A low-complexity method for fine synchronization of MSK signals

Abstract

The present invention provides a low-complexity MSK signal fine synchronization method. In the fine synchronization module, the training sequence is subjected to single MSK modulation, and then up-sampled and expanded as a local signal, and a complex correlation operation is performed with the received filter signal in a symbol interval before and after the coarse synchronization moment, and the fine synchronization moment is finally determined. The present invention can significantly reduce the amount of calculation and complexity, effectively save hardware resources, and be easy to implement in engineering without losing the fine synchronization accuracy and maintaining high mobility and adaptability.

Description

Technical Field

[0001] The invention relates to signal fine synchronization and TOA (Time Of Arrival) measurement, and is applied to a digital communication system of MSK (Minimum Shift Keying) modulation. Background Art

[0002] MSK modulation is a phase-continuous, envelope-constant modulation method with excellent spectrum efficiency and power efficiency, and is widely used in military and civilian communications. When receiving an MSK signal, the signal must first be captured synchronously, and then fine synchronization of the signal must be performed before signal demodulation can be completed. Among them, the fine synchronization accuracy has a significant impact on the demodulation performance. The higher the accuracy, the smaller the inter-code interference and phase noise caused by the synchronization error, and the better the demodulation performance. In addition, in distributed communication networking, RTT (Round-Trip Time) is used to complete network autonomous synchronization, and the timing accuracy of RTT is determined by the fine synchronization accuracy of the RTT signal; in TOA-based ranging and positioning, the ranging and positioning accuracy is determined by the measurement accuracy of TOA, and the measurement accuracy of TOA is also determined by the fine synchronization accuracy of the measurement signal. Therefore, the signal fine synchronization accuracy has a direct and decisive impact on the performance of the communication system and network.

[0003] At present, the most commonly used fine synchronization method with good comprehensive performance is the signal complex correlation synchronization method based on the training sequence. Figure 1 As shown in FIG. 1 , after the received signal is sampled by A / D (Analog / Digital, analog to digital conversion), two signals of real part and imaginary part are formed; the two signals are respectively low-pass filtered to form S I (n) and S Q (n), where n is the sampling point; then decimation is performed to obtain an decimated signal S' with a lower sampling rate I (D·n) and S' Q (D·n), D is the extraction multiple, and then the received signal is synchronously captured. If the capture is successful, the rough synchronization moment of the signal can be obtained. The accuracy of this moment is poor, so the synchronization capture is also called coarse synchronization. The subsequent fine synchronization module is also required to improve the synchronization accuracy of the signal. Among them, signal extraction can significantly reduce the calculation amount of synchronization capture; in the fine synchronization module, the training sequence is modulated by MSK to form a local signal, and the real part and imaginary part are L respectively. I (n) and L Q (n), the sampling rate is the same as A / D; the local signal L I (n), L Q (n) The filtered signal S within one symbol time before and after the coarse synchronization moment I (n), SQ (n), perform floating-point complex correlation operation to obtain the complex correlation value R(n), and finally, perform peak search on the correlation value modulus |R(n)|. The peak moment is the more accurate fine synchronization moment.

[0004] The precision of fine synchronization is subject to the sampling rate of the signal. The higher the sampling rate, the higher the synchronization precision. To obtain better synchronization precision, the A / D is usually set to a higher sampling rate. In the above conventional floating-point complex correlation method, since fine synchronization uses all the sampling signals of the training sequence for floating-point complex correlation operations, it contains a large number of floating-point multiplications, which requires a large amount of calculations and consumes a large amount of hardware resources in engineering implementation. If a serial method is used for resource optimization, the control logic will become complicated and the processing delay will be increased. If the complexity of fine synchronization calculation is reduced, the signal sampling rate needs to be reduced, and the synchronization precision will be greatly lost. At present, achieving high-precision synchronization with low computational complexity is a technical problem in the fine synchronization of MSK signals. Summary of the invention

[0005] In order to overcome the shortcomings of the prior art, the present invention provides a low-complexity MSK signal fine synchronization method, which significantly reduces the amount of calculation and complexity without losing the fine synchronization accuracy and maintaining high maneuverability and adaptability, effectively saves hardware resources, and is easy to implement in engineering.

