RU2528134C1 - Device for decoding signals passing through multibeam communication channel - Google Patents

Abstract

FIELD: radio engineering, communication.

SUBSTANCE: device for decoding signals comprises two first-stage correlators, two additional first-state correlators, two second-stage correlators, two channel pulse reaction estimate correction units and a decision device, wherein inputs of all first-stage correlators and the additional first-stage correlators are combined and are the input of the device for decoding signals.

EFFECT: enabling reception of a message regardless of the number of pulses in the transmitted message and the stability interval of the pulsed reaction of the propagation channel.

2 dwg

Description

The invention relates to the field of transmission of discrete information and is intended for use in decoders of communication signals transmitted in multipath channels.

When transmitting communication signals through a multipath channel, distortions of their shape occur, and for the coherent processing and decoding of received signals, control (estimation) of the channel impulse response (IRF) is necessary. In this regard, most of the known solutions to these problems are based on radiation along with information (i.e., unknown at the receiving end of the communication system) pulses of test pulses, which are used to evaluate the KFM or, more precisely, the KFM in the frequency band of the test pulse. Such a transmission principle is referred to as a “system with a test pulse and prediction” (SIIP) (see, for example, [1], section 3.1). This principle underlies, in particular, objects [2-4].

The disadvantage of the principle of constructing a communication system for which well-known analogues are designed is the relatively low quality of reception (decoding) of messages, caused either by a loss of time during the transmission of test and information pulses separated in time (as is the case in [1]), or by the action of test pulses interfering with the reception of information pulses (and also vice versa, the action of information pulses that interfere with the reception of test pulses).

The closest in technical essence to the claimed object is the device described in [5]. It is considered as a prototype. The prototype solves the following problem. One of two possible signals (or symbols) is transmitted - S1 (t) or S2 (t) (in relation to the digital embodiment of the decoding device, we further consider all signals as functions of discrete time, i.e., consider signals S1 (n) and S2 ( n), where n is the index of the discrete time argument). As noted above, the shape of the transmitted symbol during propagation in the multipath channel was distorted, described as a convolution of this symbol with the KFM, the shape of which is a priori unknown. At the receiving point, a decision must be made about which of the two characters was transmitted (i.e., decode the transmitted message).

The block diagram of the prototype is shown in figure 1; explanations on it are given below when describing the principle of operation of the prototype.

The principle of operation of the prototype is as follows. For each possible moment of arrival of the connected signal, a cross-correlation between the received signal and each of the two possible symbols S1 (n) or S2 (n) is calculated. This action is performed by the correlators of the first stage (positions 1-1 and 1-2 in figure 1). In this case, the correlators of the first stage are essentially matched filters, i.e. each correlator of the first stage plays the role of a filter consistent with the corresponding symbol. As a result of performing this function, temporary implementations are formed at the outputs of the first-stage correlators 1-1 and 1-2, and at the output of that correlator, the reference oscillation of which coincides with the actually transmitted symbol, this temporary implementation is an estimate of the KFM in the working frequency band (with noise ), and at the output of another correlator, only noise. Due to the fact that the prototype does not know which of the two possible symbols was transmitted, it sums up (the same time samples) time realizations generated at the outputs of both correlators of the first stage (adder 2 in FIG. 1). This result of summing up the estimate of the KFM obviously contains. The adder 2 is also cumulative over a moving time interval, i.e. it accumulates arrays of estimates of the KFM (mixed with noise), which are formed sequentially in time as a series of connected symbols arrives. Such accumulation is necessary, in particular, to ensure a decrease in the correlation of noise at the inputs of each of the second-level correlators. The number of indicated accumulated implementations is limited both by the number of characters contained in the transmitted message and (this is a more stringent condition) by the stability interval of the KFM.

Next, the calculation of the correlation between the estimate of the KFM (it is formed at the output of the adder 2 and fed to the reference inputs of the correlators of the second stage) and each of the temporary realizations formed at the outputs of the correlators of the first stage 1-1 and 1-2 is implemented. This function is performed in the correlators of the second stage (positions 3-1 and 3-2 in figure 1).

