DE1931698C - Transmission system with pulse code modulation - Google Patents

Description

E (x) = ^

has.

This means

χ / T

/ T = normalized frequency, variable frequency. Clock period of the code elements.

S | v _e M = power density of the specified PCM useful signal at the input of the route section,

S _n Jx) = required L ^ istun ^ density of the PCM useful signal ai.i output of the route section taking into account the equalizer included in the route section.

Cable attenuation factor, crosstalk factor of near-end crosstalk. Number of interfering PCM signals.

A (x)

A _n (X)

η =

The invention relates to a transmission system with pulse code modulation, in which the in both transmission directions utilized, at least two undesirably interlinking transmission lines containing transmission path is divided into preferably several route sections, which over with Equalizers provided regenerators are connected and where appropriate in the individual route section the type of signal and / or signal form by the equalizer assigned to the individual regenerator will be changed.

When transmitting PCM signals on lines, the length of a route section is through system-related transmission properties are limited. The pulses sent are frequency-dependent Attenuation and phase delay of the channel are distorted, leading to overlapping of neighboring pulses can lead. So it is to be ensured at certain intervals that the signal returns to his original shape is brought. This is achieved by means of regenerators and associated equalizers, which, as stated in "Bell System Technical Journal", 1966, p. 1007, at the beginning and / or on Arranges the end of a route section. An equalization «has in the course of the section itself the insufficient near-end crosstalk vapor proved to be unfavorable. Systems known to date for the transmission of PCM signals via cables * are related to the signal-to-noise ratio of the transmitted Message not yet optimally interpreted. Especially with the multiple use of local lines, where the PCM signals in both "directions on several lines that are contained in one cable: ± over-

there will be interference from near-end crosstalk to a particularly high degree. Line \ ersizeunü the existing signal-to-noise ratio between message and disturbance is therefore of great importance here To use.

ι ς The invention is based on the object of specifying a system for the transmission of PCM signals on lines, which allows the sum of the inaccurate line equalization and a resulting from the near-end crosstalk.

to keep it as small as possible in a simple way.

This object is achieved according to the invention in that the equalizer assigned to the individual route section to eliminate the near-end interference caused by the undesired coupling

chens a transfer factor £ (x) according to the equation

SsJx) A (X)

has.

In it means

x = / T = normalized frequency, / = variable frequency, T = clock period of the code elements, S _n Jx) = power density of the specified PCM useful signal at the input of the Strecker section,

SnoM = required power density of the PCM useful signal at the output of the route section, taking into account the equalizers included in the route section, A (x) = attenuation factor of the cable, A _N (x) - crosstalk factor of the near-end crosstalk. η = number of interfering PCM signals.

The invention is based on the consideration that in the equation for the transmission factor of the route section the quotient

into a deformation factor

and into a species change factor

[T (r \ ~ l

can be split. The change in shape refers to the change in the shape of the pulse Understood on the entrance and exit of the route section, for example, the deformation of a at the entrance of the route section existing square pulse into a pulse with approximately sinusoidal Course. The species change is

if the type of impulse differs between the input and output of the section of the route, for example a binary impulse signal is sent at the input and a quasitemary impulse signal for further off 'triune' is present at the output. Fun such a transmission system Aare then, for example, a so-called biternary in the sense of a publication in the NTZ! 965. jj _e f [; ν 141 to! 44 \ used expressions v.-ji-.e. There are thus four basic possibilities for the transmission of the characteristic signal features

ρ ·> Sisinal becomes as unec-u'irt and uiv.cr- .; transmitted i.-rt
ι; _: . means

Γ -S \, i-vi

1 .

S.,

2. The signal is in its form, but not in its type, changed in the course of the transfer; for example, a binary signal with rectangular signal form in the course of transmission into a similar with sinusoidal Signal shape can be transformed. Here, too, the digital characteristics are not changed. Pas means

. IXl

The type of signal and thus its digital characteristics are changed; For example, in the course of transmission, a binary signal becomes a pseudo-ternary or a biternary one. A pseudoternary signal, which is also variously referred to as a bipolar signal, is understood to be one in which a pseudoternary zero emerges from a binary zero, while plus one and minus one are generated alternately from the binary one. A biternary signal, which is also variously referred to as a suobinary signal, is understood to mean one in which a biternary zero results from a binary zero, while the binary one is followed by a one with a special sign, the change of which is then and only triggered if there is an odd number of consecutive zeros in between.
That means

To make this clear, will. '. next to the Optimization for case 1 and then treated for cases 2 through 4. The following definitions are used uses Is. here / u fig. 11th

.v - ■ / T aul normalized the Bi'.following period

Frequency.

.S \ _ivi - power density of the IK M-Nut / signal ... in Cable entry 1.

.S \ ,. = 2 j .s \ _t . (. Vid.x total performance of the PC M-Si

enals on the cable menu I.
.-li.x) = attenuation factor of the cable.

L - cable length / wipe the amplifier point

ten I and 11.

/:!.x) = the transfer factor reads Lnt / errers. -S _K , .I \) = power density of a disruptive IVM-

Signals, at cable input 11. η = number of interfering PCM signals.

A _s {x) = cross-talk factor of near-end crosstalk. S _Va (x) = power density of the PCM useful signal after the equalizer,

S _NA = 2 \ S _s Jx) dx total power of the PCM

(I.

Useful signal after the equalizer. SrJx) = power density of the interfering PCM signals after the equalizer.

S _R <= 2 S _Ra (x) dx total power of all disturbances

signals after the equalizer, Sy _A - distortion power (mean square error of the useful signal after the equalizer).

S RYA ⁼ total interference power.
1, 2, 3 ... m = wire pairs or lines in the cable. (All factors relate to the power, all powers to the amplitude 1.)

