CN111083082B

CN111083082B - Multiple synchronization method of FHSS-GMSK system

Info

Publication number: CN111083082B
Application number: CN201910754957.8A
Authority: CN
Inventors: 熊军; 谢琦; 郭晓峰; 王旭
Original assignee: Xi'an Yufei Electronic Technology Co ltd
Current assignee: XI'AN YUFEI ELECTRONIC TECHNOLOGY Co.,Ltd.
Priority date: 2019-08-15
Filing date: 2019-08-15
Publication date: 2022-03-29
Anticipated expiration: 2039-08-15
Also published as: CN111083082A

Abstract

The invention relates to the technical field of wireless communication, in particular to a multiple synchronization method of an FHSS-GMSK system; the method comprises the following steps: s1 sets three-stage synchronous jump structure for GMSK signal; s2, independently adjusting GMSK signals; s3, synchronously processing the GMSK signal in S2 and calculating frequency offset; s4, synchronizing the signals in each hop for 2 times to complete the final synchronization in each hop; the invention utilizes multiple synchronization, and can quickly and accurately synchronize and correct frequency offset; and an independent GMSK modulation mode is adopted, so that a plurality of correlation peak points cannot appear during correlation synchronization. The local synchronization signal uses non-gaussian filtering and no oversampling, saving multipliers, and the length of each correlation is only the length of the original synchronization symbol. And the intra-hop training sequence is subjected to sliding correlation to obtain a matched filter h, the received signal r is subjected to matched filtering and then is subjected to correlation with the local training sequence, and a synchronization point after matched filtering is found, so that the final synchronization in each hop is completed, and the triple synchronization is really realized in a seamless connection manner.

Description

Multiple synchronization method of FHSS-GMSK system

Technical Field

The invention relates to the technical field of wireless communication, in particular to a multiple synchronization method of an FHSS-GMSK system.

Background

In the prior art, the GMSK reception flow chart is that an input signal is first subjected to frequency hopping synchronization as shown in fig. 1, and then the start position and the current Time (TOD) of frequency hopping are found, and initial timing synchronization and carrier frequency offset estimation are completed at the same time. And then, carrying out debounce according to a frequency hopping rule, carrying out timing synchronization tracking and carrier frequency offset tracking when data is hopped, then carrying out GMSK demodulation, and sending the obtained soft demodulation information to an LDPC decoding module for decoding.

The system has the symbol rate of 12.5MHz, the working clock and sampling rate of 100MHz and 8 times of sampling. After an input radio frequency signal is subjected to quadrature down-conversion to a baseband, the input radio frequency signal is subjected to low-pass filtering and then is subjected to subsequent processing.

The synchronous capture is completed by using a pseudo-random code in a frequency hopping synchronous head, during searching, whether an actual signal exists or a frequency hopping sweep frequency is aligned or not is not considered, the received signal is always correlated with the synchronous head, when noise is received or the frequency is not swept, a correlation peak is very small, and only when the signal of the synchronous head is received and the frequency is swept and aligned (namely, the optimal sampling point is aligned), the correlation peak exceeds a threshold, so that the synchronous capture is completed.

When correlation is performed, waveform correlation is adopted, that is, correlation is performed between a received signal and a known GMSK modulation waveform, and if the received signal is r (n) and the local known waveform is l (n), the correlation process is as follows:

the correlation time length is half hop, and for the coarse correlation, N is 100MHz/32 × 48.64us is 152 sampling points; for fine correlation, N is 100MHz by 48.64us 4864 samples. When the correlation value exceeds a threshold, capture is considered up.

The correlation process is divided into two stages of coarse correlation and fine correlation. The initial coarse correlation is performed using the first half of the ACODE. The symbol rate of the first half part is low, and 32 times of extraction is performed during filtering, so that the time interval during searching is larger, the searching is performed once every 32 clock cycles, the number of data points used during each correlation calculation is 152, and the searching calculation amount is reduced. Each correlation requires 152 complex multiplication operations, which must be completed in 32 clock cycles, and at least 152/32-5 complex multipliers are required, each complex multiplier needs 4 real multipliers, so 5-4-20 real multipliers are required. Since there are 4 sets of a CODEs that need to be correlated in parallel at the same time, 20 x 4 to 80 real multipliers are required. Considering that 2 of the 4 sets of a CODEs are inversed by the other 2 sets, the corresponding complex signal waveforms are conjugate to each other, so that the number of multipliers is reduced by half, i.e. only 40 real multipliers are needed.