Further, for concreteness, we assume that the symbol S1 (n) is transmitted. Moreover, in the correlator of the second stage 3-1, a response is actually formed proportional to the energy of the received multipath signal (i.e., distorted in the shape of the symbol S1 (n)), which corresponds to the effect of coherent addition of rays. This response is characterized by a high level. At the output of the same correlator of the second stage. 3-2 in this situation, only the realization of noise is formed. It has a low level mainly because the noise coming to the inputs of the correlator of the second stage 3-2 noise with a large number of implementations accumulated in the adder 2 estimates of the KFM are practically uncorrelated. Further, when comparing (in the solving device 4) with each other and / or with a threshold of response levels formed at the outputs of the correlators of the second stage 3-1 and 3-2, a decision is made on the actually transmitted symbol. So, in the case under consideration, as a rule, the level at the output of the correlator of the second stage 2-1 will be large and / or higher than the threshold, which will lead (as a rule, i.e., as a trend) to the decision that symbol S1 (n), which in the situation under consideration is the correct solution.

Thus, in the prototype, the problem of temporal dispersion of the energy of the communication signal is solved, because, as noted above, an effect is achieved that is equivalent to the coherent addition of all the rays (more precisely, the signals that came along all the rays).

The disadvantage of the prototype is the low quality of decoding that occurs when receiving a message consisting of a small number of characters, or with a relatively small interval of stability of the KFM. So, when receiving a message consisting of a single character, in the adder 2 of the prototype, the result of summing the responses of the correlators of the first stage 1-1 and 1-2 to the received additive mixture of the information pulse and noise is formed. In this case, at both inputs of each of the correlators of the second stage, the noise is largely correlated. This correlation of noise significantly eliminates the differences in the responses of the correlators of the second stage 2-1 and 2-2 to the received signal and thereby reduces the tendency to make the right decision. As the number of characters in the message increases, the effect of the above factor, which reduces the quality of decoding, although decreases, but very smoothly. A small interval of stability of the KFM in the light of the above explanation of the reasons for the low quality of decoding in the prototype leads to the same consequences as a small number of characters in the message.

The purpose of the claimed device is to eliminate the specified disadvantage of the prototype, i.e. providing the ability to receive (decode) the message regardless of these factors. The goal is achieved in that in a device for decoding signals containing two correlators of the first stage and two correlators of the second stage, as well as a solver, the common input of the correlators of the first stage being the input of the decoding device, the outputs of the first and second correlators of the first stage are connected to the first inputs, respectively the first and second correlators of the second stage, the outputs of the correlators of the second stage are connected to the inputs of the resolver, and the output of the resolver is the output of the device of decoding, two additional first-stage correlators are introduced, the inputs of which are combined with the input of the decoding device, and two channel response response correction blocks, the first and second of which are connected between the output of the first and second additional correlators of the first stage, respectively, and the second inputs of the first and second correlators, respectively second stage.

The inventive object can be used in a communication system in the general case with multi-position coding (while the number of correlators of the first and second steps accordingly increases in it). With an N-position code in it, the number of correlators of the first and second stages, additional correlators, and also inputs of the resolving device 8 is N. However, the minimum composition of its attributes occurs when it is used in a binary communication system. In this case, simultaneously with each information pulse (symbol) S1 (n) or S2 (n) (which correspond to codes, for example, “1” and “0”), the corresponding test pulse S1i (n) or S2i (n) is emitted, moreover, each test pulse is emitted in a frequency band that does not coincide (or does not coincide completely) with the frequency band of the corresponding information pulse. Each test pulse may differ from the corresponding information pulse, for example, only with a carrier frequency.

The block diagram of the claimed object is shown in figure 2, where indicated:

- 5-1 and 5-2 - correlators of the first stage;

- 5-3 and 5-4 - additional correlators of the first stage;

- 6-1 and 6-2 - correction blocks for evaluating the impulse response of the channel;

- 7-1 and 7-2 - correlators of the second stage;

- 8 - decisive device.

Each correlator of the first stage (5-1 ... 5-4) is implemented, for example, in accordance with the block diagram in Fig.5.14, p.295 of the book [6]. In this case, the signal input of the correlator is the lower input in the indicated Fig. 5.14, to which the received signal x (n) is supplied. The reference function of the correlator of the first stage (in the indicated Fig. 5.14 it is designated as h (n)) is stored in its long-term memory, which is not shown in Fig. 5.14 for simplicity. In the claimed device, the support functions of the correlators of the first stage h (n) are:

- correlator 5-1 - h1 (n) = S1 (n);

- correlator 5-2 - h2 (n) = S2 (n);

- additional correlator 5-3 - h3 (n) = S1 and (n);

- additional correlator 5-4 - h4 (n) = S2and (n).