Optimization while maintaining the type and shape of the signal

According to the possibility mentioned under point 1, S _Vo ^ Sse should be, where S _RA + S _VA = S _RVA should be as small as possible. With the above definitions, the useful power

30th

35

40

S _S

= 2Js _Se

(x) -A (x) ■ £ (.x) dx

4. Combination of 2 and 3; the signal is in

Course of the transmission changed both in its type _{and di (} , _ve n: achievement as well as in its form.

That means

1:

Sy _A

2 js _Nr {x) · (1 -

dbc. (2)

If 'Jf

It is therefore possible to use a factor, provided that η is sufficiently large

If the transfer factor E (x) of the equalizer is to be reduced, the powers of the individual interferers add up;

split, whereby sicii becomes a much more advantageous 65,

Dimensioning of the system, in particular also ", f"

with regard to the resulting equalizer, er * ra- - ■ "J * ReW ' A _s (x) ■ b (x) üx (JJ

can be enough. 0

The power density of the signals is the same in both directions; but there between interfering signal and Useful signal there is no correlation

> RVA -

[n · A _N {x) ■ E (X) + (1 -

dx. (4)

The transfer factor of the equalizer E (x) should now be chosen so that S _KVA becomes a minimum. That is true for

Em - - ^{A {x} l -, (51

^{£ (X)} - [AiX) + nA _n (X)? ^l)

Equation (S) shows that the optimal equalizer is independent of the power density of the signal, but only dependent on the cable data and the number of the interfering signals. The entire disturbance performance is hereby

pulled equalizer recovered signal S _{Na is} determined.

It can be shown that it does not matter whether the The equalizer is placed at the cable entry or exit or distributed on both sides; under no circumstances may however, it can be arranged in the course of the cable. In particular, you can split the equalizer according to equation (8) into the two Quotient given factors and the effect of the

■ ο put the first factor at the cable entry. However, this in turn amounts to the case that S _n i was used to send from the start.

The invention is explained in more detail below with reference to diagrams relating to an exemplary embodiment explained.

The transmission of POM signals of a time division multiplex system is used as an application example 32 telephone channels on a 0.6 mm local line Diameter examined. The frequency-dependent Factors A (x) and A _N (x) in the transfer factor E (x) were based on the following formulas:

S * _VA - 2Y S _Nr (x) am ₊ ": A _n (X) ^dx '

the total useful power

(6)

E (X) =

S _{% m} ix)

A {x) [A (X) + η

18)

which means that equations (6) and (7) apply here for interference and useful power with the restriction that Sj, _{m is to be used} instead of S _n . If one proceeds from the assumption S " _f % S", one can basically say that if an optimal equalizer is narrowed, the signal-to-noise ratio nkbt depends on the type and shape of the signal sent from the signal S _He . but rather the one at the exit of the route section, taking into account the A (X)

(L = Län

25th <W \ 'x ♦ IJix)

in km).

It show in FIG. 2 curves 1 to 4 are a logarithmic representation of the transfer function E (x) of the equalizer for the following cases:

If _{η ■ A N} (x) <A (x \, then the quotient in equation (6) tends to zero, the quotient in equation (7) tends to 1. If, on the other hand, η - A "(x)> A (x \ , the quotient in equation (6) tends to unity and the quotient in equation (7) tends to 0. Although the optimal equalizer is independent of the power density of the signal, S _KVA becomes smaller and S _SA larger, the more the more the power density S _Sr (x) is concentrated in frequency ranges in which η-A ^ x) ■ c A (x) In practice this means choosing the type and shape of the signal so that the power density increases rapidly with increasing frequency subsides.

Optimization when the type of signal changes and / or shape

According to the possibilities of changing the type and / or shape of the signal, S _Na S _n , with the restriction that S _Sa in turn results in a signal which contains characteristic features of the signal.

The optimization requirement is now fulfilled by

30th

Cure ve

Signa lan

constant constant Binary to biternary Binary to pseudo-era

Waveform

constant Rectangle in sine Rectangle in sine Rectangle in sine

In Fig. 2, curve 1 shows the frequency response of the equalization function for the example L = 2.4 km and π = 40 with the same type and shape of the signal. If the rectification is also to cause the transition from the rectangular to the sinusoidal shape, it should be designed according to curve 2. Here the maximum and thus the cut-off frequency is a little lower. Curve 3 shows the transition from the binary to the biternary signal. Here the first zero of the equalizer is already at χ = 0.5. The transition from bi

Similar to the pscudoterar signal is shown with curve 4. Here skh results in a NuOsteOe for χ = 0, dtL, the low-frequency component of the binary signal is completely attenuated. This possibility offers the advantage of transmitting with a relatively high

_s5 lower limit frequency to be able to use. In addition, all disturbances falling in this frequency range (e.g. dialing noises) are kept away from the inputs of the regenerators. The saving in amplification compared to the case that signal type and

te -form is retained, it is also I2db With regard to the circuit design of the equalizer, it should be said that the usual, e.g. B. also in the literature mentioned at the outset are suitable and these by

6 $ special dimensioning according to the required Transmission properties can be adjusted

For this purpose i sheet of drawings

Claims (1)

I 931 698 Claim: Transmission system with pulse code modulation, in which the transmission path used in both fbertra-iunii ^ nchiunscn and containing at least two undesirably coupling transmission lines is divided into preferably several route sections. which are linked via regenerators provided with equalizers and in which, if necessary, the type of signal and / or signal form in the individual route section is changed by the equalizer assigned to the individual regenerator, characterized in that the equalizer assigned to the individual route section eliminates the coupling caused by the undesired coupling Near-end crosstalk an overhaul factor F; \ ι according to the equation