RU2486683C1 - Method of searching for noise-like signals with minimum frequency-shift keying - Google Patents

Abstract

FIELD: radio engineering, communication.

SUBSTANCE: on each search cycle, the input signal is split into quadrature components which are integrated on intervals equal to duration of the code sequence element, and the integration results are stored over a time interval equal to the signal repetition period. N values of the modulus of the cross-correlation function are calculated in real time with clock frequency selected based on the condition of enabling calculation of N values of the modulus of the cross-correlation function in real time. Results of integrating the elements are multiplied with elements of reference quadrature code sequences. Values of quadrature components of the cross-correlation function are calculated by integrating N combined multiplication results. The greatest of the N values of the modulus of the cross-correlation function is selected; code sequence delay is estimated; the code generator is synchronised with the received noise-like signal with minimum frequency-shift keying.

EFFECT: faster search for noise-like signals with high noise-immunity of reception and low hardware costs.

2 dwg

Description

The invention relates to the field of radio engineering and can be used in radio navigation and radio communication systems for code synchronization of receivers of noise-like signals with minimal frequency manipulation.

A known method for the parallel search for noise-like signals by the delay time [Radio systems / Yu.P. Grishin, V. P. Ipatov, Yu. M. Kazarinov and others. Ed. Yu.M. Kazarinova. - M .: Higher school, 1990, p.64 (Fig.3.16), p.99 (Fig.4.6)], which consists in multiplying the received signal with N pairs of reference noise-like signals, which are quadrature copies of the received signal for N discrete time values delays, integrating the results of multiplication in 2N quadrature channels over an interval equal to the duration of a noise-like signal, extracting N values of the module of the mutual correlation function (VKF) and choosing the delay signal of the reference signal in the channel with the maximum value B as the estimate of the delay time CF

The search method provides a potentially achievable noise immunity, the shortest possible search time, however, it is difficult to implement with the number of channels N >> 1.

A known method for the rapid search for noise-like signal (SHPS) [Patent RU No. 2206180, MKI H04L 7/10, publ. 06/10/2003], which consists in sequentially evaluating the characters of a pseudorandom sequence (PSP), which allows one to synthesize a signal with the required delay in any receiving unit of a PSP with a length of m characters. M estimates of the received binary symbols of the pseudo-random sequence are recorded in the register of the reference generator of the NPS of the correlation storage and then the affiliation of this m-valued combination of symbols with the desired noise-like signal is checked. If the received m-digit combination of characters does not belong to a noise-like signal (false alarm), then the register is reset and filled again with the next character estimates of the pseudorandom sequence of the desired noise-like signal for subsequent verification using a correlation accumulator.

The disadvantage of the quick search method is its low noise immunity.

A known method of searching for noise-like signals with minimal frequency manipulation [Patent RU No. 2420005, IPC H04L 27/14, publ. 05/27/2011], which consists in the fact that at each search cycle, samples of the quadrature components of the input noise-like signal are accumulated at clock intervals, they are stored, 4N samples of the quadrature components of the input noise-like signal z _1k and z _2k are generated, a two-step sign approximation of the shape of the quadrature reference video signals is performed I _k and Q _k, pairwise combined results of multiplying the quadrature component z _1k and z _2k samples approximated with quadrature reference video signals I _k and Q _k, calculated 2N samples vadraturnyh components of the cross-correlation function, isolated N values of this function module, determining a delay value of the element code sequence for setting synchronism code generator with the received signal.

The disadvantage of this method is the complexity of the implementation and a significant amount of computational operations. In addition, the known search method is not applicable in the presence of additional digital modulation of a noise-like signal.

The closest technical solution to the claimed invention is a method of searching for noise-like signals with minimal frequency manipulation. [Patent RU No. 2353064, MKI H04L 27/14, publ. 04/20/2009], which consists in the fact that at each search cycle, an input noise-like signal is divided into quadrature components, sampling, digitizing and integrating quadrature components at clock intervals m times shorter than the length of the code sequence element. The integration results are stored for a time equal to the duration of the code sequence element, during which they are multiplied with samples of the corresponding reference noise-like signals generated with a frequency M times the clock frequency of the input signal code sequence. Moreover, for a time equal to one repetition period of the code sequence, M module values of the cross-correlation function are generated, which are then accumulated over an interval equal to a fixed number of repetition periods of the code sequence. The decision on the value of the delay time of the input signal is made by choosing the maximum of the M values of the accumulated modules of the cross-correlation function, storing the maximum value, its address and cycle number and repeating the search procedure a fixed number of times with a shift by M elements of the sequence when switching to each subsequent search cycle.