When the correlator of the first stage is implemented in the spectral region (i.e., based on the fast convolution procedure), the discrete Fourier transform (DFT) operation is performed in advance on the support function of each of these correlators, and the array of the DFT result (the result of its complex conjugation) is stored in long-term memory corresponding correlator of the first stage. Above the arrays of samples of the input signal x (n), DFT is also performed, then elementwise multiplication (i.e., multiplication of the same samples) of arrays of DFT results over the reference function and input signal and the inverse DFT (DFT) from the array of the results of the specified multiplication are performed. The period of updating the array of samples of the input signal for adjacent time cycles of calculating the correlation in each of the correlators of the first stage is usually chosen equal to the duration of each of the pulses S1 (n), S2 (n), S1 and (n) and S2 and (n) (the duration of all these pulses in the simplest case, coincide), while the length of the DFT window is equal to the double duration of each of these signals. Four independently working correlators of the first stage are shown in FIG. 2 conditionally. When implemented in the spectral region, the DFT from the input signal that is part of these correlators may be common to all of them.

A variant of the block diagram of the correlator of the first stage in the time domain equivalent to the one considered is also possible; A description of this correlator option is given in [6], Fig. 6.18b, p. 418, where (in accordance with the current capabilities of the technique) instead of a recirculating delay line that stores an array of time samples of the reference signal when it is severely limited, a multi-bit shift register is implemented, storing the same samples represented by multi-bit code words.

The dynamics of updating the input and output data of the correlator under consideration is illustrated, for example, in [7, pp. 76-78.].

Each correlator of the second stage 7-1 and 7-2 is implemented similarly to the correlator of the first stage (preferably in the time-domain variant), with the only difference being that it does not have a long-term memory that stores the reference oscillation. The duration of the signal update cycle at the output of each correlator of the second stage can be, for example, one sampling period of the input signals.

$$\widehat{H}2и\left(n\right)$$

, сформированных в полосах частот испытательных импульсов S1и(n) и S2и(n) (эти оценки сформированы на выходах дополнительных корреляторов соответственно 5-3 и 5-4), например, в оценки ИРК $$\widehat{H}1\left(n\right)$$

$$\widehat{H}2\left(n\right)$$

в полосах частот информационных импульсов соответственно S1(n) и S2(n). При этом выходные сигналы блоков 6-1 и 6-2 вычисляются соответственно по формуламBlocks for estimating the impulse response of channel 6-1 and 6-2 recalculate estimates of the KFM, respectively $$\widehat{H}\mathrm{one}\mathrm{and}\left(n\right)$$

$$\widehat{H}2\mathrm{and}\left(n\right)$$

formed in the frequency bands of the test pulses S1i (n) and S2i (n) (these estimates are formed at the outputs of additional correlators, respectively 5-3 and 5-4), for example, in the estimates of the KFM $$\widehat{H}\mathrm{one}\left(n\right)$$

$$\widehat{H}2\left(n\right)$$

in the frequency bands of information pulses, respectively, S1 (n) and S2 (n). The output signals of blocks 6-1 and 6-2 are calculated respectively by the formulas

where the entry of the ODPF {Mj (k)} (for j = 1, 2) means the operation of the ODPF on the array of samples of the discrete spectrum Mj (k) (k is the discrete argument of the frequency), defined as

- массивы спектров, являющиеся результатами выполнения операции ДПФ над массивами временных отсчетов соответственно информационного сигнала S1(n), испытательного сигнала S1и(n), а также оценки ИРК в полосе частот испытательного сигнала $$\widehat{H}1и\left(n\right)$$

; S2(k), S2и(k) и $$\widehat{H}2и\left(k\right)$$

- массивы спектров, являющиеся результатами выполнения операции ДПФ над массивами временных отсчетов соответственно информационного сигнала S2(n), испытательного сигнала S2и(n), а также оценки ИРК в полосе частот испытательного сигнала $$\widehat{H}2и\left(n\right)$$

.where S1 (k), S1 and (k) and $$\widehat{H}\mathrm{one}\mathrm{and}\left(k\right)$$

- arrays of spectra that are the results of the DFT operation on arrays of time samples, respectively, of the information signal S1 (n), test signal S1 and (n), as well as estimates of the KFM in the frequency band of the test signal $$\widehat{H}\mathrm{one}\mathrm{and}\left(n\right)$$

; S2 (k), S2 and (k) and $$\widehat{H}2\mathrm{and}\left(k\right)$$

- arrays of spectra that are the results of the DFT operation on arrays of time samples of the information signal S2 (n), test signal S2and (n), respectively, as well as estimates of the KFM in the frequency band of the test signal $$\widehat{H}2\mathrm{and}\left(n\right)$$

.

