RU2192708C2 - Method for compressing spectrum width of limited- frequency-band electric information signals - Google Patents

Abstract

FIELD: data transmission in electric communications. SUBSTANCE: method used when spectrum width or radio-channel width should be compressed and/or when hiding of messages transmitted should be ensured involves time sampling of continuous source signals, artificial increase in time steps between digital readings, and generation of continuous signals having accordingly narrower frequency band; in the process repetition period of digital readings is increased not by artificial delay of each next reading relative to preceding one (without changing these readings), but by fast convolution of readings of each group into one equivalent group reading or pulse whose amplitude and polarity unambiguously illustrate analog parameters of source pulses or readings of respective groups or pulses, entire sequence of readings being divided into separate groups in advance. EFFECT: provision for compressing spectrum width of communication signals without reducing amount of information and its transmission speed. 4 cl, 1 dwg

Description

 The invention relates to techniques for transmitting information and can be used in telecommunication systems when it is necessary to compress the spectrum width of the transmitted signals to reduce the frequency band of a wired or radio channel and (or) it is required to ensure the secrecy of the transmitted messages.

 There is a known [1] method for compressing the width of the frequency spectrum of signals by recording the original signal in some storage device (for example, in a tape recorder) and then reproducing and transmitting along the channel in slow motion.

 The disadvantage of this method is the increase in message transmission time (channel occupation time) and an additional delay in the received message relative to the message received from the source, which in most communication systems is unacceptable; transmission of continuous (without pauses) signals in this way is impossible.

The closest technical solution (prototype) can be considered a method [2] for converting the width of the frequency spectrum, allowing both expansion and contraction of the spectrum, in which the write operation in the storage device is preceded by a discrete representation of the signal. In this case, discrete samples of the original signal (the following with a period of T ₁ = 1 / 2F ₁ , where F ₁ is the upper frequency in the signal spectrum) are recorded in the memory at one speed, and are read (in the mode that provides spectrum compression) at a lower speed. Then, from the sequence of samples with increased clock intervals T ₂ (T ₂ > T ₁ ), continuous signals are generated, which are fed to the input of the communication channel.

 However, this method is also characterized by the above-mentioned disadvantages (increase in channel occupation time, additional message delay, inability to transmit continuous (without pauses) signals).

 The technical result of the invention is to compress the spectrum width of telecommunication signals without reducing the amount of information per unit time of existence of the original signal, i.e. without reducing the speed of information transfer; this ensures the secrecy of the transmitted messages.

 The essence of the invention lies in the fact that, as in the prototype method [2], the initial continuous signals are sampled by time, the clock intervals between discrete samples are artificially increased and continuous signals are formed with a correspondingly narrower frequency band, but unlike the specified method [2] the period of following discrete samples is not increased by an artificial delay of each subsequent sample relative to the previous one (without changing the samples themselves), but by a fast one (using modern x high-speed computing devices) by converting (collapsing) the samples of each group into which the entire sequence of samples is previously divided into one equivalent group sample (GO) - a pulse whose amplitude and polarity collectively uniquely reflect similar parameters of the initial pulses-samples of the corresponding group . Moreover, signals that allow compression of the spectrum, that is, having a limited frequency band, can belong to one or several (different) sources; in the latter case, a composite signal can be formed based on the frequency or time principles of signal separation.

The drawing illustrates one of the options for the formation of GO. The original signal (drawing a) u (t) (it can also be composite) as a result of time sampling with a step T ₁ = l / 2F ₁ , where F ₁ is the upper frequency in the spectrum of the signal u (t), is represented by a sequence of pulses samples u ₁ (t). This sequence is divided into groups, for example, 3 samples in each group. Given just such an option, drawing b is made. Moreover, the case is considered when the pulses displaying the first samples in groups (u ₁ (t) in Figure b) retain their parameters (polarity and amplitude), and the second and third pulse samples are quantized in level and encoded, i.e. represented by binary code combinations - signals u ₂ (t) in drawing b. Then, from each group of signals, consisting of u ₁ (t) and u ₂ (t) (drawing b), a group pulse-count u ₃ (t) is generated (drawing c).