Due to the existence of doppler, there is a maximum frequency offset of ± 18kHz, so that in correlation, the influence of the frequency offset needs to be considered, and a search is also needed at frequency offset points, the interval of the search is 6kHz, and a total of 6 frequency offset points need to be searched. The frequency offset search value is:

in order to reduce the amount of calculation during frequency offset search, the sampling rate is further reduced and then correlation is performed on different frequency points. That is, after the received signal is first multiplied by the local a CODE waveform conjugate to remove the modulation information, 0 is integrated every 8 points and down-sampled to 19 data points. Again, using these 19 points, a search is made over the 6 frequency offset points. Each frequency offset point requires 19 complex multiplication operations, and 6 frequency offset points require 19 × 6-114 complex multiplication operations. Since 3 values of the 6 frequency offset points are the inverses of the other 3 values, and the corresponding complex carrier waveforms are conjugate to each other, half of the multiplication operations are reduced, that is, only 57 times of complex multiplication operations are needed. It must finish within 32 clock cycles, and at least 57/32 ═ 2 complex multipliers, i.e. 8 real multipliers are needed.

After coarse correlation, there are still 32 sampling points in the uncertainty in time, and fine correlation needs to be performed using the latter half of the a CODE. The fine correlation uses the initial position obtained by the coarse correlation to find the initial position corresponding to the latter half, and the fine search is performed by taking 16 sampling points (33 points in total) before and after the initial position as the initial position. The number of sample points used for the fine search was 4864. And comparing the 33 correlation values, wherein the position corresponding to the maximum correlation value is the optimal position. The fine correlation calculation requires 4864 × 33 complex multiplication operations, but has 4 hops to process (B CODE of the same frequency is at least 4 hops later), so 4864 × 33/(9728 × 4) ═ 5 complex multipliers, i.e., 20 real multipliers, are required.

And in the best position, carrying out finer search on the frequency deviation, wherein the search step is 100Hz, the search range is +/-3 kHz, and 30 frequency deviation points need to be searched. In order to reduce the calculation amount of searching on the frequency offset points, every 256 sampling points are integrated and cleared to be 0, 4864 sampling points are reduced to 19 points, and then searching is carried out on 30 frequency offset points. 19 × 30/2 ═ 285 complex multiplication operations are required.

After the A CODE is captured and the jump synchronization is completed, the frequency is continuously used for searching the 2 nd group of initial synchronization frequency CODEs, and the search is carried out at the corresponding position (4 jumps or integral multiple thereof) during the capture. The elimination method is that the received data is related to the corresponding local A CODE, if the related value is lower than the threshold, which indicates that the received data is not the A CODE, the timing tracking, the carrier tracking and the demodulation decoding are carried out on the hop according to the processing mode of the common data hop, and then the TTOD is extracted [22:17 ].

Under high speed (8M, 4M and 2M), timing tracking is carried out by utilizing the head and tail of each jump, and because the contents of the head and the tail are known, the received data of the head and the tail are correlated with the locally known waveform of the head and the tail at the current optimal sampling point and two points before and after the current optimal sampling point by utilizing the received data of the head and the tail, and the correlation value of which position is the maximum is judged to be the optimal sampling point.

At low rates (125k, 250k), spreading is used (spreading ratios of 16 and 32), so timing tracking is performed with 1024 data symbols. Still, at the current optimal sampling point and two points before and after the current optimal sampling point, the received data symbols and the local spreading code waveform are used for correlation at 3 positions: in a symbol (spreading period), the received data is correlated with the spreading code waveform, and because of 0 and 1, the correlation value of each symbol at 0 and 1 is compared, the larger one is taken as the correlation value of the symbol, and then the correlation values of all the data symbols are summed to be the final correlation value.

Disclosure of Invention

Aiming at the defects of the prior art, the invention discloses a multiple synchronization method of an FHSS-GMSK system, which is used for solving the problems that the coarse synchronization, the coarse frequency offset, the accurate synchronization and the fine frequency offset are too complex and the anti-noise effect is limited when the coarse frequency offset, the accurate synchronization and the fine frequency offset are carried out on GMSK signals in the prior art.