[0006] The technical solution adopted by the present invention to solve the technical problem comprises the following steps:

[0007] (1) Perform single MSK modulation with an initial phase of 0 on a training sequence of length K. The modulated signal is recorded as L(k), which contains the real part L I (k) and the imaginary part L Q (k), the value can be only ±1, ±j, where k is the modulation time;

[0008] (2) Upsample L(k) by a factor of U to obtain an upsampled signal L'(n), where n is the sampling time.

[0009] (3) Expand and shape the up-sampled signal L'(n) by M times to obtain a shaped signal L"(n), where M≤U and is an odd number;

[0010] (4) According to the local signal L”(n), the coarse synchronization time T C The floating point signal S(n) received in each of the preceding and following symbol intervals is subjected to complex correlation operation to obtain the correlation value R(n);

[0011] (5) Take the modulus of the correlation value and search for the maximum value of |R(n)| within the range. The sampling time of the maximum value is recorded as T off , then the fine synchronization time

[0012] In the step (2), the up-sampled signal

[0013] In the step (3), the forming signal

[0014] In the step (3), M=3.

[0015] In the step (4), the relevant value Where, L"(n) * is the conjugate of L”(n).

[0016] In step (4), it is assumed that the real part and imaginary part of S(n) are S I (n) and S Q (n), the real and imaginary parts of the correlation value R(n) are R I (n) and R Q (n), then

[0017]

[0018] Among them, the value range of sampling time n is T C -U+1≤n≤T C +U-1;a k 、b k The way in which the multiplier on the right side participates in the accumulation operation is defined, and the value of L(k) is ﹢1, a k The value is ﹢1, b k The value is 0; L(k) is -1, a k The value is -1, b k The value is 0; L(k) takes the value of +j, a k The value is 0, b k The value is ﹢1; L(k) is ﹣j, a k The value is 0, b k The value is -1; +1 means that the multiplier on the right participates in the addition operation as the original number; -1 means that the multiplier on the right participates in the addition operation after being inverted; 0 means that it does not participate in the operation.

[0019] In the step (5), the correlation value modulo

[0020] In the step (5), when searching for the peak value of the correlation value of the signal, the square root operation of the last modulus when calculating the correlation value at each moment is omitted, and the square value of the modulus of the correlation value is directly calculated, |R(n)| 2 =|R I (n)| 2 +|R Q (n)|2 .

[0021] The beneficial effect of the present invention is that since the local signal value has only four cases of ±1, ±j, its signal constellation diagram is as follows: Figure 3 As shown, the multiplication and addition in the complex correlation operation of the floating-point signal is simplified to addition, and in the case of high-multiple sampling, only part of the U sampling points of each symbol of the received signal participates in the operation, which significantly reduces the amount of calculation, reduces the complexity of engineering implementation, and saves hardware resources. In order to facilitate the comparison of the amount of calculation, the present invention and the conventional floating-point complex correlation method do not use the implementation method of serial optimization. The comparison results are shown in the following table:

[0022] Fine-grained synchronization methods Number of multipliers Number of adders Conventional floating point complex correlation method 4KU+2 About 4KU The present invention 2 (used when calculating the square of the modulus value) About 2KM

[0023] As can be seen from the above table, no matter how long the training sequence is, the present invention uses 2 multipliers, and the number of adders used is reduced to about M / (2U) of the conventional floating-point complex correlation method. For example, if the training sequence length K=256, the sampling rate U=20, and M=3, the conventional floating-point complex correlation method requires 20482 multipliers and about 20480 adders, while the present invention only requires 2 multipliers and 1536 adders.