As can be seen from the drawing, the signals u ₃ (t) follow with a period T ₂ = 3T ₁ , therefore, to convert a sequence of pulses u ₃ (t) into a continuous signal u ₄ (t), a low-pass filter (LPF) with a frequency of cutoff F ₂ = 1 / 2T ₂ = 1/2 • 3T F ₁ = _1/3. If, when dividing the samples (u ₁ (t)) of the original signal (u (t)) into groups, each of them will contain not 3, but n> 3 samples, then after the formation of group samples (GO) and processing them in the low-pass filter, it will increase and the compression ratio of the spectrum: F ₂ = F ₁ / n.

 The considered method is generally free from the disadvantages inherent in the prototype method [2]: a constant slight delay in received messages regarding messages received from the source is due only to the time of the formation of the GO and restoration of the original signals, i.e. It is connected with the technical perfection of computing devices that implement the corresponding algorithms, and does not interfere with the transmission of continuous (without pauses) signals.

 Since the amplitude, polarity, and repetition rate of pulses reflecting the GO are fundamentally different from similar characteristics of the samples of the initial signal, the newly formed continuous signal with a compressed spectrum becomes completely unrecognizable. And if we take into account that the parameters of the algorithm that provides the convolution of several samples into one GO can be changed from time to time, then it becomes almost impossible for the uninitiated to extract any information from the new signal.

 In systems with modems, to compress the width of the spectrum of the signals entering the communication line and / or to ensure the secrecy of the transmitted information, it is sufficient to apply the proposed method to modulating signals. In this case, the degree of compression of the spectrum is determined not only by the parameters of the algorithm for generating GOs, but also by the type of modulation (manipulation).

 The proposed method without any restrictions is compatible with other known methods of reducing the frequency band of signals (elimination of redundancy, linear prediction, multi-position coding, the use of speech-converting devices, etc.) and can be used as a means of increasing the efficiency of these methods.

=[H_i-1+C(1+2y_i)] (|H_i| модуль Н_i; ширина каждой области равна Δ == 2С(1+2у_min); центр области соответствует величине С(1+2у_i); число областей равно

r_i - число двоичных разрядов в кодовой комбинации, которую отображает у_i) и несущее полную или частичную информацию о параметрах (знаке и амплитуде) одного или нескольких квантованных отсчетов j-й группы исходного сигнала (или другую информацию); на каждом i-м этапе формирования ГО_j у_i может принимать одно из М значений (позиций); различные значения у_i от y_min до у_max изменяются с одинаковым шагом δ, например у_i={ 0,1,2,3,4,5,6,7} (здесь δ= 1; М=8; r₁=log₂8=3) или y_i={0,5; 2,5; 4,5; 6,5} (здесь δ= 2; M=4; r₁=log₂4=2); с учетом х_i комбинация {х_i, у_i} несет информацию, представленную r_1i=(r₁+1) двоичными разрядами;
m - число этапов формирования ГО_j; максимальное (без перекрытия областей |H_i|/k, которым принадлежат различные значения y_i) значение k равно
k_max=(1+2y_min)/2(1+y_min+y_max), (*)
где у_min и у_max соответственно - минимальное и максимальное значения у_i;

по полученным с помощью (2) численным значениям х_i и у_i в обратной последовательности восстанавливаются кодовые комбинации КК_i,j, ГКК_j, и КК_отсj,k, отсчеты ИO'_j1, ИО'_j2,..., ИО'_j,n (j=1,2,3,...) и сигнал ИС', отличающийся от исходного сигнала ИС тем, что он формируется из квантованных отсчетов.To implement the convolution of each j-th (j = 1,2,3, ...) groups of samples EUT _j1, IO _j2, ..., AI _{j, n)} of the original signal (IP) in one group count _j is provided a method of GO , the essence of which is that all the original samples (or all, except the first samples in the groups IO _j1 ) are quantized by levels and are represented by binary code combinations KK _{otj, k} (k = 1,2, ... n), forming in aggregate group code combination GKK _j , subsequently divided into parts KK _{i, j} , where i is the number of the stage of formation of GO _j , i = 1,2, ..., m; each of the code combinations KK _{i, j} , taking into account its numerical value, is associated with an individual combination of information numbers x _i and y _i used in the recurrence algorithms (1) and (2) of m-stage processes of formation and separation of GO _j , establishing a one-to-one correspondence of parameters GO _j and quantized samples IO ' _j1 , IO' _j2 , ..., IO ' _{j, n} :
H _i = x _i k [H _i-1 + C (1 + 2y _i ], i = 1,2, ..., m, (1)
where H _i is the intermediate value of GO _j ; H _m is the final value of GO _j ;
x _i - binary number ± 1, which displays the characters "1", "0" and carries part of the information about the amplitude and (or) sign of the next sample of the original signal or other information;
k is a positive fractional decimal number (<1), which plays the role of a normalizing coefficient to limit the range of changes in the values of H _i ;
Н ₀ - (Н _i-1 at i = 1) - the value of the 1st sample in the group (ИО _j1 ) or the verification number, which is used to assess the correctness of restoration of samples (ИО ' _j1 , ИО' _j2 , ..., ИО ' _{j, n} );
С - normalizing coefficient (positive decimal number), its purpose is to make the value [H _i-1 + С (1 + 2y _i )] always positive;
y _i is an integer or fractional decimal number indicating a particular range of values