The invention is realized by the following technical scheme:

a multiple synchronization method of FHSS-GMSK system includes the following steps:

s1 sets three-stage synchronous jump structure for GMSK signal;

s2, independently adjusting GMSK signals;

s3, synchronously processing the GMSK signal in S2 and calculating frequency offset;

s4 synchronizes the signals in each hop for 2 times, and finally completes the synchronization in each hop.

Furthermore, in S1, three-level synchronization is set, where each second of GMSK is set to be an initial synchronization hop, and 4 independent GMSK modulation signals are used for synchronization hops;

the second stage adopts 2 groups of independent GMSK modulation signals in each time slot for the second stage synchronization and frequency offset calibration;

and the third stage adopts a training sequence for synchronization of each hop, and the third stage does not carry out frequency offset calibration.

Furthermore, in S2, 4 sets of training sequences modulated by independent GMSKs are used, and GMSKs of each segment are independently modulated, and the processing flow of the slave modulator is that the input data sequence is a binary data sequence consisting of a {0,1} sequence; before GMSK modulation, differential encoding is first required, and then the return-to-zero signal is converted into a non-return-to-zero sequence, i.e.:

where d e 0,1 and a e-1, 1 represent the differentially encoded input and output sequences, respectively.

Further, the modulation process is as follows:

before modulation, shaping filtering is carried out, proper shaping filtering function g (t) is selected,

wherein rect (x) is defined as:

h (t) definition

Wherein:

BT＝0.3

b is a filter with impulse response h (T) and 3db bandwidth, T is the delay of one input data bit;

symbol sequence a [ n ]]Convolving with g (t) to obtain frequency function, multiplying with pi to obtain phase function

Is composed of

Wherein h is a modulation coefficient; and (3) carrier modulation processing:

I＝cosΘ

Q＝cosΘ

further, GMSK modulation is divided into two cases: the first is the modulation of the transmitted synchronous signal GMSK, the modulation of the transmitted signal requires oversampling, the oversampling factor OSR being defined by f_s/r_bWherein f is_sIs the sampling frequency, r_bIs the symbol rate and the second is the local pre-stored synchronization signal loc _ S, without oversampling.

Further, in S3, the matched and filtered data is captured, and the capturing is performed at intervals of 1/OSR symbol, and each 1/OSR symbol is correlated with the known sync header by the received data.

Further, the segmentation search algorithm is as follows,

the length onel of each section in the formula, the number of each hop section is M, and the OSR is the oversampling multiple of the GMSK signal; the received signal r is over-sampled, the over-sampling multiple is OSR multiple, and the local synchronizing signal loc _ s is not over-sampled, so that when the received signal is related to the local signal, every other OSR signals are related to the corresponding local signal;

the search algorithm samples and searches in a segmented mode, relevant numerical value vectors in the segments are overlapped, and signals between the segments are overlapped in a scalar mode; the segmentation search algorithm finds out the initial synchronization point of the signal by finding out the maximum correlation peak point, namely maxPOS, namely the initial position of the synchronization head;

[maxValue,maxPOS]＝max(sum_sample(t))

maximum frequency offset that can be corrected at this time

The frequency offset search value is:

when calculating the frequency offset by adding the two sections of conjugate multiplication after pinching, the original modulation signal needs to be removed firstly;

r_pss(t)＝loc_S^*(n+m·oneL).*r([n+m·oneL]*OSR+t)

and (3) carrying out conjugate multiplication on the front section and the rear section to obtain frequency offset, if 4 sections exist, obtaining 3 frequency offset values, and averaging the three frequency offset values to obtain more accurate frequency offset values:

Δf＝mean(Δf_m)；

OSR is an oversampling factor, where fs is the sampling frequency, fb is the symbol rate, Tb is the sampling interval of the physical layer, and Tb is 1/fb.

Furthermore, in S4, the synchronization is train obtained according to the correlation characteristic of the TRAINING sequence, and T _ SEQ is obtained after GMSK non-interpolation and high-speed filtering mapping; after the training sequence passes through the channel, the signal received at the receiving end is:

where h is the channel impulse response, w is the channel noise, and T is the noise_SEQ[-]^*Convolution with the above equation:

the approximation of the third step in the above equation is based on w being white noise and T_SEQHas white noise characteristics; if the received burst signal is T_SEQ[-]^*And (4) performing convolution, wherein the result is represented by v, so that the v contains the channel impulse response, and the synchronization and the channel estimation are simultaneously completed by adopting a sliding window technology.