The disadvantage of this method is the significant search time with a large base noise-like signal (pseudo-random sequence length N >> 1), due to the need for repeated repetition of the search procedure.

The present invention is intended to solve the problem of reducing the search time for a noise-like signal with minimal frequency manipulation with high noise immunity and low hardware costs.

The problem is solved in that in a method for searching for a noise-like signal with minimal frequency manipulation, in which the input noise-like signal is divided into quadrature components by multiplying it with reference harmonic signals of the carrier frequency shifted relative to each other by a phase angle π / 2, integration of quadrature elements components, remember the results of the integration of the elements of the quadrature components, form the reference quadrature video signals, calculate N values of the module of the cross-correlation function of the input and reference noise-like signals, select the maximum of N values of the module of the cross-correlation function, estimate the delay of the code sequence corresponding to the maximum value of the module of the cross-correlation function, set the code generator in a state of synchronism with the received noise-like signal using the estimate of the delay of the code sequence , according to the invention, on each of n periods of repetition of a noise-like signal, starting from the second, in each channel e search form a periodic code sequence with a length of N elements and reference quadrature code sequences by sign approximating the samples of the reference quadrature video frequency signals, multiply the results of integration of the elements of the quadrature components with the elements of the corresponding reference quadrature code sequences, combine the results of multiplying the integrally accumulated quadrature components and the elements of the corresponding reference quadrature code followed values, calculate the values of the quadrature components of the cross-correlation function by integrating N combined results of multiplication, calculate the module of the cross-correlation function, accumulate the values of the module of the cross-correlation function for n periods of repetition of a noise-like signal, and the integration of the elements of the quadrature components is carried out at intervals equal to the duration of the element of the periodic code sequence , the results of integrating the elements of quadrature components are stored on the inter a time shaft equal to the period of repetition of a noise-like signal, and the formation of a periodic code sequence, readings of reference quadrature code sequences and reading the results of integration of elements of quadrature components are performed with a clock frequency M times the clock frequency of the periodic code sequence of the input noise-like signal.

Figure 1 shows a diagram of a device for implementing the proposed method for searching for noise-like signals with minimal frequency manipulation, and figure 2 is a block diagram of an algorithm for searching for noise-like signals with minimal frequency manipulation.

A device for implementing a method for searching for noise-like signals with minimal frequency manipulation contains the first and second multipliers 1 ₁ and 1 ₂ , the signal inputs of which are combined and are the input of the device, and the reference input of each is connected to the corresponding output of the reference generator 2. To the outputs of the multipliers 1 ₁ and 1 ₂ , the inputs of the first and second integrators 3 ₁ and 3 ₂ , respectively, are connected, the outputs of which are connected to the information inputs of the first and second random access memory devices 4 ₁ and 4 _2, respectively. The outputs of the storage devices 4 ₁ and 4 _{2 are} connected to pairwise combined signal inputs of the third 1 ₃ and fifth 1 ₅ and fourth 1 ₄ and sixth 1 ₆ multipliers, respectively. The outputs of the third and sixth (1 ₃ , 1 ₆ ) and fourth and fifth (1 ₄ , 1 ₅ ) multipliers are combined, respectively, through a subtractor 5 and a first adder 6 ₁ , to the outputs of which are connected the corresponding inputs of the module 7 for generating the cross-correlation function modules. This unit 7 contains the third and fourth integrators 3 ₃ and 3 ₄ , the first and second quadrators 8 ₁ and 8 ₂ , as well as the second adder 6 ₂ connected in series, the square root extraction element 9, the fifth integrator 3 ₅ and the third random access memory 4 ₃ . The output of random access memory 4 _{3 is} connected to the input of the decision unit 10, the output of which is connected in series to the control unit 11, the controlled delay element 12, the code generator 13 and the synthesizer 14 samples. Moreover, the clocked input of the controlled delay element 12 is combined with the clocked input of the synthesizer 14 samples and the clock inputs of the first and second random access memory 4 ₁ and 4 ₂ and connected to the first output of the block 15 forming time intervals. The synchronizing inputs of the first and second integrators 3 ₁ and 3 ₂ are interconnected and connected to the second output of the block 15 forming time intervals, the synchronizing inputs of the third 3 ₃ , fourth 3 ₄ and fifth 3 ₅ integrators and the third random access memory 4 ₃ are interconnected and connected to the third output of block 15 forming time intervals. The clock input of the control unit 11 is connected to the fourth output of the time interval generating unit 15. The reference inputs of the third 1 ₃ and fourth 1 ₄ and fifth 1 ₅ and sixth 1 ₆ multipliers are pairwise connected to each other and to the outputs of the synthesizer 14 samples, respectively, through the symbolic elements 16 ₁ and 16 ₂ .