= [H _i-1 + C (1 + 2y _i )] (| H _i | module Н _i ; the width of each region is Δ == 2С (1 + 2у _min ); the center of the region corresponds to the value С (1 + 2у _i ) ; the number of areas is equal

r _i is the number of binary digits in the code combination, which is displayed by _i ) and carrying full or partial information about the parameters (sign and amplitude) of one or more quantized samples of the jth group of the original signal (or other information); at each i-th stage of formation of GO, _{j i} can take one of M values (positions); different values of _i from y _min to y _max vary with the same step δ, for example, _i = {0,1,2,3,4,5,6,7} (here δ = 1; M = 8; r ₁ = log ₂ 8 = 3) or y _i = {0.5; 2.5; 4,5; 6.5} (here δ = 2; M = 4; r ₁ = log ₂ 4 = 2); taking into account x _{i, the} combination {x _i , y _i } carries the information represented by r _1i = (r ₁ +1) binary digits;
m is the number of stages of formation of GO _j ; maximum (without overlapping areas | H _i | / k to which various values of y _i belong) the value of k is
k _max = (1 + 2y _min ) / 2 (1 + y _min + y _max ), (*)
where y _min and y _max, respectively - the minimum and maximum values of _i ;

using the numerical values of x _i and y _i obtained using (2), the code combinations KK _{i, j} , GKK _j , and KK _{ot j, k} , samples OO _j1 , OO _j2 , ..., OO _{j , n} (j = 1,2,3, ...) and the IS signal ', which differs from the original IS signal in that it is formed from quantized samples.

Using algorithm (1) assumes that the original signal is limited in amplitude (this condition is satisfied in almost all systems, both single-channel and multi-channel). Moreover, the value of C (1 + 2y _min ) can, in particular, be chosen equal to the accepted (actual) level of the amplitude limitation. In this case, the dynamic range of the signal after compression of the spectrum will practically remain the same.

Notes:
1) in addition to compressing the spectrum and ensuring the secrecy of message transmission, the information parameters x _i and y _i and the corresponding binary digits or their combinations can be used as service messages (for example, to increase the accuracy of recovery, service communication, channel testing), to organize multi-channel transmission from several sources, etc. ;
2) the formation and restoration of the values of N _i can be done either by analog computing devices (AVU) or digital (CV), for example, using a microprocessor, but in the latter case, the values of N _i must be represented as decimal numbers with a maximum number of decimal places (digits) ;
3) ie. To the presence of distortions and noise in the channel in the last stages of the recovery H _i to errors in the determination of x _i and y _i, for reducing the significance of these errors in the formation of H _i recommended first "pack" LSBs all merged sample, and then - senior, using interleaving discharges; in this case, with the correct restoration of the high-order bits, all restored samples will be distorted approximately equally and to a lesser extent than without spectrum compression, since noise power is directly proportional to the channel frequency band, and when the spectrum is compressed, this band can be correspondingly reduced.