Further, in the GSM system, a sliding window technique is used for synchronization of dedicated synchronization bursts, which determines the sampling time of the received signal; the sliding window technique is implemented by r and T_SEQ[-]^*Convolution obtains a signal h:

h＝r*T_SEQ[-]^*

after obtaining the sampling synchronization and estimating the channel impulse response, the output of the matched filter is:

y＝r*h^*[-]

after the output is finished, because the signal y output by the matched filtering is an over-sampling signal, the correlation with the transmitting signal is continued at the moment

At this point y continues and T_SEQ[-]^*Carry out correlation

yT＝y*T_SEQ[-]^*

Finally, the maximum value position of yT is calculated, and the position is the initial position of the final jump signal

[maxV,maxPOSyT]＝max(yT)。

The invention has the beneficial effects that:

the invention utilizes multiple synchronization, and can quickly and accurately synchronize and correct frequency offset; and an independent GMSK modulation mode is adopted, so that a plurality of correlation peak points cannot appear during correlation synchronization. (ii) a The local synchronous signal adopts non-Gaussian filtering and no oversampling, so that the generated local synchronous sequence is correlated with the receiving sequence without multiplication operation, a multiplier is saved, and the length of each correlation is only the length of the original synchronous symbol. And the intra-hop training sequence is subjected to sliding correlation to obtain a matched filter h, the received signal r is subjected to matched filtering and then is subjected to correlation with the local training sequence, and a synchronization point after matched filtering is found, so that the final synchronization in each hop is completed, and the triple synchronization is really realized in a seamless connection manner.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings are obtained according to the drawings without creative efforts.

Fig. 1 is a flow chart of GMSK reception in the prior art;

FIG. 2 is a three level synchronous frame architecture diagram;

fig. 3 is a GMSK frame structure diagram;

FIG. 4 is a GMSK independent modulation diagram;

fig. 5 is a diagram of a GMSK modulator transmitting a signal;

fig. 6 is a diagram of a locally pre-stored GMSK signal loc _ S for synchronization;

FIG. 7 is a plot of multiple correlation peaks occurring when the initial phase is the same;

FIG. 8 is a plot of only one correlation peak occurring at different initial phases;

FIG. 9 is a peak point plot of the final sliding correlation;

figure 10 is a flow chart of multiple synchronization of GMSK-FHSS signals.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In this embodiment, in order to improve the reliability of system frequency offset measurement and improve the accuracy of system synchronization acquisition, three-level synchronization is set, an initial synchronization hop is set every second in GMSK, and 4 independent sets of GMSK modulation signals (4 × 345) are used for synchronization hop. The second stage uses 2 independent sets of GMSK modulated signals (2 x 621) per slot for the second stage synchronization and frequency offset calibration. And the third stage adopts a training sequence for synchronization of each hop, and the third stage does not carry out frequency offset calibration.

Fig. 2 is a three-level synchronous frame mechanism diagram, the first level initial synchronization hop is much and short, and large frequency offset is easy to apply. The second-stage intermediate synchronization has few skip sections and each section is long, so that the estimation can be more accurate. The training sequence of each hop of the third stage can be slid within the window to estimate the exact phase offset and the main path position of this hop for later demodulation.

Note that: the middle synchronization jump needs to be windowed according to the position of the previous initial synchronization jump, the influence of the following multipath is considered, and the main path can jump in a certain range, so the main path is positioned according to the initial synchronization jump; the front and back 8 OSR points search for the maximum correlation value. And the initial synchronization jump recalculates the updates each time.

Of course, for the sake of system simplification, a 2-level synchronization process may be adopted, and fig. 3 is an example of a GMSK frame structure diagram.

GMSK independent adjustment principle and implementation steps: in the above synchronization hops, whether the initial synchronization hop or the intermediate synchronization hop, each hop is composed of a plurality of independent GMSK modulation data blocks, and each block is an M sequence or a GOLD sequence, and the phases are different between different blocks.

Fig. 4 shows that GMSK independent modulation means that the input GMSK modulators are all independently input, that is, the d1 sequence is input, after modulation is completed, the d2 sequence is input, and then the d3 and d4 sequences are input in sequence. The sequences after the respective modulation are connected in series again. Continuous modulation is that firstly [ d1, d2, d3 and d4] are serially connected and then modulated. The independent modulation is firstly modulated in series. The advantage of independent modulation is that each segment of the signal can be processed independently and there is no phase correlation with the preceding and succeeding segments.