The output of the code generator 13, connected to the control input of the synthesizer 14 samples, is the output of the device for searching for noise-like signals with minimal frequency manipulation.

The method of searching for noise-like signals is as follows. At the input of the search device (Fig. 1), after sampling and digitization _, samples of the received periodic noise-like signal with minimal frequency manipulation are received with a step T _d :

$$\begin{array}{l}s(t_i)=D(t_i-\tau )\cos \left[{\omega }_0{t}_i+\Theta (t_i-\tau )-\varphi \right]=\\ =D(t_i-\tau )\left[I(t_i-\tau )\cos ({\omega }_0{t}_i-\varphi )-Q(t_i-\tau )\sin ({\omega }_0{t}_i-\varphi )\right],\\ I(t_i)=\cos \Theta (t_i),Q(t_i)=\sin \Theta (t_i),i=\mathrm{1,2}...,\text{(one)}\\ \Theta \text{(t)}=\frac{\pi }{\text{2T}}{\displaystyle \underset{0}{\overset{t}{\int }}d(t^{\prime })d{t}^{\prime },d(t)={\displaystyle \sum _{j=0}^{N-\mathrm{one}}{d}_{j}rect(t-jT)}}\end{array}$$

where ω ₀ is the average frequency;

τ is the delay time;

φ is the initial phase (the amplitude is assumed to be unity);

Θ (t) is the function that determines the law of angular modulation;

I (t _i ) and Q (t _i ) - quadrature video frequency ShPS;

t _i = iT _d , i = 1, 2, ... are the sampling times of the input and reference signals;

D (t) is the function that determines the law of additional digital modulation;

d (t) is a binary signal corresponding to a code pseudo-random sequence d ₀ , d ₁ , ..., d _N-1 ;

T is the duration of the SRP element;

rect (t) - rectangular function (pulse of unit amplitude and duration T);

N - the length of the SRP, which determines the period T _p = NT repetition of SHPS.

Input multipliers 1 ₁ and 1 _{2 have} made multiplying the signal samples (1) to the carrier frequency counts of a reference signal cosω ₀ t _i and sinω ₀ t _i, are generated by the reference oscillator 2. At the outputs of multipliers 1 ₁ and 1 ₂ form the input signal samples constituting respectively videochastotnyh :

$$\begin{array}{c}{x}_i=\frac{\mathrm{one}}{2}\left[I(t_i-t)\cos \varphi -Q(t_i-t)\sin \varphi \right],\\ y_i=\frac{\mathrm{one}}{2}\left[I(t_i-t)\sin \varphi +Q(t_i-t)\cos \varphi \right],\end{array}\text{(2)}$$

as well as frequency components 2ω ₀ , which are filtered out by the subsequent processing path.

The samples of the video-frequency components (2) of the input signal x _i and y _i are received at the signal inputs of the integrators 3 ₁ and 3 _2, respectively, at the outputs of which the values are formed:

$$X_j={\displaystyle \sum _i{x}_i,Y_j={\displaystyle \sum _i{y}_i}.\text{(3)}}$$

The summation over i in (3) is carried out according to m = T / T _d samples within each j-th element of the received SHPS: j = 1, 2, ....

The gating of integrators 3 ₁ , 3 _{2 is} carried out by clock pulses generated by the block 15 of the formation of time intervals.

