SU1525919A2 - Method of correcting inter-symbol interference - Google Patents

Abstract

The invention relates to telecommunications and can be used in communication systems with a high specific information transfer rate encoded with a convolutional code. The invention consists in the refinement of constant factors, which take into account the correction signal of intersymbol interference, which allows to improve the accuracy of the correction in the conditions of a priori uncertainty of the channel characteristics. 3 il.

Description

The invention relates to telecommunications and can be used in communication systems with a high specific information transfer rate encoded with a convolutional code.

The aim of the invention is to improve the accuracy of the correction under the conditions of a priori uncertainty of the channel characteristics.

Figure 1 shows signal plots explaining the operation of the method; figure 2 - structural diagram of the device; figure 3 - functional diagram of the averager.

The device contains (FIG. 2) analog-digital converter 1, shift register 2, multipliers 3, memory elements 4, adder 5, averager 6, delay block 7, multipliers 8 and 9, subtractor 10, multipliers 11 and 12, threshold block 13, a pseudo-random sequence generator 14, an adder I 15, multipliers 16 and averagers 17.

Averagers 6 and 17 contain (FIG. 2) an accumulator 18,

a pulse counter 19, a divider 20, a key 21, and a memory register 22.

The method includes the following steps (Figure 1).

The conversion of the input analog signal (from the output of the demodulator) into each discrete time interval into a digital signal consisting of a sign pulse signal (S; +1) whose amplitude is equal to one; the polarity is determined by the polarity of the input analog signal1 (FIG. 1a), and an information pulse signal of positive p-polarity with amplitude (m-) equal to the magnitude of the amplitude of the input analog signal (Fig. 16), UjS i P |,;

Memorization of n, sign pulse signals preceding, and n sign pulse signals, is followed by a given 1-digit pulse signal, where n and n correspond to the values of the memory and backup of the signal in the canapeline (Fig. 1a);

S

(L

Formation of the corrected sign pulsed signals by changing the unit amplitude of each of the filled sign pulsed signals by K-Ci) times (Fig. 1c, d), where K- () are constant factors calculated for each discrete time interval taking into account the pulse communication channel characteristics;

The formation of the total signal (ui), equal to the sum of the corrected sign pulse signals (Fig. 1a);

The summation of N sequentially received information signals (figb), where N is an arbitrary number of more (p - p);

Formation of the signal averaged over N clocks (Fig. 1e) with amplitude m

The formation of the resulting signal by the formula (Sjmj - Lim) (fig.1zh

Output Recovery

l a

S, t; by multiplying the sign of S by modulus m;

The formation of the total signal and the recovered output signal and the signal of a pseudo-random sequence (PS), the amplitude of which is equal to the average amplitude n of information pulse signals (Fig.13, and);

The formation of the total signal (Fig. 1k) of the sign signal (Fig. 1l), whose amplitude is equal to a predetermined fixed value, and the polarity corresponds to the polarity of the total signal:

The formation () of the sign correlation signals between the sign pulsed signal (Fig. 1m of the sum signal and n, the pre-existing and Uj subsequent sign pulse signals relative to the i-ro sign pulse signal;

Formation () of the signals of mutual correlation averaged over N clock cycles (Fig. &O;

The formation of (,) refined signals characterizing the impulse response of the communication channel by summing up the new values of the mutual correlation signals with the old values;

Conversion of the resulting signal into an output digital signal consisting of an output sign pulse signal whose amplitude is equal to 1, and an output information pulse signal of positive polarity with an amplitude equal to the magnitude of the amplitude of the resultant signal.

The device works as follows.

Analog-to-digital converter 1 generates pulse and glitch signals from the input signal. The sign signals fill register 2 on n,., + N, + 1 bits, the numbers n, n characterize the channel delay and memory, respectively, and are weighed in multipliers 3 with previously calculated

f, the K: () coefficients of the channel impulse response stored in memory elements 4.

The weighted signals are summed in adder 5 to form an averaged corrective signal, as the signals adjacent to the signal relative to which the decision is currently made are weighted and summed.

5 The correction signal from the output of the adder 5 is generated in the multiplier 9 by the average signal of the module from the output of the averager 6.

The signal from the output of analog-to-digital converter 1 after a delay in block 7 (necessary to compensate for the delay time of the sign signal in register 2) and change its polarity in multiplier 8 (in case of a negative sign pulse corresponding to this module) is corrected in the subtractor 10 signal from the output of the multiplier 9.

Signals and module signals from the subtractor 10 are output signals of the device (decoder input signals) and simultaneously received at the input of multiplier 11, where they are multiplied (the analog signal samples are reconstructed) and fed to the first input of the threshold unit 13 for comparison with the normalized signal received to the second input of the threshold unit 13 of the multiplier 12, where the sign signal of the pseudo-random generator 14 is normalized by the amplitude of the signal from the output of the averager 6. To form (ni + n) sign correlation signals tion on per5 stems 16 receives inputs of multipliers preceding claim and front, with respect to the subsequent i-ro sign pulse sign pulse signals ko0

five

0

five

0

The second ones are multiplied with the sign pulse signal of the sum signal from the output of the threshold unit 13. Then these signals are summed in (). averaging 17 through N clock cycles and the thus obtained new coefficients, which refine the impulse response of the communication channel, are added to the old values of the coefficients and are recorded in memory elements 4.

The proposed method of correction of intersymbol interference provides an increase in the reliability of information in soft solutions (metrics) of convolutional code symbols under conditions of a priori uncertainty of channel characteristics and thereby improves the reliability and decoding of a convolutional code.

Constant multipliers (K :), which take into account the intersymbol interference correction signal, are determined by calculating the correlation signals between the sign of the output sum signal, representing the sum of the output signal and the pseudo-noise sequence signal with an amplitude equal to the average amplitude of the input signal, and the corresponding sign of the input signal, separated from the corrective i-ro signal by + j symbols.

The use of symbols when multiplying instead of the full values of the signals greatly simplifies the calculation of correlation signals, thereby increasing the speed of the device and

the possibility of matching it with high-speed communication channels.

Summation of the output signal with a pseudo-random sequence with an amplitude equal to the average amplitude of the input signal ensures the invariance of the estimate of constant factors (K :) in a wide range of signal-to-noise ratios when using correlation of signs:

Claims (1)

Invention Formula Correction method of intersymbol interference by auth.St. P 1300646, characterized in that, in order to improve the accuracy of the correction, under conditions of a priori uncertainty of channel characteristics, a pulse signal of unit amplitude and a random sign of PS ± 1 is formed in each clock interval of time and the constant factors are calculated by the formula K-a) K- (t- i) + & Kj (C), Where 1 N g UR- (I) - sign | PSt; + s (8; w; s; PG) j S; -j, a is the clock number of the clock interval. mn rm Phi2.1 entrance f / nut FIG. g