RU27766U1 - MODEM FOR MULTI-BEARING SIGNALS - Google Patents

Description

The technical field to which the utility model relates.

The invention relates to the technique of communication systems, mainly radio communications in the high frequency (HF) band, and can be used to construct devices for transmitting and receiving discrete information in channels characterized by multipath propagation and Doppler scattering.

State of the art

A prototype device for transmitting and receiving multi-carrier signals is known (S. B. Weinstein and Paul M. Ebert, Data Transmission by Frequency-Division Multiplexing Using the Discrete Fourier Transform / LEEEE Transactions on Communication Technology, vol. 19, no. 5 , 1971, pp. 628-634), which uses the discrete Fourier transform for generating and demodulating signals with many carriers. The device (modem) contains

on the transmitting side, a serial-parallel converter, a generator of complex values of the transmitted symbols with relative phase shift keying (OFM), a block of the inverse discrete Fourier transform (ODPF), a parallel-serial converter, a guard interval addition unit, and a digital-to-analog converter (DAC). On the receiving side, the device contains: an analog-to-digital converter (ADC), a synchronizer, a guard interval removal unit, a serial-parallel converter, a discrete Fourier transform (DFT) block, OFM symbol demodulation block, and a parallel-serial converter. Such a modem provides a spectral efficiency of 1 bit / Hz for binary OFM or 2 bit / Hz for four-position OFM (without taking into account the duration of the guard interval).

A further increase in spectral efficiency in multi-carrier modems, as well as in single-carrier modems, is possible in two ways (Prokis J. Digital Communication. - M.: Radio and Communication, 2000):

1) through the use of multi-position methods of relative phase manipulation (for example, through the use of 8- or 16-position OFM),

2) through the use of quadrature amplitude manipulation (QAM), i.e., simultaneous manipulation of the carrier in phase and amplitude.

It turns out to be problematic due to the presence of Doppler scattering in the RF communication channels. So, the presence of a small Doppler frequency shift leads to a rotation of the points of the OFM signal constellation. It is possible to compensate for such a turn by averaging the Doppler phase shift over all carriers and turning the points of the signal constellation by an angle opposite to the obtained value. However, the maximum Doppler frequency shift that can be compensated in this way is

TS - the duration of the clock interval (symbol),

t is the number of signals transmitted by one OFM symbol.

For example, if an 8-position OFM is used, with a modulation rate

20 Baud, then the maximum frequency of the corrected Doppler shift is not

exceeds 1.25 Hz. In practice, for RF channels this value

may be several times larger.

Using a QAM with a signal constellation of the “star” type, it is possible to increase the spectral efficiency of transmission without imposing more stringent requirements on the maximum Doppler shift in the communication channel. However, QAM demodulation is possible only using coherent reception methods, which is not always possible in the high frequency range due to phase fluctuations and frequency selective fading in the channel. There are systems using

f j

J dop are fabric equalizers, which allows the use of coherent modulation of the KAM type (see, for example, patent EP0734132), however, they are highly complex, require a rather large amount of a priori information on the statistical characteristics of the channel and, therefore, turn out to be ineffective in real channels of the high-frequency range.

A device is known (patent EP1021019), which allows to simplify the formation and processing of QAM signals by applying the so-called. quasi-relative modulation. The term “quasi-relative” implies the use of a relative modulation technique for encoding and decoding the phases of signals and coherent technique for encoding and decoding signal amplitudes. The system assumes the transmission of information in frames, each of which contains one control symbol without amplitude modulation (with constant carrier amplitudes), and several information symbols modulated both in amplitude and phase. The control symbol is used to evaluate and correct the signal amplitude under the assumption that during the duration of one frame the characteristics of the communication channel do not change. Looking at a simpler (and, as a result, more stable) modulation / demodulation scheme, the device is unsuitable for operation in high-frequency communication channels due to the significant unsteadiness of the characteristics of such channels and the presence of frequency-selective fading.

, / (7

The essence of the utility model:

The purpose of the utility model is to develop a multi-carrier modem that

a) has increased spectral efficiency compared to modems using relative phase shift keying with multiple carriers,

b) is capable of operating under conditions of Doppler scattering, phase fluctuations, and frequency-selective fading in the communication channel,

c) does not require increased hardware and software costs for the implementation of modulation / demodulation functions.