The results of bitwise processing (3), obtained at intervals equal to the duration T of the SRP element, are stored in random access memory 4 ₁ and 4 ₂ , after which the integrators 3 ₁ , 3 ₂ are reset and the next element of the NPS is integrated with a duration of T. Record the results of integration into random access memory devices 4 ₁ , 4 ₂ and reset of integrators 3 ₁ , 3 _{2 are} performed with a clock frequency of the memory bandwidth f _t = 1 / T.

The results of bitwise processing (3), arrays of values {X _j } and {Y _j } of volume N each, obtained during the observation time equal to the repetition period of the SHPS, are stored in random access memory 4 ₁ and 4 ₂ until the end of the subsequent repetition period of the ShPS.

At each SHPS repetition period, starting from the second, the synthesizer 14 samples generates samples of the reference quadrature video frequency signals I (t _j ) and Q (t _j ), which are received at the inputs of the symbol elements 16 ₁ and 16 _2, respectively. In multipliers 1 ₃ , 1 ₄ , 1 ₅ , 1 ₆ , the results of (3) bitwise processing stored in memory devices 4 ₁ and 4 ₂ are multiplied by elements C _j = sign (I (t _j )) and S _j = sign (Q (t _j )) reference quadrature code sequences formed by the symbolic elements 16 ₁ and 16 ₂ . The operation sign (•) corresponds to the sign approximation of the reference quadrature video frequency signals I (t _j ) and Q (t _j ). Elements C _j are supplied to the reference inputs of multipliers 1 ₃ and 1 ₄ , and elements S _j are supplied to the reference inputs of multipliers 1 ₅ and 1 ₆ .

The frequency of data from the outputs of random access memory devices 4 ₁ and 4 ₂ to the corresponding inputs of multipliers 1 ₃ , 1 ₅ and 1 ₄ , 1 ₆ is determined by the clock frequency Mf _T , which is selected from the condition that it is possible to calculate N values of the module of the cross-correlation function in real time time: in the first approximation, M ≈ 4N. With the same frequency Mf _T , reference quadrature code sequences {C _j } and {S _j } are formed.

The signals from the outputs of the multipliers 1 ₃ and 1 ₆ , 1 ₄ and 1 ₅ are combined in pairs in the subtractor 5 and the adder 6 ₁ , forming elementwise quadrature correlations. The latter arrive at the corresponding inputs of block 7 of the formation of modules of the function of cross-correlation of the received and reference SHPS. Block 7 contains integrators 3 ₃ and 3 ₄ , quadrants 8 ₁ and 8 ₂ , as well as an adder 6 ₂ , square root extraction element (module selection) 9, fifth integrator 3 ₅ and random access memory 4 ₃ for storing N values of the module of the mutual function correlations. The integrators 3 ₃ , 3 ₄ integrate at intervals equal to the period T _{n of the} repetition of the SHPS, the corresponding element-wise quadrature correlations arriving at their inputs, forming the quantities

$$\begin{array}{c}{z}_{\mathrm{one}kl}={\displaystyle \sum _{j=\mathrm{one}}^N(C_j{X}_{j+(l-\mathrm{one})N+k}-S_j{Y}_{j+(l-\mathrm{one})N+k}),}\\ z_{2kl}={\displaystyle \sum _{j=\mathrm{one}}^N(S_j{X}_{j+(l-\mathrm{one})N+k}-C_j{Y}_{j+(l-\mathrm{one})N+k}),}\end{array}\text{(four)}$$

where k is the number of the search channel: k = 0, 1, 2, ..., N-1;

l is the number of the ShPS repeating period: l = 1, 2, ..., n.

The summation over j in (4) is carried out over N samples of the corresponding quadrature components {X _j } and {Y _j }, taking into account the sign of the elements C _j = ± 1 and S _j = ± 1 within each of the n periods of the repetition of the SHPS.