[0024] Although the amount of calculation is greatly reduced in the present invention, the synchronization accuracy is not lost. The conventional floating point complex correlation method and the present invention method are simulated and compared for fine synchronization error performance. The simulation conditions are: the sampling rate of A\D is set to 100Msps (megasamples / second), the sampling interval is 10ns; the symbol rate is set to 5MBaud (megabaud), and the sampling multiple is 20; the training sequence length is 256; the signal-to-noise ratio range is -5 to 20dB; a Gaussian white noise channel is used; the number of simulations is 10,000; the expansion multiples in the present invention are set to 1, 3, and 5 respectively. The simulation results are as follows Figure 4 , Figure 5 As shown, Figure 4 is the error mean curve of fine synchronization. The mean is used to examine the unbiasedness of fine synchronization. It can be seen from the figure that the error mean of the conventional method under different signal-to-noise ratios fluctuates around 0, and the error mean of the present invention under different signal-to-noise ratios is within 0.4ns, which is much smaller than the sampling interval. As the signal-to-noise ratio increases, the error mean of both the conventional method and the present invention tends to 0. It can be seen that the fine synchronization of the present invention is the same as the conventional method and can be considered unbiased. Figure 5The error root mean square value curve (RMSE) of fine synchronization is used to examine the error performance of fine synchronization. It can be seen from the figure that the error performance of the present invention is basically the same when the expansion multiples are 3 and 5, and is not degraded compared with the conventional method, and is even slightly better than the conventional method. When the expansion multiple is 1, the error performance of the present invention is degraded by 0.5ns compared with the conventional method. Therefore, it is appropriate to take the expansion multiple of 3 in the present invention (the calculation amount when the expansion multiple is 3 is smaller than that when the expansion multiple is 5). When the expansion multiple is 3, the accuracy of fine synchronization is not lost.

[0025] The present invention is similar to the conventional floating point correlation method in that it is applicable to high-speed mobile communications and has high mobility adaptability. The present invention method (taking a 3-fold expansion multiple as an example) and the conventional floating point complex correlation method are simulated and compared under mobile conditions. Under the above simulation conditions, a 1kHz Doppler frequency deviation is added, and the simulation results of the error performance are as follows: Figure 6 As shown in the figure, under the predetermined moving conditions, the error root mean square value of the method of the present invention is consistent with that of the conventional floating point complex correlation method, and is only degraded by less than 0.5ns compared with the case of no Doppler frequency deviation, which is much smaller than the signal sampling interval. Therefore, the present invention can also be applied to high-speed moving communication scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of a conventional floating point complex correlation fine synchronization method;

[0027] Figure 2 It is a schematic diagram of the method of the present invention;

[0028] Figure 3 It is the constellation diagram of single MSK signal and quadruple MSK signal;

[0029] Figure 4 It is a mean value curve diagram of the fine synchronization error of the conventional floating point complex correlation method and the present invention;

[0030] Figure 5 It is a curve diagram of the RMS value of the fine synchronization error of the conventional floating point complex correlation method and the present invention;

[0031] Figure 6 It is a curve diagram of the RMS value of fine synchronization error of the conventional floating point complex correlation method and the present invention (when the expansion factor is 3) at 1 kHz Doppler frequency shift. DETAILED DESCRIPTION

[0032] The present invention is further described below in conjunction with the accompanying drawings and embodiments. The present invention includes but is not limited to the following embodiments.

[0033] In view of the problems of high computational complexity and large consumption of hardware resources for engineering implementation of conventional floating-point complex correlation method in fine synchronization of MSK signals, the present invention achieves high-precision fine synchronization of MSK signals with lower computational complexity, which not only ensures good synchronization performance, facilitates demodulation of optimal sampling points of MSK signals and accurate measurement of TOA, but also facilitates engineering implementation.

[0034] In the present invention, the A / D is configured to a relatively high sampling rate, generally an integer multiple of the signal symbol rate, and the sampling multiple is recorded as U (unit: number of sampling points / symbol). In the fine synchronization module, the training sequence is subjected to single MSK modulation, and then it is up-sampled and expanded as a local signal, and a complex correlation operation is performed with the receiving filter signal in each symbol interval before and after the coarse synchronization moment. Since the single MSK modulated signal has only four values ​​of ±1 and ±j (where j is an imaginary unit), the floating-point multiplication and addition operation in the complex correlation can be simplified to addition, which significantly reduces the amount of calculation and complexity.

[0035] The technical solution of the present invention comprises the following steps:

[0036] (1) Single MSK modulation of the training sequence: Single MSK modulation with an initial phase of 0 is performed on a training sequence of length K. The modulated signal is recorded as L(k). The length of L(k) is K and contains the real part L I (k) and the imaginary part L Q (k) has only four possible values: ±1, ±j, where k is the modulation time, which is an integer in the range of 0 to K-1.