The goal is achieved due to the fact that:

a) the transmitted information is laid in the relative phase change of each carrier from symbol to symbol, as well as in the ratio of the amplitudes of two or more simultaneously transmitted carriers (and not in the absolute value of the amplitudes of the carriers, as is the case with QAM),

b) for demodulation and decoding of information embedded in the values of the carrier amplitudes, incoherent methods are used, which reduces the requirements for the stability of the communication channel and simplifies the software and hardware implementation of the modem.

The essence of the utility model is that in a modem with many carriers, the structural diagram of which (see figure 1) contains on the transmitting side:

a) series-parallel Converter (1),

b) shaper of complex values of transmitted characters with relative phase shift keying (2),

g) block inverse discrete Fourier transform (3),

d) parallel-serial Converter (4),

e) a block for adding a guard interval (5),

g) digital-to-analog converter (6), as well as on the receiving side:

a) analog-to-digital Converter (7),

b) synchronization unit (8),

c) a block for removing the guard interval (9),

g) series-parallel Converter (10),

e) a unit of discrete Fourier transform (11),

e) a block of symbol demodulation with relative phase shift keying (12)

g) parallel-serial Converter (13),

in the transmitting part, an amplitude modulator (14) is additionally introduced, and in the receiving part, a relative amplitude demodulator (15), designed to transmit an additional stream of information.

The input of the amplitude modulator (14) receives binary symbols of the additional information stream, divided into blocks of L binary symbols, each input block being encoded with values

The coding is performed in such a way that each combination of bits of the input block corresponds to a set of values of the amplitudes of the carriers, such that M / 2 of them has an amplitude of min, and the remaining M / 2 have an amplitude of v4max min. stored in the storage device of the modem and used both during transmission and reception. The structure of the transmitted signal (in the frequency domain) illustrating the principle of relative amplitude modulation for a multi-carrier signal is shown in FIG.

In the relative amplitude demodulator (16), “soft decoding” is performed, for which the following operations are sequentially performed:

a) the amplitudes of the set M of received neighboring carriers are calculated

A ,, k 1, ..., M

b) the centered values of the amplitudes are found by subtracting from 4 the average value of the amplitude for the group:

,, ..., M (1)

c) 2 correlation metrics are calculated:

. (2)

where the weighting factors are determined by the totality of the amplitudes of the carriers formed when the amplitude modulator appears at the input of the combination of the input L-bit block, according to the rule:

  1, if the / -th carrier, when coding the ith input combination, had

the amplitude of Atah,

c.J -1, if the / -I carrier when coding the u-th input combination had

amplitude Attg), the metric CMj is determined, which has a maximum value, and a decision is made that a combination corresponding to this metric is transmitted.

The implementation of the relative amplitude demodulator can be simplified, at the cost of small energy losses, if you use “tough decisions. In this case, the centered values of the amplitudes A ,,

kF, ..., M are quantized into two levels according to the rule: A sig, n (A /), where the sign () function returns the sign of the argument. In this case, instead of calculating the CL / metrics, one can count the number of distinct bits E A and c. Such

the operation is equivalent to calculating the Hamming distance between A and c

The transferred combination with this method of demodulation is considered an allowed combination that differs from the one accepted in the least number of digits.

the input bit of the additional information note is encoded by the amplitudes of only two adjacent carriers. The operation of the relative amplitude demodulator can be represented at the level of a simple block diagram depicted in FIG. The relative amplitude demodulator in this case contains, for each pair of neighboring bearing blocks, amplitude calculation units (16) and a comparator (17), from the output of which bits of the additional information stream are read.

As follows from the description of the modem and its structural diagram (Fig. 1), all transmitted information is divided into two streams: primary and secondary. The noise immunity of the transmission of information on the main and secondary streams is determined by the ratio of the amplitudes max and min - Moreover, the stronger the difference between the ax and min values, the higher the noise immunity of the transmission in the additional stream, and the lower the noise immunity of the transmission in the main stream. Thus, the choice of parameters and min should be based on a compromise. As shown by theoretical and experimental analysis of the proposed schemes, it is impossible to select a ratio for max and at which the probability of error in the main and additional information flow will coincide in the entire range of signal-to-noise ratios. However, it is possible to select such values of lax and / 4min at which the indicated probabilities will be approximately the same for large signal-to-noise ratios. In particular, the author theoretically and experimentally

it was found that when using binary OFM and an additional amplitude modulator / demodulator with parameters for transmitting the main stream, the noise immunity of the information transmitted along the main and additional streams will be approximately the same if the condition