(соответственно $$\widehat{H}2и\left(k\right)$$

) формируется, как промежуточный результат. При этом данный результат может быть использован для расчета по формуле (2) без реализации в блоке 6-1 (соответственно 6-2) операции вычисления ДПФ.When implementing the correlator of the first stage 5-3 (5-4) in the spectral region (i.e., based on the fast convolution procedure), the discrete spectrum used in relation (2) $$\widehat{H}\mathrm{one}\mathrm{and}\left(k\right)$$

(respectively $$\widehat{H}2\mathrm{and}\left(k\right)$$

) is formed as an intermediate result. Moreover, this result can be used for calculation by the formula (2) without the implementation of the DFT calculation operation in block 6-1 (respectively 6-2).

To ensure the correctness of the operation (1) when calculating all the DFTs, all the arrays initial for these calculations are supplemented with zero samples in accordance with the convolution calculation rules.

As a result, each of the correction blocks for estimating the impulse response of channel 6-1 and 6-2 performs such a conversion of the KFM estimates (mixed with noise) in the frequency ranges of the test pulses, respectively, S1i (n) and S2i (n), at which the KFM estimates are generated ( also mixed with noise) in the frequency ranges of information pulses, respectively, S1 (n) and S2 (n),

The solver 8 is a circuit for comparing the current signal levels at its inputs with each other and / or with a predetermined threshold stored in its long-term memory. If the level of one of the signals exceeds the level of the second signal (and, possibly, the threshold) at the output of the resolving device 8, for example, a code corresponding to the specified first signal (for example, “1”) is generated, otherwise, a code corresponding to the second signal ( for example, "0").

It should be noted that implemented by the correlators 5-1 ... 5-4 and 7-1. 7-2, the procedure for calculating the correlation between the input and reference signals is linear (the reference signals (or their spectra) of the correlators 5-1 ... 5-4 are stored in the long-term memory of these blocks, and the reference signals of the correlators 7-1 and 7-2 are operatively are generated by the correction blocks for evaluating the impulse response of the channel, respectively, 6-1 and 6-2), and therefore the correlators 5-1 and 5-2 without changing the principle of operation of the claimed device can be rearranged with the correlators 7-1 and 7-2, respectively. In this case, the connections between each of the correlators 7-1 and 7-2 with the output of the correction unit for estimating the pulse response of the channel, respectively, 6-1 and 6-2 are saved.

The principle of operation of the claimed device is as follows. When transmitting the information pulse S1 (n) and the corresponding to the test pulse S1 and (n) the output of the correlator of the first stage 5-1 (the reference oscillation of which coincides with the information pulse S1 (n)), a response is formed equal to the convolution of the KFM with the autocorrelation function of the pulse S1 ( n). At the same time, at the output of the correlator of the first stage 5-3 (the reference oscillation of which coincides with the test pulse S1i (n)), a response is formed equal to the convolution of the KFM with the autocorrelation function of the signal S1и (n). The mentioned responses of the correlators of the first stage 5-1 and 5-3 in this situation are estimates of the KFM in the frequency band of the information pulse S1 (n) and the test pulse S1 and (n), respectively. If the frequency ranges occupied by the information pulse and the corresponding test pulse do not coincide (i.e., if their autocorrelation functions do not coincide), the KFM estimates in the frequency ranges of these pulses are uncorrelated, regardless of which symbol was transmitted. As a result of recalculation of the estimate of the KFM in the frequency band of the test pulse to the estimate of the KFM in the frequency band of the information pulse carried out in the correction unit for the estimation of the impulse response of channel 6-1, formed at the inputs of the correlator of the second stage 7-1 in the case of the transmission of the pulse S1 (n) paired with a pulse, S1and (n)) become correlated (coinciding up to the noise occurring at these inputs). Because of this, when transmitting the symbol S1 (n), the response level of the correlator of the second stage 7-1 is high.