The integration results (4), which are the values of the quadrature components of the cross-correlation function of each of the N search channels for each of the n SHPS repetition periods, are squared and combined in the adder 6 ₂ . The cross-correlation function module is formed at the output of the square root extraction element 9:

$$z_{kl}=\sqrt{z_{\mathrm{one}\mathrm{<figure-callout\; id="2"\; label="k\; l"\; filenames="00000011.png"\; state="\{\{state\}\}">k\; </figure-callout>}l}^2+z_{2\mathrm{<figure-callout\; id="2"\; label="k\; l"\; filenames="00000011.png"\; state="\{\{state\}\}">k\; </figure-callout>}l}^2},k=\mathrm{0,1,2}...,N-\mathrm{one};l=\mathrm{1,2}...,n.\text{(5)}$$

The integrator 3 ₅ accumulates the values of the module of the cross-correlation function (5) in each search channel:

$$Z_k={\displaystyle \sum _{i=\mathrm{one}}^n{z}_{kl}},k=\mathrm{0,1,2}...,N-\mathrm{one.}\text{(6)}$$

The number n of SHPS repetition periods in (6) is one less than the total number of SHPS repetition periods in the observation interval of the input signal.

The values of Z _{k are} stored in the random access memory 4 ₃ , forming an array of volume N. The decisive unit 10 selects the maximum value of the module of the cross-correlation function from N values:

$$Z_{\max }=\underset{k}{\max }{Z}_k,k=\mathrm{0,1,2}...,N-\mathrm{one.}\text{(7)}$$

принятого ШПС относительно временной шкалы, задаваемой блоком 15 формирования временных интервалов. Эта оценка используется блоком 11 управления для установки генератора 13 кода с использованием элемента 12 управляемой задержки в состояние синхронизма с принятым ШПС с точностью не хуже ±T/2 (при условии, что аномальные ошибки отсутствуют).The number k _{m of} the search channel, in which the value Z _max of the cross-correlation function module is observed, corresponds to a delay estimate $$\widehat{\tau }=k_{m}T$$

adopted SHPS relative to the timeline specified by block 15 of the formation of time intervals. This estimate is used by the control unit 11 to set the code generator 13 using the controlled delay element 12 in a state of synchronism with the received BPS with an accuracy of not worse than ± T / 2 (provided that there are no anomalous errors).

, соответствующую значению Z_max (7) модуля функции взаимной корреляции.The search algorithm (4) - (7) explains the block diagram (figure 2). The first search channel (k = 0) uses data (X ₁ , X ₂ , ..., X _nN ) and (Y ₁ , Y ₂ , ..., Y _nN ). In the block for calculating the quadrature components of the cross-correlation function, the values z _10l and z _20l (4) are calculated for each repetition period of the NPS: l = 1,2, ..., n. In the unit for calculating the module of the cross-correlation function, the module Z ₀ (6) of the cross-correlation function is obtained by summing the modules z _0l (5) of the cross-correlation function calculated for each NPS repetition period: l = 1, 2, ..., n. In the second channel (k = 1), the search procedure, including the execution of operations (4) - (6) in the mentioned blocks, is performed similarly to the first channel, differing only in that the data is used for calculations (X ₂ , X ₃ , ..., X _{nN +1} ) and (Y ₂ , Y ₃ , ..., Y _{nN + 1} ). The described procedure is repeated N times (taking into account the shift of the data sets {X _j } and {Y _j } by one position with the transition to the next channel), and in the search channel with the number k = N-1, the data is used (X _N , X _{N + 1} , ..., X _{(n + 1) N-1} ) and (Y _N , Y _{N + 1} , ..., Y _{(n + 1) N-1} ). Upon completion of the calculations in the channel with the number k = N-1, the delay determines the delay of the code sequence $$\widehat{\tau }=k_{m}T$$

corresponding to the value of Z _max (7) module of the cross-correlation function.

An example implementation of a synthesizer of 14 samples of reference quadrature signals I (t _i ) = cosΘ (t _i ) and Q (t _i ) = sinΘ (t _i ) using an accumulating adder (phase accumulator) and read-only memory for storing samples of quadrature signals is given in monographs [Digital phase synchronization systems / M.I.Zhodzishsky, S.Yu.Sila-Novitsky, V.A. Prasolov and others. Ed. M.I.Zhodzishsky. - M .: Owls. Radio, 1980, p. 55-57].