4na r: 4m 0.414. (3)

Lpah +

Similarly, when using the four-position OFM, the condition

Lvah-Lpsh, OD77. (4)

tach + aiin

The technical result that is achieved when using the utility model is to increase the speed of information transfer in modems with many carriers without expanding the occupied frequency band. At the same time, the considered schemes for constructing a modem with many carriers do not tighten the requirements for the maximum allowable Doppler shift in the communication channel, which, in particular, allows their use in the RF channels.

The figures characterizing the relative increase in spectral efficiency when using the utility model (PM), for some parameter values are given in table. 1.

Tab. 1.

Information confirming the feasibility of implementing the utility model: The practical implementation of the described modem can be performed on the basis of a digital signal processor (DSP), a microcontroller or a general-purpose processor, as well as on the basis of programmable logic integrated circuits (FPGAs). The author of the utility model implemented mock-ups of modems based on ADDS-2106x-EZ-LITE evaluation boards, the core of which is 32-bit DSP ADSP-21061 from Analog Devices. All experimental results described below were obtained using these mock-ups.

One of the main issues that affect the feasibility of efficiently implementing a utility model is the correct selection of the parameters L and M. When choosing the values of L and M, proceed from the following considerations. First of all, the number of binary characters in a block must be such that the total

the number of combinations in a block of L binary symbols did not exceed the number of different combinations of M binary values having the same weight equal to M / 2, i.e. prerequisite must be met

.(5)

In addition, it should be borne in mind that when implementing the proposed modem, the spectral transmission efficiency increases compared to the prototype modem at L / M bits / Hz. It is easy to see that an increase in spectral efficiency can be achieved by increasing the parameters L and M. However, on the other hand, the number of M carriers combined into one subgroup must be such that the signal fading in the frequency band occupied by these carriers is not of a frequency-selective nature. Otherwise, demodulation on the basis of the ratio of the amplitudes of the carriers in the subgroup will be impossible. It is also necessary to take into account that with an increase in L, hardware and software costs for the implementation of amplitude modulator and demodulator blocks increase.

Thus, the final choice of parameters is determined by the modem designer based on:

a) processor resources,

b) the required increase in spectral efficiency (transmission speed),

(A // 2) (M / 2)

As an example of the practical implementation of the utility model, one can cite the modem models developed on the basis of ADDS2106X-EZ-LITE evaluation boards. In this example, the communication channel was the RF radiotelephone channel with a standard bandwidth of 3.1 kHz. The sampling frequency of the channel signal was taken equal to 8 kHz. The signals were digitized using a 16-bit ADC / DAC and processed in a 32-bit circuit. The size of the DFT block was 128 samples; the duration of the guard interval was 24 samples. The total number of carriers was 40. Six different multi-carrier modems were experimentally studied:

1) a modem without additional modulation of carrier amplitudes using binary OFM;

2) a modem without additional modulation of carrier amplitudes using a four-position OFM;

3) a modem using binary OFM with additional modulation of carrier amplitudes and parameters, Z, l;

4) a modem using a four-position OFM with additional modulation of carrier amplitudes and parameters,

5) a modem using binary OFM with additional modulation of carrier amplitudes and parameters,

6) a modem using a four-position OFM with additional modulation of carrier amplitudes and parameters, 6;

When implementing modems according to p. 5) and 6), 40 carriers were divided into 5 groups of 8 carriers in each group. The number of different combinations of amplitudes of 8 carriers, in which half of them has an amplitude of min and the other half of max, is equal to 70. Of these 70 combinations, 64 combinations were selected and tabulated, which were used to transcode the 6-bit blocks of the additional information stream into values amplitudes of 8 carriers. For each group, modulation and demodulation was carried out separately, but the encoding / decoding table was used alone for all five groups. Thus, it took only 64 bytes of read-only memory to store the encoding / decoding table. All functional units (with the exception of the blocks of analog-to-digital and digital-to-analog conversion) of the modems were implemented in real time by means of software based on circuits with a capacity of 40 million operations per second.