[0037] (2) Upsampling of single MSK signal: Upsample L(k) by U times to obtain the upsampled signal L'(n), whose length is KU, where n is the sampling time, which is an integer in the range of 0 to KU-1. The value of L'(n) is:

[0038]

[0039] (3) Expansion and shaping of upsampled signal: The upsampled signal L'(n) is expanded and shaped by M times to obtain a shaped signal L"(n), whose length is still KU, where M≤U and is an odd number. The value of L"(n) is:

[0040]

[0041] (4) Simplified complex correlation operation of received floating-point signal: Since the training sequence has good autocorrelation characteristics, the fine synchronization position cannot be more than one symbol before or after the coarse synchronization position. Therefore, according to the local signal L”(n), the coarse synchronization time T C The floating point signal S(n) is received in each symbol interval before and after, and a complex correlation operation is performed to obtain the correlation value R(n), as follows:

[0042]

[0043] Where, L"(n) * is the conjugate of L”(n). Since only some sampling points in L”(n) have values, and the values ​​are only ±1, ±j, the complex correlation operation can be simplified to include only addition operations. Assume that the real and imaginary parts of S(n) are S I (n) and S Q (n), the real and imaginary parts of the correlation value R(n) are R I (n) and R Q (n), the simplified related operations are as follows:

[0044]

[0045] Among them, the value range of sampling time n is: T C -U+1≤n≤T C +U-1, and n is an integer; where a k 、b k The way in which the multiplier on the right side participates in the accumulation operation is defined, which is fundamentally determined by L(k), as shown in the following table:

[0046] L(k) <![CDATA[a k ]]> <![CDATA[b k ]]> ﹢1 ﹢1 0 ﹣1 ﹣1 0 ﹢j 0 ﹢1 ﹣j 0 ﹣1

[0047] In the table, +1 means that the multiplier on the right side participates in the addition operation as the original number; -1 means that the multiplier on the right side participates in the addition operation after being inverted; 0 means that it does not participate in the operation. Therefore, the simplified complex correlation operation does not contain multiplication.

[0048] (5) Correlation value peak search: modulus the correlation value, as follows:

[0049]

[0050] Search for the maximum value of |R(n)| in the range, and the sampling time of the maximum value is recorded as T off , then the fine synchronization time is:

[0051]

[0052] Optimization scheme 1 of the present invention:

[0053] In order to further reduce the computational complexity of the fine synchronization method when receiving the MSK signal, in step 5 of the technical solution of the present invention, when searching for the peak value of the correlation value of the signal, the square root operation of the last modulus when calculating the correlation value at each moment is omitted, and the square value of the modulus of the correlation value is directly calculated. The optimized calculation method is as follows:

[0054] |R(n)|2 =|R I (n)| 2 +|R Q (n)| 2

[0055] Here we directly search for |R(n)| in the range 2 The maximum value of the correlation value is searched for the peak value to obtain the fine synchronization moment, which further reduces the calculation complexity of the fine synchronization module, saves the hardware resources for engineering implementation, and simplifies the overall implementation difficulty of the communication system.

[0056] Optimization scheme 2 of the present invention:

[0057] In the present invention, a flexible selection scheme for the expansion multiple M is provided. In order to comprehensively consider the fine synchronization performance and the amount of calculation, according to the subsequent simulation experimental results, it can be found that when M=3, the two can achieve a relatively good balance, and the amount of calculation is significantly reduced without losing the synchronization accuracy; if it is necessary to further save computing resources, M=1 can be taken, at which time the fine synchronization performance is slightly reduced, and the error performance is degraded by 0.5ns compared with the conventional method. Figure 4 Therefore, in the practical application of the communication system, a trade-off between synchronization accuracy and computational complexity can be made according to specific relevant indicator requirements.

[0058] In view of the huge amount of calculation in the conventional MSK signal floating-point complex correlation fine synchronization method, the present invention simplifies the floating-point complex correlation operation by constructing a simple local signal. Without losing the fine synchronization accuracy and maintaining high mobility and adaptability, the amount of calculation is significantly reduced, the complexity of fine synchronization is reduced, and the hardware resources for engineering implementation are effectively saved.

