A method for time-frequency synchronization using chirp sequence

Abstract

The present invention relates to a method for time-frequency synchronization using a chirp sequence. The method comprises the following steps: sending two repeated chirp sequences; the receiving end generates a local up and down sweep frequency signal; the up and down sweep frequency signals are respectively segmented and multiplied with the received signal, and then FFT is performed to find the power spectrum peak; when the power spectrum peak exceeds the threshold, it is considered that the chirp sequence is detected, and then the power spectrum of the subsequent two segments of the signal is calculated, and the power spectrum peak and position of the subsequent two segments of the signal are obtained; according to the relationship between the peak and position of the power spectrum of the previous and next three segments, the timing deviation and frequency deviation are calculated. The performance of the present invention is independent of the starting position of the segment, and there is no problem of position ambiguity; the calculation complexity of the present invention is much lower than that of the sliding correlation; the present invention has a low delay, and the result can be obtained in one synchronization cycle.

Description

Technical Field

[0001] The invention relates to a time-frequency synchronization method in the communication field, and in particular to a method for performing time-frequency synchronization using a chirp sequence. Background Art

[0002] Time-frequency synchronization is a problem that any communication system will face. Its performance will have a decisive impact on the entire communication system. Efficient and accurate time-frequency synchronization is the prerequisite for the communication system to achieve reliable data transmission. In order to improve performance, time-frequency synchronization usually adopts a data-assisted method. The transmitter estimates the timing deviation and frequency deviation by adding the synchronization sequence known to the receiver. For different synchronization sequences, the time-frequency synchronization method is also different.

[0003] Chirp sequence is used in IEEE802.15.4a standard, and its time-frequency synchronization principle is as follows:

[0004] The baseband representation of the Chirp signal is

[0005]

[0006] Where, μ0 is the sweep rate, T is the sweep time;

[0007] The corresponding up and down sweep signals are:

[0008]

[0009]

[0010] When there is a timing deviation Δt and a frequency deviation Δf at the receiving end, the received signal is

[0011]

[0012] Multiply the received signal with the local up and down sweep signal

[0013]

[0014] The multiplication results P(t) and Q(t) are transformed into the frequency domain through FFT, and the peak positions of their power spectra are measured as F p and F q According to the derivation, the relationship between the power spectrum peak position and the timing deviation Δt and frequency deviation Δf is as follows:

[0015]

[0016] From this we can get

[0017]

[0018] However, there are two problems in implementing this principle:

[0019] 1) Position F p and F q There is ambiguity

[0020] When Δf is 0, According to the above relationship, However, there is no negative value at the peak position of the power spectrum, so According to the formula, the final estimated result is Δt=0. This is obviously inconsistent with the actual situation. This is because the peak position of the power spectrum cannot distinguish between positive and negative, and there is ambiguity.

[0021] 2) Peak value is related to timing deviation

[0022] When there is only one Chirp sequence, the timing deviation is different, and the peak value after the above calculation will also be different. When the deviation is large, because the data for FFT is not a complete Chirp sequence, it will cause the peak value to drop, which is prone to misjudgment when the signal-to-noise ratio is low. Figure 2 As shown in the figure, T is the length of a chirp sequence, the horizontal axis is different timing deviations, and the vertical axis is the peak value. The initial deviation may be at any position, resulting in a waveform in the peak value. When the peak value is small, it is easy to misjudge under low signal-to-noise ratio.

[0023] In view of the above problems, the following documents are disclosed in the prior art:

[0024] 1) A Chinese patent with publication number CN105162571A proposes a method, a receiving end device and a communication system for synchronization based on a chirp signal. The patent sends a chirp signal to generate a complex up-and-down sweep signal with the same parameters, and uses the sliding correlation of the complex up-and-down sweep signal to the received signal to determine the best sampling point, and then extracts the received chirp signal from the best sampling point based on the best sampling point, and calculates the frequency deviation and delay with the local complex up-and-down sweep signal. However, before calculating the frequency deviation and delay, sliding correlation must be performed first. Sliding correlation is a complex multiplication and addition of the local sequence length for each sampling point, which requires a lot of calculations.

[0025] 2) The Chinese patent with publication number CN101304260A proposes a synchronization method and component for a CHIRP spread spectrum communication system. The patent generates a local reference signal, multiplies the received CHIRP signal with the local reference signal, and then calculates the intensity interpolation of the two strongest frequency components in the product. The intensity interpolation can reflect the timing error, thereby generating a control signal to gradually reduce the time difference between the received signal and the local reference signal until the time difference is 0, thereby completing the synchronization of the received signal. Although the patent does not require sliding correlation, it only needs to multiply two signals, but it cannot get the timing result at one time. It needs to gradually adjust the time difference to 0 through multiple synchronization sequence cycles, and the timing synchronization time is relatively long. Summary of the invention

[0026] In view of this, the present invention proposes a method for time-frequency synchronization using a chirp sequence. The performance of the method is independent of the segment start position, there is no problem of position ambiguity, and the computational complexity and delay are lower.