Example 1 - compression of the spectrum by 2 times, i.e. the number of _j samples combined into one GO is equal to two; channel distortion and interference are negligible; the number of quantization levels of the signal IP '256; the number of binary digits (dv) per one IP sample '1 = log ₂ 256 = 8; the values of _{i i} = {0.5; 2.5; 4,5; 6.5} (log ₂ 4 = 2, that is, at each stage of the formation of GO _{j, the} combination {x _i , y _i } can display ("pack" in GO _j ) a maximum of 3 bits); in each civil defense _j “official” 1 service dv. R. ; C = 4; the number of stages of the formation of GO _j is 6: at 5 stages, “3 steps" are "introduced" and at stage 1 2 dv (total 17 dv; if necessary, this combination can be recoded); at 5 stages (in accordance with (*)) k _max1 = (1 + 2y _min ) / 2 (1 + y _min + y _max ) = (1 + 2 • 0.5) / 2 (1 + 0.5 + 6.5) = 1/8 (k ₁ = 1/10 is assumed); at one stage (for 2 doors) k _max2 = (1 + 2 • 0.5) / 2 (1 + 0.5 + +2.5) = 2/8 = 1/4 (k ₂ = k _max2 = 1/4); H ₀ (H _i-1 for i = 1) is taken equal to +6 and is used as a verification number, according to which the correctness of recovery is checked.

 Tables 1 and 2 show the information that makes up the database necessary for the formation and separation of any civil defense, in relation to the above conditions of example 1.

The specific group code combination (GKK _j ) adopted for a certain j-th group, representing IO ' _j1 , IO' _j2 and 1 service rank (r.r.), as well as dividing this GKK _j into parts (KK _{i, j} ), used at individual stages of the formation of GO _j , and selected for each such part of the HCC _j from Table 1 numbers x _i and y _i are given in table 3.

The stages of formation of GO _j in accordance with (1):
1. H ₁ = x ₁ [H ₀ + C (1 + 2y ₁ )] k ₁ = + [+ 6 + 4 (1 + 2 • 4,5)] / 10 = + 4,6;
2. Н ₂ = x ₂ [Н ₁ + С (1 + 2у ₂ )] k ₁ = - [+ 4,6 + 4 (1 + 2 • 6,5)] / 10 = -6,06;
3. Н ₃ = x ₃ [Н ₂ + С (1 + 2у ₃ )] k ₁ = + [- 6.06 + 4 (1 + 2 • 2.5)] / 10 = + 1.794;
4. H ₄ = x ₄ [H ₃ + C (1 + 2y ₄ )] k ₁ = - [+ 1,794 + 4 (1 + 2 • 4,5)] / 10 = -4,1794;
5. H ₅ = x ₅ [H ₄ + C (1 + 2y ₅ )] k ₁ = - [- 4.1794 + 4 (1 + 2 • 6.5)] / 10 = -5.18206;
6. H ₆ = x ₆ [H ₅ + C (1 + 2y ₆ )] k ₂ = - [- 5.18206 + 4 (1 + 2 • 0.5)] / 4 = -0.704485.

Thus, GO _j , equivalent to the selected specific code combinations for IO ' _j1 , I0' _j2 and one service bit ("1"), corresponds to a value of (-0.704485). Other GOs _j are formed similarly.

