SU1540008A2 - Device for separation of transmission direction in duplex system - Google Patents

Abstract

The invention relates to the field of telecommunications, may find application in the transmission of data over communication channels. The purpose of the invention is to improve separation accuracy by suppressing echoes. The device contains an input unit 1, a switch 2, two DACs 3, 11, ADC 4, a training sequence shaper 5, memory blocks 6, 10, r-7, a subtractor 8, an adder 9, an automatic level adjustment block 12 is entered into the device, a comparator 13, a nonlinear signal processing unit 14, an interpolator 15, a D / A converter 16, a key 17 and a rectifier 18. The echo suppression is performed in the input units using nonlinear processing, and by interpolation to restore the form of the received signal. 1 hp f-ly, 1 ill.

Description

The invention relates to telecommunications and can be applied when transmitting data over communication channels, ile.iT; inventions - increasing the accuracy of separation by suppressing echo signals.

The drawing shows a structural electric circuit of the device.

The device contains an input unit -ig I, switch 2, the first digital-to-analog converter (PAP) 3, analog-to-digital converter (ADC) 4,. “Training sequence generator 5, the first memory block 6., generator 15, subtractor 8, adder 9, second memory unit 10, second NLP ί j,. 'Needle 1 ζ of automatic level control, comparator 13, non-linear signal processing unit 14, interpolator 15, third PAP 16, key 17, rectifier 18, while non-linear processing unit 14 consists of a logarithm 19 and a multiplier 20, and the interpolator 15 - from delay line 21, adder 25, and divider 23.

The device operates as follows.

The operation of the device can be divided into the process of preliminary adaptation of the iodine parameters of the communication channel and the process of duplex transmission of information, ’The process of preliminary adaptation is as follows.

From the control output of the data terminal equipment at the moment of training, 0. This signal connects the output of the training sequence generator 5 through the switch 2 to the input of the first DAC 3, and also forcibly resets the second memory unit 10. The clock pulses from the output of the generator 7 are continuously fed to the input of the generator 5 learning sequence, which under their action continuously changes its state, forming binary combinations ϋ, (Kit), These binary combinations are fed to the input of the first DAC 3 and converted to tension U; (t), / 'the tax voltage Uj (t) goes to the opposite station and at the same time the ADC input 4 “The training signal I'i (t) is converted (collapsed) in accordance with the input parameters of the two-wire communication channel. Therefore, at the input of the ADC h there is a signal, the value of which can be estimated by the following expression:

where x ('t)

Y (C) ¹ Xt) signal at the input of the ADC 4; impulse response of near echo (input parameters of the communication channel);

communication channel pum.

At the output of the ADC 4, we observe the same signal as at its output, only in binary form. Binary combinations with the output of the ADC 4 are fed to the input of the first block 6 of the memory. With this first binary combination and the output of the shaper 5 of the training sequence, Щ (Κίκΐ;) corresponds to l ’« (t) on s ap and Values U, 'К. ftt)

that is, which are the tamper switch of the communication channel on the connection with the user ;; If you want to use the worms and ₃ worms are detected '' block 6 of the network according to expression (2), by the value with the address Ut (K _f t), the response UD (Ked) flies in Yes , II at the collection of all types of other ombinapils A o m and r o v a l e t o b o o <; ·, The 5th sequence controls the exit of the intramural trench · about- / up to 5m:;!) The data comes in S ' ¹ , which indicates the termination of the learning process, Thus, after training the device under the channel parameters are asleep and the nerve block about the memory stores the responses of the communication channel to the corresponding digital combinations, and the second block 10 of the memory is completely reset.

Let us consider how duplex transmission of information and compensation of the signals of its transmitter at the input of the receiver are made. After training, the signal from the output of the data terminal equipment (Г’.1 switches the switch 2 so that the output of the input unit I is connected to the input of the first DAC 3, if the learning driver 5 moves off, although the latter changes its link state continuously

After preliminary training, duplex transmission of information begins. which consists in the fact that its transmitter begins to transmit a signal in the direction of the opposite side, and the signal y (t) comes from the communication channel.