[0027] To achieve the above object, the technical solution adopted by the present invention is:

[0028] A method for time-frequency synchronization using a chirp sequence comprises the following steps:

[0029] Step 1, send two repeated Chirp sequences;

[0030] Step 2, the receiving end generates local up and down sweep frequency signals;

[0031] Step 3, multiply the upper and lower sweep frequency signals by the received signal in sections, and then perform FFT to find the power spectrum peaks of the upper and lower sweep frequency signals;

[0032] Step 4: When the power spectrum peak value exceeds the threshold, it is considered that the Chirp sequence is detected. At this time, the power spectrum of the subsequent two sections of up and down sweep signals is calculated to obtain the power spectrum peak value and position of the subsequent two sections of up and down sweep signals;

[0033] Step 5, inferring the power spectrum peak positions of the real up-and-down sweep frequency signals according to the relationship between the power spectrum peaks and their positions of the three preceding and following up-and-down sweep frequency signals;

[0034] Step 6: Calculate the timing deviation and frequency deviation based on the actual power spectrum peak positions of the up and down sweep frequency signals.

[0035] Furthermore, the inference method in step 5 is:

[0036] 1) When Fq1=Fq2=Fq3 or Fp1=Fp2=Fp3:

[0037] Fq=Fq1

[0038] Fp=Fp1-N

[0039] 2) When Fp2=Fp3≠Fp1 and Fq2=Fq3≠Fq1:

[0040] If Fp2>N / 2, then Fp=Fp2-N, otherwise Fp=Fp2

[0041] If Fq2>N / 2, then Fq=Fq2-N, otherwise Fq=Fq2

[0042] 3) When Fq1=Fq2≠Fq3 and Fq1=Fq2≠Fq3:

[0043] If A1>K1*A2, then

[0044] Fp=Fp1+N

[0045] Fq=Fq1-N

[0046] If A2>K1*A1, then

[0047] Fp=Fp1

[0048] Fq=Fq1-N

[0049] If A1>K1*A2 and A2>K1*A1 are both false:

[0050] If Fp1>N / 2, then Fp=Fp1-N, otherwise Fp=Fp1

[0051] If Fq1>N / 2, then Fq=Fq1-N, otherwise Fq=Fq1

[0052] Among them, Fp1, Fp2, and Fp3 are the power spectrum peak positions of the three previous and next sweep signals, Fq1, Fq2, and Fq3 are the power spectrum peak positions of the three previous and next sweep signals, Fp is the real power spectrum peak position of the up-sweep signal, Fq is the real power spectrum peak position of the down-sweep signal, N is the length of the Chirp sequence, A1 is the sum of the power spectrum peaks of the first up- and down-sweep signals, A2 is the sum of the power spectrum peaks of the second up- and down-sweep signals, and K1 ranges from 0.5 to 0.8.

[0053] Furthermore, the calculation method in step 6 is:

[0054] Δf=(Fp+Fq) / 2

[0055] Δt=(Fp-Fq) / 2

[0056] Wherein, Fp is the peak position of the power spectrum of the real up-sweep signal, Fq is the peak position of the power spectrum of the real down-sweep signal, Δt is the timing deviation, and Δf is the frequency deviation.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] 1) The present invention adopts two Chirp symbols in series to solve the problem that the peak value changes with the timing deviation, so that the synchronization performance is independent of the timing deviation.

[0059] 2) The present invention uses two Chirp symbols in series, and can use the relationship between the power spectrum peak value and position changes for three consecutive times to eliminate the position ambiguity and quickly calculate the timing deviation and frequency deviation.

[0060] 3) The computational complexity of the present invention is greatly reduced compared to the sliding correlation method.

[0061] 4) The present invention has low latency and can obtain results in one synchronization cycle, thus meeting the real-time nature of time-frequency synchronization. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a method according to an embodiment of the present invention.

[0063] Figure 2 This is a schematic diagram of the relationship between the power spectrum peak size and timing error.

[0064] Figure 3 This is a schematic diagram of using segmented FFT to obtain the power spectrum.

[0065] Figures 4 to 7 It is a graph showing the relationship between the three power spectrum peaks and their positions and timing deviations.

[0066] Figure 8 It is another form of a graph showing the relationship between the power spectrum peaks and their positions and timing deviations. DETAILED DESCRIPTION

[0067] The following will be combined with the embodiments to clearly and completely describe the concept, technical solution advantages and technical effects of the present invention, so as to fully understand the purpose, features and effects of the present invention. It should be noted that the specific embodiments described herein are only used to explain the present invention and do not limit the present invention.