4 sets of training sequences modulated by independent GMSK are adopted, and GMSK of each segment is modulated independently, namely, the slave modulator executes the processing flow.

The process flow of the modulator is that the input data sequence is a binary data sequence consisting of a {0,1} sequence. Before GMSK modulation, differential coding is first required, and then the return-to-zero signal (RTZ) is converted into a non-return-to-zero sequence (NRZ), i.e.:

The modulation process comprises the following steps:

shaping filtering is performed before modulation. To select the appropriate shaping filter function g (t),

wherein rect (x) is defined as:

h (t) is defined as

Wherein:

BT＝0.3

b is a filter with impulse response h (t) and 3db bandwidth. T is the delay of one input data bit.

Is composed of

Where h is the modulation factor, and the exemplary system value is 0.5.

And (3) carrier modulation processing:

I＝cosΘ

Q＝cosΘ

GMSK modulation is divided into two cases: the first is the modulation of the transmitted synchronous signal GMSK, the modulation of the transmitted signal requires oversampling, the oversampling factor OSR being defined by f_s/r_bWherein f is_sIs the sampling frequency, r^bType 2 is a locally pre-stored synchronization signal loc _ S, not oversampled. Fig. 5 is a GMSK modulator transmitting a signal; FIG. 6 is a diagram for local prestoringIn a loc _ S diagram of the synchronized GMSK signal; the GMSK synchronous signal loc _ S pre-stored locally does not need to pass through a Gaussian interpolation filtering module. The base band rate fb is used. Therefore, the storage capacity is reduced, the subsequent correlation operation amount is greatly reduced, and the correlation precision also reaches Tb/OSR.

It is simulated below that a plurality of correlation peak points occur when 4 segments of GMSK signals are not independently modulated and the original signals modulated by each segment of GMSK are the same. If independent modulation is used for the GMSK signal and the original synchronization signal is initially out of phase, only one correlation peak occurs, so that the correlation peak is easily found. The calculation is simple and accurate. FIG. 7 shows that multiple correlation peaks occur when the initial phases are the same; fig. 8 shows that only one correlation peak occurs when the initial phases are different.

Synchronization processing and frequency offset calculation: a synchronization jump is divided into a plurality of synchronization blocks, and a plurality of segment correlation results are superposed in absolute numerical value in order to enhance noise immunity. And the PN sequence of each section is uncorrelated, and the phase difference of the front section and the rear section is utilized to synchronously calculate the magnitude of the frequency offset. Although GMSK signals are transmitted, 1-1 alternative signals are still adopted locally, multiplication operation is avoided, high innovativeness is achieved for synchronization and correlation of the GMSK signals at the time, and synchronization and frequency offset measurement is carried out in a frequency offset preset parallel mode in order to expand subsequent compatibility of larger frequency offset.

The match filtered data is captured by searching at 1/OSR symbol intervals, i.e., every 1/OSR (e.g., OSR: 4,8, etc.) symbol is correlated with the known sync header using the received data.

The magnitude of the doppler plus clock difference of the present system is about 2 × 10-6, and at a carrier frequency of 2GHz, the generated frequency offset value is about 2GHz × 2 × 10-6 kHz, the maximum moving speed is known to be 1.5km/s, so the maximum frequency offset is 10kHz, the maximum phase rotation generated at 345 symbols during acquisition is 2 × pi 14kHz 345/12.5MHz × 14 × 345/12500 pi 2 pi 0.38 × 2 pi, and the loss caused by acquisition is limited and ignored. The segmentation search algorithm is as follows,

the length oneL of each segment in the above formula (e.g. equal to 345,691),

the number of segments per hop is M (e.g. equal to 2, 4)

OSR is the oversampling multiple of the GMSK signal (in general 4, 8)

The received signal r is oversampled, the oversampling multiple is OSR multiple, and the local synchronization signal loc _ s is not oversampled, so that when the received signal is correlated with a local signal, every other OSR signal is correlated with the corresponding local signal. The aim of speed consistency is achieved, and meanwhile, the correlation length of the whole system is reduced. The search algorithm samples segmented search, intra-segment correlation value vector superposition, and inter-segment signal scalar (absolute value) superposition.

The segmentation search algorithm finds the initial synchronization point of the signal, i.e. the maxPOS, i.e. the initial position of the synchronization header, by finding the maximum correlation peak point.