Suppose that at the first moment of time the input of input unit 1 receives such an analog signal, which is converted, for example, into a digital combination and ₍ (К ^ fit). This digital combination turns the first DAL 3 into

Analog voltage U ₄ (t). At this 15th time, the signal y (t) and the echo signal (far echo) come from the communication channel. due to the training signal, Ka output of the ADC 4 there is a value of about 20

Li (K _rt ,, At) = and <(k _n., At) + + y. (K ^, fit) + (K _n4 Mt) (3)

A.

where 1m (K _{and (} yX) - near echo _{;? 5}

U ₍ (K _n * i flt) - received signal;

U eyuDKpyyS) - distant echo,

The value E | (K _{P4 (} At) is supplied to the first input by subtracting 8. Since the input unit converted the signal to be transmitted to the first digital ³⁰ combination U <(Kfit), the content of the communication channel response (near echo ) to the corresponding digital combination, that is, U <(K <4t). Since U <(K _n , (fit) and 35 b \ (K _t fit) are the responses of the communication channel to the corresponding digital combination U ((Kfit ), then subtractor 8 compensates the near echo

N ((K p ♦ <it) = L <(K _{n 4 (} fit) -

- and I (k ₍ m) = y, (k _{n +)} at) + ⁺ UjxoiiKnn & t). (4)

According to the signal from the output of generator 7, a new value L, (K _{n +} , it) is written to the first block (5 memory instead of the previous contents (K, fit) at the address U ^ Kit), and in the second memory block 10, the same address is written result Μ, ίΚη ,, fit). Thus, in the first memory unit 6, a new sample of the transmitted signal is stored (together with the received and echo signal) 55

P <(K _{p + (} ^for digital combination

U <(Kat). If the input unit 1 then transmits other allowed combinations, the compensation process does not differ from the compensation process in the (n + 1) -th clock interval. Consider the ongoing processes if the input unit 1 again forms the first allowed combination Ui (Kfit). Then, at the output of ASCT 4 a signal

L, (K (^. 5 it) = U <(K _rt . ₅ fit) + + ygiKn ^ at) + П _{e> с5} (K _n ,, at). (5)

Subtractor 8 signal output

MDKn ^ oO = L ((K, wi at) -

- v. ((K _{n +} <4t) = y ₂ (K _{n +} . ₂ at) -

- for I (κ _η * <fit) Evil (k | _H - · at)

- Ujvo, (K, ..,. \ T \ (6) ila the output of the adder 9 sons

Nf (K _n , ₉ at) = m ₍ (Кп. _Г ^) ⁺ + N ((K _n . ₍ Fit) = + ⁺ Пе ко2 (K n + i) (7), etc. According to the signal with the output of the generator 7, the first memory block 6 records the value L <(K _n , „fit), and in the second memory block 10 the value

those. recovered received signal and echo.

Thus, no matter what combination the input unit 1 _s forms in the first memory unit 6, there is a response of the communication channel (near echo to this combination). For a transmitted signal, such a combination of nodes represents a filter plug, and for a signal coming from a communication channel, a direct wire with a transmission coefficient equal to unity. However, an echo signal is received towards the receiving device, which noise the received signal, reducing the reception quality.

Echo cancellation is performed in the input units using non-linear processing. In fact, samples of the total signal Nj (Kfit), consisting of samples of the received signal and echo signals, are supplied to the rectifier 18. The rectifier 18 converts the signals in absolute value, i.e. if at the input of the rectifier 18 there is a countdown of the total signal of a positive value, then at the output its value is stored. If the input of the rectifier 18 has a negative count, then it turns into a positive one.

The signal from the output of the rectifier 18 is then simultaneously supplied to the logarithm 19 and the automatic level control unit 12. In the logarithm 19, the logarithm of the value of KDKDb). If the value of [Nl (Kftt)] is shown at the input of the logari- ed fattor 19, then the value lg [N; (Kftt)] [appears at the output of this block. Since the echo signal in amplitude is approximately three times smaller than the received signal, then the i-S LN signal at the output of the logarithm 17. (KM)] log [Y ((KM)]. A peculiar absorption of the energy of the echo signal occurs. This energy is converted using non-linear processing of 20 g of the total output signal of the logarithm! 9.

For example, let N; (KAt) = +32. Then, for the benefit of the logarithm 19, a signal count of 1.5 appears. 25

If the value of the constant coefficient m is chosen, for example, equal to 40, then at the output of the non-linear processing unit 14, a signal reading equal to 1.5x40 = 60 appears. If, at the output of the rectifier 18, a reading equal to Nj (Kflt) = - 64, then this counting turns into rectifier 18 in the countdown equal to +64. At the output of the log of the rifmator 19, a countdown of the signal β 1β · (-64) = 1.8 is observed. At the output of the non-linear processing unit 14, there is a signal sample equal to 40 x log (-64) = - 1.8x40 = 72. In addition to processing in the non-linear processing unit 14, the sign is preserved for the values formed. The sign of the quantity is supplied from the output of the adder 9 to the second input of the interpolator 15. Thus, if a positive 45th count appears at the output of the adder 9, then the sign of this count is directly fed to the second input of the interpolator 15. In the first case, when N; (KAt) = +32, a value is observed (with the key 17 closed) + +40 log (32) = +60. In the second case, when N; (KDb) = -64, at the output of the interpolator 15, a value of -40 log (-64) = -72 is observed.

samples Nj (Kat) with a threshold '.pchchchchchiy generated by the unit 12 Aug' + - optical level control. If the threshold value at the output of the automatic level control unit 12 is greater than the reference value N- ₍ (K ^ t), then the signal ”1 is present at the output of the comparator 1 3, if the threshold value is less than the value of N _; (Kftt), then the output comparator 13 - 0. ”With '1 from the output of comparator 13, key 17 is closed, and with zero output signal of comparator 13, key 17 is open. If there is an error;;. :: the threshold generated in block 12 of automatic level control depends on the level of received and echo signals.