The stages of restoring the values of x _i , y _i and H ₀ (in accordance with the above algorithm (2)):
1.1) H ₆ = -0.704485⇒x ₆ = -1, | H ₆ | = 0.704485; 2) | H ₆ | / k ₂ = 0.704485 • 4 = 2.81794; 3) from Table 2 it follows that | H ₆ | / k ₂ = 2.81794 belongs to the region {0 .. .16} ⇒y ₆ = 0.5, C (1 + 2y ₆ ) = 8; 4) H ₅ = | H ₆ | / k ₂ -C (1 + 2y ₆ ) = 2.81794-8 = -5.18206;
2. H ₅ = -5.18206⇒x ₅ = -1, | H ₅ | = 5.18206⇒ | H ₅ | / k ₁ = 5.18206 • 10 = 51.8206⇒ {48 ... 64} ⇒y ₅ = 6.5, C (1 + 2y ₅ ) = 56⇒H ₄ = 51.8206-56 = -4.1794;
3. H ₄ = -4.1794⇒x ₄ = -1, | H ₄ | = 4.1794⇒ | H ₄ | / k ₁ = 4.1794 • 10 = 41.794⇒ {32 ... 48} ⇒y ₄ = 4.5, C (1 + 2y ₄ ) = 40⇒ H ₃ = 41.794 -40 = + 1.794;
4. H ₃ = + 1,794⇒x ₃ = + 1, | H ₃ | = 1,794⇒ | H ₃ | / k ₁ = 1,794 • 10 = 17.94⇒ {16 ... 32} ⇒y ₃ = 2.5, C (1 + 2y ₃ ) = 24⇒H ₂ = 17.94 -24 = -6.06;
5. H ₂ = -6.06 x x ₂ = -1, | H ₂ | = 6.06⇒ | H ₂ | / k ₁ = 6.06 • 10 = 60.6⇒ {48 ... 64} ⇒y ₂ = 6.5, C (1 + 2y ₂ ) = 56⇒H ₁ = 60.6-56 = + 4.6;
6. H ₁ = + 4,6⇒x ₁ = +1, | H ₁ | = 4.6⇒ | H ₁ | / k ₁ = 4.6 • 10 = 46⇒ {32 ... 48} ⇒y ₁ = 4.5, C (l + 2y ₁ ) = 40⇒N ₀ = 46 -40 = + 6.

Thus, the results of the restoration of x _i , y _i and H ₀ coincide with the original data, according to the obtained values of x _i and y _i in accordance with table. 1, code combinations (CC _{i, j} ) are determined from 3 and 2 digits (the number of digits r _1i for any stage is determined by the initial conditions), two 8-bit combinations are formed from them, displaying IO ' _j1 and IO' _j2 , and the service discharge is fixed. According to the code combinations for IE _'j1 and IO' _j2 restored quantized values IO _'j1 and IO' _j2. Other groups from IO ' _j1 and IO' _j2 (j = 1,2,3, ..,) are similarly restored, and then IS '.

в табл.7.Example 2 - spectrum compression 3 times; the influence of distortion and interference in the channel cannot be neglected; H ₀ = IO _j1 = +2; the number of levels of quantization IP '128; number of doors R. for 1 count of the IC 'l = log ₂ 128 = 7, while 6 digits indicate the value of the count, and 1 digit - its sign (-1⇒ "0", + 1⇒ "1"); y _i values = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15} (log ₂ 16 = 4, taking into account х _i on each at the stage of formation of GO _{j, the} combination {x _i , y _i } can display a maximum of 5 doors); the correspondence of x _i , y _i and code combinations (QC _{i, j} ) - in table 4; C = 10; the number of stages of formation of GO _j is 3: at 2 stages, “5 steps” are “introduced” and at one stage 4 doors. R. ; at 2 stages (in accordance with (*)) k _max1 = (1 + 2u _min ) / 2 (1 + at _min + at _max ) = (1 + 2 • 0) / 2 (1 + 0 + 15) = 1/32 (k ₁ = 1/32 is assumed); at one stage k _max2 = (2 • 1 + 0) / 2 (1 + 0 + 7) = 1/16 ( taken k ₂ = 1/16); Group assignment of bits in a codeword (KG _j) (for AI _'j2 and IO' _j3) and received their specific values - in Table 5; the redistribution of discharges in the MCC, taking into account recommendation 3 of the Notes, the separation of discharges by stages and the specific values of x _i and y _i (in accordance with Table 4) are given in Table 6; borders and centers of regions

in table 7.

The stages of formation of GO _about in accordance with (1):
1. H ₁ = x ₁ [H ₀ + C (1 + 2y ₁ )] k ₁ = - [+ 2 + 10 (1 + 2 • 11)] / 32 = -7.25;
2. H ₂ = x ₂ [H ₁ + C (l + 2y ₂ )] k ₁ = - [- 7.25 + 10 (1 + 2 • 6)] / 32 = -3.8359375;
3. Н ₃ = x ₃ [Н ₂ + С (1 + 2у ₃ )] k ₂ = - [- 3,8359375 + 10 (1 + 2 • 2)] / 16 = -2,8852539.

Thus, GO _j , equivalent to the selected specific code combinations for IO ' _j2 and IO' _o3 , corresponds to a value of (-2.8852539). Other GOs _j are formed similarly.

Sources of information
1. Kharkevich A.A. Selected Works, vol. 2. M .: Nauka, 1973.