As the results of experiments showed, modems that use additional modulation of carrier amplitudes lose from 2 to 5 dB signal-to-noise ratio, compared with modems in which such modulation is not used. The indicated loss is an inevitable payment for increasing the information transfer rate (see Table 1), and is comparable to the loss that occurs when the number of points in the OFM signal constellation increases.

All further calculations will be valid under the assumption that the channel is subject to additive white Gaussian noise (ABGS).

An error probability estimate for the modem shown in FIG. 1, includes the calculation of the probability of error for the main and additional information flows. The indicated probabilities are related quantities and depend both on the average signal-to-noise ratio (SNR) and on the choice of values and 4max. We introduce the following notation:

A 4t. JJ 4na l4nin (6)

2+

 min

where Afn - represents the average amplitude of the carriers, and the parameter k can be interpreted as the coefficient of auxiliary amplitude modulation.

Suppose that the main information stream is transmitted using binary OFM. As is known, in systems with binary OFM, the probability of an error per bit can be found by the formula (Prokis J. Digital Communication. - M.: Radio and Communication, 2000):

where y is the signal-to-noise ratio per symbol.

Since the amplitude of each carrier can take two possible values - and, the signal-to-noise ratio is also a random variable taking two possible values jmm and Utah- If

P 1 exp (-7), (7)

to denote the average SNR by y r T gD is the spectral density of the noise, then the values and Utax can be expressed as:

r.in (l-) V., r.ax (l +) V. (8)

/ yVQZ / VQ

Since with such a modulation scheme for half the SNR carriers will be Ymin, and for the other half Ymax), the resulting error probability can be found by simple averaging over all signal carriers, i.e.

Pm - J-exp (- (H-) Vj + iexp (- (l-) Vj (9)

For a four-position OFM, the probability of error per bit is (Prokis J. Digital Communication. - M.: Radio and Communications, 2000):

P Q, (a, b), ((10)

(((i.

where ei (a,) exp - x - 1 (ab),

  )

Fn (n :) is a modified Bessel function of the ith order of the first kind, and the parameters a and b are defined as:

) X1 + lA) - (11)

As in the case of the binary OFM, the average probability of error is found by averaging the probabilities of the corresponding Ymin and for ah. The probability of error for the main stream, calculated by the above method, will be valid for any values of the parameters Mi Z-.

It is rather difficult to calculate the probability of error per bit for the additional information flow for arbitrary values of Mi L, generally, to say the least. However, in the particular case when, L, the indicated probability (P is easy to obtain, given that in this case the binary correlated signals are incoherently received. With this interpretation, the probability of error per bit can be calculated using formula (10), where the values of the parameters an are determined differently :

"F. | -Cl-Jl, (l-Jl + p, (12)

„2NQ) V)

ES is the signal energy of the carrier; its one information bit,

p is the cross-correlation coefficient of the signals corresponding to their transmission

zero and single characters.

Since in an additional stream each bit of information is transmitted in two

hen; by them, one of which always has an amplitude of mm, and the other an amplitude

max, FOR energy (E will be fair as follows; its equality:

. (l -) (l +) (l + A:) (13)

p j {(l + A:) sin6;, / + (l-A :)

 0. (14)

X - A:) sin6y, + (l + A;) sin () af /

where 6U1 and 6U2 are the frequencies of two adjacent carriers.

The result is a consequence of the orthogonality of any two carriers

signal from a group signal.

Considering (13) and (14), the parameters a & b can be expressed in a compact form:

«L / (1-) T., v (l + A:) V .. (15)

The above relations (9) - (11) and (15) allow an accurate calculation of the error probability per bit for the main and additional information flow for a given depth of auxiliary amplitude modulation (A :). It is easy to verify that with increasing k, the probability of error for the auxiliary information stream decreases, while the probability of error for the main information stream increases.

The parameter k can be selected in two ways. The first approach is to select the minimum possible modulation depth at which the probability of error per bit for the main information stream (for a fixed SNR) does not exceed a predetermined value. In this case, the noise immunity of the information transmitted through the additional information channel may be significantly lower than the noise immunity of the main information stream. Such an approach

appropriate in cases where auxiliary or control information is transmitted via an additional channel, which is of less importance and / or a higher degree of protection against channel errors. This approach may also be interesting in cases where the modem is used to transmit voice information. As you know, when encoding speech signals, part of the bits of the output stream of the speech codec is more sensitive to errors in the communication channel. In such systems, the auxiliary channel can be used to transmit the bits of the speech codec least sensitive to channel errors. Another approach is to choose the parameter k from the condition of equal noise immunity of the information in the main and additional channels, i.e. R R. Unfortunately, in this case, the optimal value of k depends on

SNR. To find the value of the parameter k, which leads to the equality of the probability of errors in the primary and secondary channels, it is possible by numerically solving equation f) 7. 7. But, as the analysis showed, for y x

the quantity k tends to some boundary value. This boundary value can be calculated analytically for both the binary OFM and the four-position OFM at and.

