the time constant of the filter.

The net effect of combining the absolute value circuit 90 with low-pass filter 100 is to produce an output which is the average of the absolute value of the input. The ENABLE line of the low-pass filter 100 is connected to the determinator 40 which activates it only when an incoming signal is Yo corresponding to Xo, i.e, the filter 100 is active only when D=N. Therefore the output n of circuit 80 is ##EQU2## which is approximately the desired rms value of the noise. This value is then fed into divider 160 which also receives an input x from determinator 40 to generate the signal-to-noise ratio X/N. One skilled in the art will appreciate the fact that the monitor presented herein performs its function without interfering in any way with the normal functioning of the modem.

The circuit presented herein accomplishes all the desired objectives of the invention. Other configurations may be made without departing from the scope of the invention as defined in the appended claims.