[0068] A method for time-frequency synchronization using chirp sequences, such as Figure 1 As shown, the specific steps are as follows:

[0069] Step 1, send two repeated Chirp sequences;

[0070] The discrete signal of Chirp sequence is expressed as

[0071]

[0072] Wherein, N is the length of the Chirp sequence;

[0073] Step 2: The receiving end generates local up and down sweep frequency signals, where the up sweep frequency signal is

[0074]

[0075] The lower sweep signal is

[0076]

[0077] Step 3: If Figure 3 As shown, the upper and lower sweep frequency signals are segmentally multiplied with the received signal, and then FFT is performed to find the power spectrum peak;

[0078] Step 4: When the power spectrum peak exceeds the threshold, it is considered that the Chirp sequence is detected. At this time, the power spectra of the next two signal segments are calculated to obtain the power spectrum peak and its position.

[0079] The threshold can be obtained through simulation. That is, by observing the size of the power spectrum, setting a value range that is smaller than the peak value and larger than the noise, and then selecting the best value in the range by simulating the false detection rate and missed detection rate.

[0080] In addition, the power spectrum of an FFT can be averaged as the noise value, and then multiplied by a register-adjustable K value as the threshold. The K value can also be obtained by simulation, and the simulation method is the same as the above simulation method. K is a number greater than 1, generally between 4 and 8.

[0081] Starting from the current power spectrum that exceeds the threshold, plus the power spectrum of the next two segments of the signal, the peaks and positions are recorded as:

[0082] Up sweep signal power spectrum peaks: Ap1, Ap2, Ap3

[0083] Downward sweep signal power spectrum peak: Aq1, Aq2, Aq3

[0084] The peak positions of the power spectrum of the upper sweep signal: Fp1, Fp2, Fp3

[0085] Downward sweep signal power spectrum peak position: Fq1, Fq2, Fq3

[0086] Step 5: Infer the true Fp and Fq based on the relationship between the peak values ​​and positions of the three power spectra.

[0087] The relationship between the three-segment power spectrum peaks and their positions and timing deviations is as follows: Figures 4 to 7As shown, the timing deviation in the figure refers to the deviation between the starting position of the FFT window of the first power spectrum exceeding the threshold and the starting position of the first Chirp signal.

[0088] The timing deviation can be further explained as follows:

[0089] The FFT starting position of the local up and down sweep sequence segmented power spectrum calculation is arbitrary, but if the power spectrum peak exceeds the threshold, there must be a Chirp sequence in the FFT window. Figure 8 As shown, the power spectrum peak that exceeds the threshold may occur only when the timing deviation is between -N and N.

[0090] from Figures 4 to 7 It can be seen that the rules of Aq and Ap are consistent. To improve performance, the subsequent judgment can be made by summing two terms, and let A1 = Ap1 + Aq1, A2 = Ap2 + Aq2, and A3 = Ap3 + Aq3. There are four types of timing deviation:

[0091] 1) The timing deviation is near -N, that is Figure 5 , 6 , the leftmost of 7, its characteristics are:

[0092] Fp2=Fp3, and the values ​​of Fp2 and Fp3 are near N ( Figure 6 );

[0093] Fq2=Fq3, and the values ​​of Fq2 and Fq3 are near 0 ( Figure 7 );

[0094] A2>A3( Figure 5 )

[0095] 2) The timing deviation is between -N and 0, that is, Figure 5 , 6 , the left half of 7 (not close to the leftmost side nor close to the middle), its characteristics are:

[0096] Fp1=Fp2=Fp3( Figure 6 );

[0097] Fq1=Fq2=Fq3( Figure 7 );

[0098] A2>A1 and A2>A3( Figure 5 )

[0099] 3) When the timing deviation is near 0, that is Figure 5 , 6 , 7, its characteristics are:

[0100] Fp1=Fp2, and the values ​​of Fp1 and Fp2 are near 0 ( Figure 6 );

[0101] Fq1=Fq2, and the values ​​of Fq1 and Fq2 are near 0 ( Figure 7 );

[0102] A1>A3 and A2>A3( Figure 5 )

[0103] 4) The timing deviation is between 0 and N, that is, Figure 5 , 6 , the right half of 7 (not close to the middle nor to the rightmost), its characteristics are:

[0104] Fp1=Fp2( Figure 6 );

[0105] Fq1=Fq2( Figure 7 );

[0106] A1>A2 and A1>A3( Figure 5 )

[0107] The premise for applying the above relationship is that the frequency deviation will not be too large, and will cause a single-digit position offset of Fp and Fq at most, which is satisfied in most communication environments.