[0059] The present invention acts on the fine synchronization module. After the MSK signal is sampled at a high rate by A / D, low-pass filtered and roughly synchronized, the low-pass filtered signal within one symbol before and after the rough synchronization moment is simplified by complex correlation operation and correlation value peak search, and finally the fine synchronization moment is given. The whole processing process is as follows: Figure 1 As shown, the present invention is further described below with reference to an example. For simplicity, it is assumed that the training sequence contains 8 symbols, specifically: {1, 1, 1, 0, 1, 0, 0, 0}, the sampling rate U = 4, and M = 3. The specific implementation process of the present invention is as follows:

[0060] [Step 1]

[0061] The training sequence is modulated by single MSK with an initial phase of 0 and a modulation signal length of 8, as follows:

[0062] {j,-1,-j,-1,-j,-1,j,1}

[0063] [Step 2]

[0064] The single MSK modulated signal is upsampled, and the upsampling ratio is the A\D sampling ratio to form an upsampled signal with a length of 32, as follows:

[0065] {j,0,0,0,-1,0,0,0,-j,0,0,0,-1,0,0,0,-j,0,0,0,-1,0,0,0,j,0,0,0,1,0,0,0}

[0066] [Step 3]

[0067] The up-sampled signal is expanded and shaped, and the expansion multiple M=3 is formed to form a local signal, as follows:

[0068] {j,j,j,0,-1,-1,-1,0,-j,-j,-j,0,-1,-1,-1,0,-j,-j,-j,0,-1,-1,-1,0,j,j,j,0,1,1,1,0}

[0069] [Step 4]

[0070] Perform simplified complex correlation operation on the received floating-point signal to obtain the rough synchronization time T C The correlation value within each symbol interval before and after, for example, calculate the correlation value at sampling point n, where T C -3≤n≤T C +3, the specific calculation method is shown in the following table:

[0071]

[0072] Among them, the “—” in the table means not participating in the accumulation operation.

[0073] [Step 5]

[0074] The square of the modulo correlation value, |R(n)| 2 =|R I (n)| 2 +|R Q (n)| 2 , then at T C -3≤n≤T C Search within +3 range |R(n) 2 The maximum value of the maximum value is recorded as: T off , then the fine synchronization time is:

[0075] T r =T C +T off +1.

Claims

1. A low-complexity MSK signal fine synchronization method, characterized in that: The following steps are involved: (1) Perform single MSK modulation with an initial phase of 0 on a training sequence of length K. The modulated signal is recorded as L(k), which contains the real part L I (k) and the imaginary part L Q (k), the value can be only ±1, ±j, where k is the modulation time; (2) Upsample L(k) by a factor of U to obtain an upsampled signal L'(n), where n is the sampling time. (3) Expand and shape the up-sampled signal L'(n) by M times to obtain a shaped signal L"(n), where M≤U and is an odd number; (4) According to the local signal L”(n), the coarse synchronization time T C The floating point signal S(n) received in each of the preceding and following symbol intervals is subjected to complex correlation operation to obtain the correlation value R(n); (5) Take the modulus of the correlation value and search for the maximum value of |R(n)| within the range. The sampling time of the maximum value is recorded as T off , then the fine synchronization time 2. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In the step (2), the up-sampled signal 3. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In the step (3), the forming signal 4. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In the step (3), M=3.

5. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In the step (4), the relevant value Where L″(i) * is the conjugate of L″(i).

6. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In step (4), it is assumed that the real part and imaginary part of S(n) are S I (n) and S Q (n), the real and imaginary parts of the correlation value R(n) are R I (n) and R Q (n), then Among them, the value range of sampling time n is T C -U+1≤n≤T C +U-1;a k 、b k The way in which the multiplier on the right side participates in the accumulation operation is defined, and the value of L(k) is ﹢1, a k The value is ﹢1, b k The value is 0; L(k) is -1, a k The value is -1, b k The value is 0; L(k) takes the value of +j, a k The value is 0, b k The value is ﹢1; L(k) is ﹣j, a k The value is 0, b k The value is -1; +1 means that the multiplier on the right participates in the addition operation as the original number; -1 means that the multiplier on the right participates in the addition operation after being inverted; 0 means that it does not participate in the operation.

7. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In the step (5), the correlation value modulo 8. The low-complexity MSK signal fine synchronization method according to claim 1, characterized in that: In the step (5), when searching for the peak value of the correlation value of the signal, the square root operation of the last modulus when calculating the correlation value at each moment is omitted, and the square value of the modulus of the correlation value is directly calculated, |R(n)| 2 =|R I (n)| 2 +|R Q (n)| 2 .

Images