Qualitative parameters described search method characterized PNS probability P _err anomalous errors greater than a value T / 2 (in absolute value), and _search the search time t. When the code memory bandwidth is N >> 1, the problem of searching for NPS by the delay time can be reduced to the problem of recognizing N orthogonal signals, with respect to which the error probability can be estimated as [L.E. Varakin. Theory of signal systems. - M .: Owls. Radio, 1978, p. 60 (f-la (2.34))]:

$$P_{\mathrm{about}w}\cong N\left[\mathrm{one}-F\left(\raisebox{1ex}{$q$}\!\left/ \!\raisebox{-1ex}{$\sqrt{2}$}\right.\right)\right],\text{(8)}$$

where Ф (х) is the probability integral, q is the signal-to-noise ratio at the output of the "synchronous" channel (with a relative delay of the received and reference signals τ = 0). Formula (8) is written under the assumption that the number of NPS repetition periods is n >> 1 (this allows us to approximate the distribution of the output quantity by the normal distribution).

The loss in the signal-to-noise ratio due to the sign approximation of the reference quadrature signals I (t) and Q (t) in comparison with the optimal correlation processing does not exceed 1 dB [V.N. Bondarenko / Optimal algorithm for searching for a noise-like signal with minimal frequency manipulation. - M., "Radio Engineering and Electronics", 2008, vol. 53, No. 2. S.222-229], i.e. the proposed search method provides noise immunity close to potentially achievable.

Due to the use of the sign approximation of the reference quadrature signals, the operations of multiplication in calculating the cross-correlation function are replaced by the operations of addition and subtraction. This reduces computational costs and computation time. The search time for the proposed method is fixed and is determined by the accumulation time required to ensure a given probability of the correct completion of the search:

t _search = (n + 1) T _p .

When implementing a device for implementing a method for searching for noise-like signals with minimal frequency manipulation in real time with the number of NPS repetition periods n = 25, the bandwidth N = 16383, the NPS repetition period T _p = 40 ms, the search time is t _search ≈ (n + 1 ) T _p ≈1 s, whereas when using the multi-channel search method with the number of channels K = 100, it is t _search = NnT _p / K≈160 s. If the element base is not fast enough, the search time exceeds the value (n + 1) T _p for the post-processing time. For example, when using programmable logic integrated circuits of the Xilinx ™ Virtex-4 ^® series [Knyshev, D.A. FPGA of the company "XILINX": Description of the structure of the main families / D.A. Knyshev, M.O. Kuzelin. - M .: Dodeka XXI, 2007. - 238 p.] With the maximum possible clock frequency f _tm = 400 MHz with the same signal parameters, the search time is t _search ≈T _p + N ² / f _tm ≈17 s.

Thus, the proposed method for the search for SHPS with minimal frequency manipulation allows almost 10 times or more to reduce the search time compared to the prototype method. This is the technical and economic effect in comparison with the known methods for searching for noise-like signals.

Claims (1)

A method for searching for noise-like signals with minimal frequency manipulation, which consists in dividing the input noise-like signal into quadrature components by multiplying it with reference harmonic signals of the carrier frequency shifted relative to each other by the phase angle π / 2, integrating the elements of the quadrature components, remember the results integrating elements of quadrature components, form reference quadrature video frequency signals, calculate N values of the function module in the correlation of the input and reference noise-like signals, select the maximum of the N values of the cross-correlation function module, estimate the code sequence delay corresponding to the maximum value of the cross-correlation function module, set the code generator in the synchronism state with the received noise-like signal using the code sequence delay estimate, characterized in that on each of n periods of repetition of a noise-like signal, starting from the second, a period is formed in each search channel a code sequence with a length of N elements and reference quadrature code sequences by sign approximation of samples of reference quadrature video frequency signals, multiply the results of integration of elements of quadrature components with elements of the corresponding reference quadrature code sequences, combine the results of multiplying integrally accumulated quadrature components and elements of the corresponding reference quadrature code sequences in pairs calculate the value the quadrature components of the cross-correlation function by integrating N combined multiplication results, calculate the cross-correlation function module, accumulate the cross-correlation function module values for n periods of the noise-like signal repetition, and the elements of the quadrature components are integrated at intervals equal to the duration of the periodic code sequence element, the integration results elements of quadrature components are stored for a time interval equal to a period the repetition of a noise-like signal, and the formation of a periodic code sequence, readings of the reference quadrature code sequences and reading the results of the integration of the elements of the quadrature components is performed with a clock frequency M times the clock frequency of the periodic code sequence of the input noise-like signal.

Images