 2. GB 1,438,716 A, H 04 B 11/66, 1976.

Claims (4)

 1. A method of compressing the width of the spectrum of information electric signals with a limited frequency band, in which the original signals are sampled by time, artificially increase the clock intervals between discrete samples so many times, how many times reduce the frequency band of the original signal, and limit the spectrum to this resulting reduced frequency band a sequence of discrete samples with increased clock intervals, characterized in that the clock intervals between discrete samples increase rapidly th coagulation samples of each group into which all share the initial sequence of signal samples in one group equivalent count pulse, the magnitude of the amplitude and polarity of which together uniquely represent mutually same parameters raw counts - counts corresponding group. 2. The method according to p. 1, characterized in that for the implementation of the convolution of each j-th (j = 1,2,3,...) Group of samples of the original signal IO _jl , IO _j2 ,. . . , IO _{j, n} into one group sample of GO _j, all the original samples in the group, or all but the first sample of IO _jl , are quantized by levels and represented with binary code combinations of the QC _{j, k} (k = 1,2, ..., n) , forming in aggregate the group code combination GKK _j from which m-stage transformations form GO _j , for which the GKK _j are divided into code combinations KK _ij (i = 1,2, ..., m) and each new one of these combinations match the individual combination of information numbers x _i and y _i used in the recurrence algorithms (1) and (2) m - stage processes the formation and separation of GO _j , establishing a one-to-one correspondence between the parameters GO _j and the quantized samples IO ' _jl , IO' _j2 ,. . . , OO ' _{j, n} . H _i = x _i k [H _i-1 + C (1 + 2y _i )], i = 1,2,. . . , m (1)
where H _i is the intermediate value of GO _j ;
H _m - the final value of GO _j ;
x _i is a binary number ± 1, which, together with y _i, displays the code combination of QC _ij ;
k is a positive fractional decimal number (<1) that plays the role of a normalizing coefficient to limit the range of variation of the values of H _i ;
Н ₀ - (Н _i-1 at i = 1) - the value of the 1st sample in the group of ИО _jl or the verification number, which is used to assess the correctness of the restoration of the samples ИО ' _jl , ИO' _j2 ,. . . , OO ' _{j, n} ;
C - normalizing coefficient - a positive decimal number, its purpose is to make the value [H _i-1 + C (1 + 2y _i )] always positive;
y _i is an integer or fractional decimal number indicating a particular range of values

= [H _i-1 + C (1 + 2y _i )], where | H _i | - module H _i ; the width of each region is Δ == 2C (1 + 2y _min ); the center of the region corresponds to the value of C (1 + 2y _i ); the number of areas is equal

where r _i is the number of bits in the code combination that y _i displays; at each i-th stage of the formation of GO _j y _i can take one of M values; different values of _{i i} from y _min to y _max vary with the same step, taking into account x _{i the} combination {x _i , y _i } carries the information represented by r _li = (r _i +1) binary digits;
m is the number of stages of formation of GO _j ;
maximum without overlapping areas

to which different values of y _i belong, the value of k is
k _max = (1 + 2y _min ) / 2 (1 + y _min + y _max ),
where _min and _max are respectively the minimum and maximum values of _i ;

using the numerical values of x _i and y _i obtained using (2), the code combinations KK _ij , GKK _j and KK _{otj, k} , quantized impulse samples IO ' _jl , IO' _j2 , are restored in the reverse sequence. . . , ИО ' _{j, n} , from which a signal is formed with the initial value of the frequency band. 3. The method according to p. 2, characterized in that if it is necessary to transmit over the channel, in addition to the main signals, the spectrum of which is compressed, service information and / or messages from other sources, this additional information is represented by binary characters, which are included in the group code combinations of the CCM _j and process along with binary symbols representing the main signals. 4. The method according to p. 2, characterized in that if there is distortion and / or interference in the transmission channel of the QC code combinations, the _{j, k} (k = 1,2, ..., n) are changed to the reverse order of binary bits and during the formation of group code combinations GKK _j , binary bits are interleaved from the converted KK combinations _{otjj, k} (k = 1,2, ..., n), and in the process of reconstructing signals with the original frequency band, the inverse transforms are performed.

Images