Calculations show that the value of the optimal depth of auxiliary modulation for large SNR for modems with binary OFM is: k K. 0.414, and for modems with four-position OFM k k, 0.277.

The error probabilities per bit for systems with binary and four-position OFM for optimal values of k (0.414 and 0.277) are shown in Fig. 4 and Fig. 5, respectively. As the graphs show, when using the proposed modulation / demodulation scheme, the energy loss will be about 4 dB for a system with binary OFM (the transmission rate increases by 1.5 times) and approximately 3 dB for a system with a four-position OFM (transmission speed increases by 1.25 times).

List of drawings

FIG. 1 is a general block diagram of a multi-carrier modem

FIG. 2 is a drawing explaining the principle of relative amplitude

modulation for a multi-carrier modem (signal is shown schematically

in the frequency domain).

FIG. 3 is a block diagram of a relative amplitude demodulator for

modem with parameters, L.

Figure 4 - probability of error per bit for a modem with binary OFM, with

depth of auxiliary amplitude modulation, 414.

FIG. 5 - probability of error per bit for a modem with a four-position OFM,

at a depth of auxiliary amplitude modulation, 277.

Claims (5)

1. A modem for signals with many carriers, containing on the transmitting side a serial-parallel converter, a shaper of complex values of the transmitted symbols with relative phase shift keying, a block of inverse discrete Fourier transform, a parallel-serial converter, a block for adding a guard interval and a digital-to-analog converter, and on the receiving analog-to-digital converter, synchronization unit, guard interval removal unit, serial-parallel conversion a caller, a discrete Fourier transform unit, a symbol demodulation unit with relative phase shift keying, and a parallel-serial converter, characterized in that it further comprises an amplitude modulator in the transmitting part, to the input of which an additional binary information stream is supplied, and the output amplitudes of the carriers further to the block of the inverse discrete Fourier transform, and the values of the amplitudes are formed by the information block in such a way that in each sub an uppe of several carriers, half of the carriers has amplitudes A _min and the other half has amplitudes A _max > A _min , and, in addition, additionally contains a relative amplitude demodulator on the receiving side, to the input of which complex symbol values from the discrete Fourier transform block are supplied, and the decoded binary information stream is removed from the output, and the decoding of the information block in each subgroup is carried out in a “soft” way by calculating the correlation metrics for each of the allowed combinations amplitudes in a subgroup and determining the combination corresponding to the maximum metric.  2. The modem according to claim 1, characterized in that the relative amplitude demodulator comprises a unit for calculating the average value of the amplitudes of the carriers in the subgroup, a binary quantizer, the quantization threshold of which corresponds to the average value of the amplitudes in the subgroup, and the decoding of the information block corresponding to one subgroup is carried out by comparison a binary-quantized set of amplitudes in a subgroup with all allowed binary-quantized sets of transmitted amplitudes and determining the allowed combination and having the smallest Hamming distance.  3. The modem according to claim 1, characterized in that each bit of the additional information stream is transmitted by a subgroup consisting of only two carriers, and the relative amplitude demodulator in this case contains, for each pair of adjacent carriers, amplitude calculation units and a comparator from which bits are read out additional information flow. 4. The modem according to claim 3, characterized in that a binary relative phase shift keying driver and demodulator is used to transmit the main information stream, and a relative amplitude modulation modulator and demodulator are used to transmit the additional information stream, the ratio of the difference and the sum of the amplitude values A _max and And _min , formed in the modulator, is 0.414. 5. The modem according to claim 3, characterized in that a modulator and a demodulator of four-position relative phase shift keying is used to transmit the main information stream, and a modulator and a demodulator of relative amplitude modulation, and the ratio of the difference and the sum of the amplitudes A _max and And _min formed in the modulator is 0.277.