[0108] Based on the above relationship, the real Fp and Fq can be inferred according to Fp1, Fp2, Fp3, Fq1, Fq2, Fq3, A1, A2, A3. The specific method is as follows:

[0109] 1) When Fq1=Fq2=Fq3 or Fp1=Fp2=Fp3:

[0110] Fq=Fq1

[0111] Fp=Fp1-N

[0112] 2) When Fp2=Fp3≠Fp1 and Fq2=Fq3≠Fq1:

[0113] If Fp2>N / 2, then Fp=Fp2-N, otherwise Fp=Fp2

[0114] If Fq2>N / 2, then Fq=Fq2-N, otherwise Fq=Fq2

[0115] 3) When Fq1=Fq2≠Fq3 and Fq1=Fq2≠Fq3:

[0116] If A1>K1*A2, then

[0117] Fp=Fp1+N

[0118] Fq=Fq1-N

[0119] If A2>K1*A1, then

[0120] Fp=Fp1

[0121] Fq=Fq1-N

[0122] otherwise

[0123] If Fp1>N / 2, then Fp=Fp1-N, otherwise Fp=Fp1

[0124] If Fq1>N / 2, then Fq=Fq1-N, otherwise Fq=Fq1

[0125] K1 is a decimal less than 1 and greater than 0, and is generally between 0.5 and 0.8. The specific value can be selected by simulating the missed detection rate and the false detection rate to find the best value.

[0126] Step 6: According to the real Fp and Fq, get the final timing deviation and frequency deviation:

[0127]

[0128] In summary, the performance of the present invention is independent of the starting position of the segment, and there is no problem of position ambiguity; the computational complexity of the present invention is much lower than that of the sliding correlation; the present invention has a low delay, and the result can be obtained in one synchronization cycle.

Claims

1. A method for time-frequency synchronization using a chirp sequence, characterized in that: The following steps are involved: Step 1, send two repeated Chirp sequences; Step 2, the receiving end generates local up and down sweep frequency signals; Step 3, multiply the upper and lower sweep frequency signals by the received signal in sections, and then perform FFT to find the power spectrum peaks of the upper and lower sweep frequency signals; Step 4: When the power spectrum peak value exceeds the threshold, it is considered that the Chirp sequence is detected. At this time, the power spectrum of the subsequent two sections of up and down sweep signals is calculated to obtain the power spectrum peak value and position of the subsequent two sections of up and down sweep signals; Step 5: Infer the power spectrum peak positions of the real up and down sweep signals according to the relationship between the power spectrum peaks and their positions of the three previous and next sweep signals. The inference method is: 1) When Fq1=Fq2=Fq3 or Fp1=Fp2=Fp3: Fq=Fq1, Fp=Fp1-N, 2) When Fp2=Fp3≠Fp1 and Fq2=Fq3≠Fq1: If Fp2>N / 2, then Fp=Fp2-N, otherwise Fp=Fp2, If Fq2>N / 2, then Fq=Fq2-N, otherwise Fq=Fq2, 3) When Fp1=Fp2≠Fp3 and Fq1=Fq2≠Fq3: If A1>K1*A2, then: Fp=Fp1+N, Fq=Fq1-N, If A2>K1*A1, then: Fp=Fp1, Fq=Fq1-N, If A1>K1*A2 and A2>K1*A1 are both false: If Fp1>N / 2, then Fp=Fp1-N, otherwise Fp=Fp1, If Fq1>N / 2, then Fq=Fq1-N, otherwise Fq=Fq1, Among them, Fp1, Fp2, and Fp3 are the power spectrum peak positions of the three previous and next sweep signals, Fq1, Fq2, and Fq3 are the power spectrum peak positions of the three previous and next sweep signals, Fp is the real power spectrum peak position of the up-sweep signal, Fq is the real power spectrum peak position of the down-sweep signal, N is the length of the Chirp sequence, A1 is the sum of the power spectrum peaks of the first up-sweep signal and the down-sweep signal, A2 is the sum of the power spectrum peaks of the second up-sweep signal, and K1 ranges from 0.5 to 0.8; Step 6: Calculate the timing deviation and frequency deviation based on the actual power spectrum peak positions of the up and down sweep frequency signals.

2. The method for time-frequency synchronization using a chirp sequence according to claim 1, characterized in that: The calculation in step 6 is: Δf=(Fp+Fq) / 2, Δt=(Fp-Fq) / 2, Wherein, Fp is the peak position of the power spectrum of the real up-sweep signal, Fq is the peak position of the power spectrum of the real down-sweep signal, Δt is the timing deviation, and Δf is the frequency deviation.

Images