[maxValue,maxPOS]＝max(sum_sample(t))

Maximum frequency offset that can be corrected at this time

The maximum corrected frequency offset is calculated to obtain +/-12.5 +/-10 +/-6/(2 +/-345) +/-18 kHz

If larger doppler exists, the maximum frequency offset will be ± 18kHz × 3 ═ 54kHz, so in the correlation, first, search is performed on the frequency offset points with large intervals, the intervals of search are 18kHz, and 3 frequency offset points are needed to be searched. The frequency offset search value is:

and carrying out sectional correlation in the frequency points searched by the two. Of course, the length of each segment is also reduced to meet the requirement of larger frequency offset.

For frequency offset.

When calculating the frequency offset by adding the two sections of conjugate multiplication after pinching, the original modulation signal needs to be removed first.

r_pss(t)＝loc_S^*(n+m·oneL).*r([n+m·oneL]*OSR+t)

Δf＝mean(Δf_m)

the frequency deviation is larger in the initial synchronization, and the frequency deviation and the synchronization are obtained by adopting a 4 x 309 mode.

The synchronization head reduces the pressure of the training sequence in the subsequent data hop after completing the frequency offset calibration and synchronization. The frequency variation within a time slot is typically small (e.g. less than 1000HZ),

the intermediate synchronization jump further corrects the frequency offset, so that the frequency offset is controlled within 500 HZ.

Such small frequency offsets

The demodulation performance of the system is affected to a limited extent. Therefore, the subsequent training sequence does not perform the frequency offset calibration.

The 2 times synchronization of the signals in the jump is found by the initial synchronization and the intermediate synchronization, but in order to adapt to each jump, the relative position of the main path may change due to the multipath, and the jump synchronization is performed again by windowing before and after the approximate position determined in the previous jump. At this point, the windowing is performed, and the matched filter coefficients are obtained as long as the signals are less correlated, and the matched filter coefficients are searched according to the initial position determined by the synchronization jump.

The synchronization, channel estimation and matching filtering are performed in two steps. In order to accomplish matched filtering, synchronization and channel estimation must first be performed.

Both channel estimation and matched filtering are input with a received signal r, which is a sequence of samples of a received GSM burst signal. The oversampling factor OSR is defined as f_s/r_bWherein f is_sIs the sampling frequency, r_bIs the symbol rate, L_hWhich represents the expected length of the channel impulse response in bit time. The channel estimator inputs the channel impulse response h to the matched filter while passing the estimated burst position in the received signal r.

The following is the extraction of points around the matched filter peak point for subsequent matched filtering

And matching the filtered signal, re-correlating with the training sequence again, and finding the optimal sampling point after matched filtering.

It is possible that matched filtering will shift the optimal sampling point.

The synchronization is TRAINING obtained according to the correlation characteristic of the TRAINING sequence, and T _ SEQ is obtained through gMSK non-interpolation and high-speed filtering mapping. After the training sequence passes through the channel, the signal received at the receiving end is:

r_TSEQ＝T_SEQ*h+w

the approximation of the third step in the above equation is based on w being white noise and T_SEQHas white noise characteristics. If the received burst signal is T_SEQ[-]^*And (4) performing convolution, wherein the result is represented by v, and the channel impulse response is contained in v, so that the synchronization and the channel estimation are simultaneously completed by adopting a sliding window technology.

In the GSM system, a sliding window technique is used for synchronization of dedicated synchronization bursts, which determine the sampling time of the received signal. The sliding window technique is implemented by r and T_SEQ[-]^*Convolution obtains a signal h:

h＝r*T_SEQ[-]^*

y＝r*h^*[-]

At this point y continues and T_SEQ[-]^*Carry out correlation

yT＝y*T_SEQ[-]^*

And finally, calculating the position of the maximum value of the yT, wherein the position is the initial position of the final jump signal. [ maxV, maxPOSyT ] ═ max (yT)

FIG. 9 is a peak point plot of the final sliding correlation; the whole algorithm processing flow is shown in fig. 10, and the whole flow involves 3 windowing positions:

the first position is: after initial synchronization, determining the starting point and the end point of the relevant position of the middle synchronization jump, and windowing for the first time.

The second position is as follows: and after the middle synchronous jump correlation is determined, determining a starting point and an end point of the matched filtering, and windowing for the second time.