In the situation under consideration, i.e. when transmitting the pulse S1 (n) (paired with the pulse S1i (n)) at the outputs of the correlators of the first stage 5-2 and 5-4 (the reference vibrations of which coincide with the information pulse S2 (n) and S2i (n), respectively), responses representing realizations of noise (we consider that all considered pulses are quasi-orthogonal). These implementations are uncorrelated with each other, and they remain uncorrelated even after the conversion carried out by block 6-2. In this regard, in the situation under consideration, the response level of the correlator of the second stage 7-2 is low.

As a result, in the situation under consideration, the response level of the correlator of the second stage 7-1, as a rule, exceeds the response level of the correlator of the second stage 7-2, which is recorded by the decisive device 8, as a result of which, as a rule, the code "1" is generated at its output, corresponding to the momentum transfer situation S1 (n) (paired with the impulse S1and (n)).

When transmitting the information pulse S2 (n) and the corresponding test pulse S2i (n), the ratio of the response levels of the correlators of the second stage 7-1 and 7-2, as a rule, will be the reverse of the above. At the same time, at the output of the solving device 8, a code (0) is generated (as a rule) corresponding to the situation of transmission of the pulse S2 (n) (paired with the pulse S2 and (n)). In cases where the inventive device is used in communication systems of the SIIP type with multi-position coding (N-positional; in the above description of the claimed device, a particular case was considered with N = 2), the number of correlators of the first and second steps in it is equal to 2N and N, respectively, and correction blocks channel impulse response estimates - N.

If in the prototype there was a formation of an estimate of the KFM common for two processing chains, each of which is consistent with one of the possible information symbols (pulses), then the claimed device implements an individual for each variant of the transmitted information symbol (pulse) the formation of the estimate of the KFM.

The inventive device provides high quality decoding even with a small number of characters transmitted in the message and a small interval of stability of the KFM (just one character and the interval of stability of the KFM, not less than the duration of this character). In addition, there is also the following factor, which provides an advantage in the quality of decoding of the claimed device over the prototype. The quality of decoding is determined almost exclusively by differences (contrast) in the levels of responses (to received signals) of the correlators of the second stage. These levels, all other things being equal, the higher, the greater the correlation value of the signals at their inputs, as well as the higher the levels of these signals. In the prototype, the levels of the signals supplied to the reference inputs of both correlators of the second stage coincide, since the same signal of a relatively high level containing an estimate of the KFM is supplied to these inputs. In the claimed device, such a large-level signal (containing the KFM estimate) is supplied to the reference input of only the second-stage correlator, the signal input of which is connected to the output of the first-stage correlator, the reference oscillation of which coincides with the actually transmitted information pulse (for each of the second-stage correlators 7 -1 and 7-2, we consider the input that is connected to the output of one of the additional correlators of the first stage 5-3 and 5-4 to be reference; we consider the second input of each of their correlators of the second stage to be signal). This factor contributes to an additional "emphasis" of the signal level just at that input of the resolver 8, the increase of which is desirable for making the right decision.

Thus, the object of the invention is achieved.

Literature

1. D.D. Klovsky. Transmission of discrete messages over the air. M.: Communication, 1969.

2. A device for receiving discrete signals in a multipath communication channel. RF patent No. 2048701.

3. Digital device for demodulating discrete signals in a multipath communication channel. RF patent No. 2267230.

4. Device for transmitting discrete signals in a multipath communication channel. RF patent No. 959291.

5. Sussman S.M. A matched filter communication system for multipath channels // IEEE Trans. IT - 6. N 3. June 1960.

6. “The use of digital signal processing”, ed. E Oppenheim. M .: World. 1980.

7. L. Rabiner, B. Gould. Theory and application of digital signal processing. M .: World. 1978.

Claims (1)

A device for decoding signals transmitted through a multipath communication channel containing two correlators of the first stage and two correlators of the second stage, as well as a solver, the common input of the correlators of the first stage being the input of the decoding device, the outputs of the first and second correlators of the first stage are connected to the first inputs of the first and the second correlators of the second stage, the outputs of the correlators of the second stage are connected to the inputs of the solver, and the output of the solver is the output of the decoding triad, characterized in that it consists of two additional first-stage correlators, the inputs of which are combined with the input of the decoding device, and two channel impulse response estimation correction blocks, the first and second of which are connected between the output of the first and second additional first-stage correlators, respectively and second inputs of the first and second correlators of the second stage, respectively.

Images