Amplitude value! ·. ' the received signal in the worst case is always about three times larger than the amplitude of the echo-signal (by 10 dB), therefore, the automatic level control unit 12 first averages the samples coming from the output of the rectifier 18. In addition, the threshold value is set at the output of the automatic level control unit equal to 2/3 of the average value of __ (at _Hz -, 4 Uey?). Coefficient ; A transmission equal to 2/9 is set by the corresponding voltage divider in the output circuit of block '2 of automatic level control.

Thus, the threshold value in block 12 of the automatic level control is equal to the average value of the echo signal. Moreover, in the comparator, the modules of the values ί N i (Kfit) j are compared with the threshold value П _Пюру = = | ^ί [Ctrl < ^£ > ^{+ and} elo (t) K Therefore, if Idru> ιΝ '<(К & t) I, then the key 17 opens. This indicates that the sum of the received and echo signals is near zero. Therefore, the signal from the output of the non-linear processing unit 14 does not arrive at the input of the interpolator 15, as a result, the signal region that lies at the output of the non-linear processing unit is artificially suppressed. 14 signal is sent to. key 17. The signal from the output of the comparator 13 is supplied to the second input of the key 17. In the comparator 13, the echo value is compared. In this case, the received signal sample is completely noisy by the echo signal.

This count with great maturity can probably be accepted. therefore his

54 C 008 '10 must be erased. The operation of erasing unreliable samples is performed by key ключ 7 in combination with comparator 13 and automatic adjustment unit 12

I level. If! I _APq z | , then the received signal is much larger than the echo. Therefore, this sample passes to the input of the interpolator 15 hours. key 17. Interpolator 15 rests on the Ί0 remaining reliable samples of the signal the shape of the received signal. When separating phase-shifted, frequency-modulated, or amplitude-modulated spins, 15 large and small amplitudes are always alternated. Therefore, at the output of the key 17, reliably received samples of N-, (KpS) and zero (erased) samples are always alternated. The flow of samples from the output of the key is supplied to the interpolator 15. In the delay line 21, the adder 22 and the divider 23, the arithmetic average is calculated. Suppose, at the input of the interpolator 15 there is a sample stream in the form of ... 12, 10, 7, 0, -7, -10, -11 ... At the output of the interpolator 15, the sample stream is (12 + 7) / 2; (10 + 0) / 2; (7-7) / 2; (0-10) / 2; (-7-10 / 2; ... etc. etc. At the output of the interpolator 15, successively reconstructed samples of the received signal appear sequentially, which are then converted into an analog voltage and given out to a consumer,

Thus, using non-linear processing, the echo samples are suppressed, and the shape of the received signal is restored by interpolation 40.

Claims (2)

Claim 1. Device for separating transmission directions in duplex communication systems according to ed. St., No. 1133675, characterized in that, in order to increase the separation accuracy by suppressing echo signals, a series-connected block of nonlinear signal processing, a key, an interpolator and a third digital-to-analog converter connected in series with rectifiers * · block, are introduced into it automatic level control and a comparator, the output of which is connected to the BTopoity key input, the adder output is connected to the second input of the interpolator and the rectifier input, the output of which is connected to the input of the nonlinear signal processing unit, the output The second converter is connected to the second input of the comparator, and the non-linear signal processing unit is made in the form of series-connected logarithm and multiplier, the second input of which is the constant coefficient input of the non-linear signal processing unit, whose signal input and output are respectively the input logarithm and multiplier output, 2 _; Device by Π. ί, T L I · cha ue 1 with the fact that the interpolator is made in the form of series-connected delay lines, an adder and a divider, and the first and second inputs of the delay line are connected to the second and third inputs of the adder and are respectively the first and the second inputs of the interpolator, the output of which is the output of the divider. Compiled by 0. Andrushko Editor L. Pcholinskaya Tehred L _with Oliynyk. Proofreader M. Sharoshi Order 231 Circulation 520 Subscription VNIIIPI State 113035, Committee for Inventions and Discoveries under the USSR SCST Moscow, Zh-35, Raushskaya nab., D. 4/5 Production and Publishing Plant Patent, Uzhgorod, st. Gagarina, 101