A third position: the match filtered signal y is correlated with the local sync sequence, specifying the start and end of the correlation, and windowed for a third time.

The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments are still modified, or some technical features are equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. A multiple synchronization method for FHSS-GMSK system is characterized in that the synchronization method includes the following steps:

s1 sets three-stage synchronous jump structure for GMSK signal;

s2, independently adjusting GMSK signals;

s4, synchronizing the signals in each hop for 2 times to complete the final synchronization in each hop;

in the step S1, three-level synchronization is set, an initial synchronization hop is set every second for GMSK, and 4 independent sets of GMSK modulation signals are used for synchronization hops;

the third stage adopts a training sequence for synchronization of each hop, and the third stage does not carry out frequency offset calibration;

in S2, 4 sets of independent GMSK modulated training sequences are used, the GMSK of each segment is independently modulated, and the processing flow of the slave modulator is that the input data sequence is a binary data sequence consisting of a {0,1} sequence; before GMSK modulation, differential encoding is first required, and then the return-to-zero signal is converted into a non-return-to-zero sequence, i.e.:

2. The multiple synchronization method for FHSS-GMSK system as claimed in claim 1, wherein the modulation process is:

wherein rect (x) is defined as:

otherwise

h (t) definition

Wherein:

BT＝0.3

Is composed of

Wherein h is a modulation coefficient; and (3) carrier modulation processing:

I＝cosΘ

Q＝cosΘ。

3. the multiple synchronization method for FHSS-GMSK system according to claim 2, wherein the GMSK modulation is divided into two cases: the first is the modulation of the transmitted synchronous signal GMSK, the modulation of the transmitted signal requires oversampling, the oversampling factor OSR being defined by f_s/r_bWherein f is_sIs the sampling frequency, r_bIs the symbol rate and the second is the local pre-stored synchronization signal loc _ S, without oversampling.

4. The multiple synchronization method for FHSS-GMSK system of claim 1, wherein in S3, the matched and filtered data is captured, the capturing is performed by searching at intervals of 1/OSR symbol, each 1/OSR symbol is correlated with the received data and the known synchronization header, and the OSR is defined as f_s/r_bWherein f is_sIs the sampling frequency, r_bIs the symbol rate.

5. The multiple synchronization method for FHSS-GMSK system according to claim 4, wherein the segmentation search algorithm is as follows,

the length oneL of each segment in the formula, the number of segments per hop is M, and OSR is the oversampling multiple of GMSK signal; the received signal r is over-sampled, the over-sampling multiple is OSR multiple, and the local synchronous signal local _ s is not over-sampled, so when the received signal is correlated with the local signal, every other OSR signals are correlated with the corresponding local signal;

the search algorithm samples and searches in a segmented mode, relevant numerical value vectors in the segments are overlapped, and signals between the segments are overlapped in a scalar mode; a segmentation search algorithm is used for finding out an initial synchronization point of a signal by finding out a maximum correlation peak point, wherein maxPOS is the initial position of a synchronization head;

[maxValue,maxPOS]＝max(sum_sample(t))

maximum frequency offset that can be corrected at this time

The frequency offset search value is:

r_pss(t)＝loc_S^*(n+m·oneL).*r([n+m·oneL]*OSR+t)

and (3) carrying out conjugate multiplication on the front section and the rear section to obtain frequency offset, and averaging the frequency offset value to obtain an accurate frequency offset value:

Δf＝mean(Δf_m)；

6. The multiple synchronization method for FHSS-GMSK system as claimed in claim 1, wherein in S4, the synchronization is TRAINING obtained according to the correlation property of TRAINING sequence, and T _ SEQ is obtained after GMSK non-interpolation and high speed filtering mapping; after the training sequence passes through the channel, the signal received at the receiving end is:

7. The multiple synchronization method for FHSS-GMSK system according to claim 6, wherein in GSM system, sliding window technique is used for synchronization of dedicated synchronization burst, which determines the sampling time of the received signal; the sliding window technique is implemented by r and T_SEQ[-]^*Convolution obtains a signal h:

h＝r*T_SEQ[-]^*

y＝r*h^*[-]

At this point y continues and T_SEQ[-]^*Carry out correlation

yT＝y*T_SEQ[-]^*

Finally, the yT maximum value position is calculated, and this position is the starting position of the final jump signal [ maxV, maxPOSyT ] ═ max (yT).