LoRa multi-user collision decoding method combining spectral phase and amplitude characteristics

Abstract

The invention provides a LoRa multi-user collision decoding method combining spectral phase and amplitude characteristics, and belongs to the technical field of Internet of Things communication. In order to solve the problem of high bit error rate of the existing LoRa collision decoding technology under a low signal-to-noise ratio, the invention provides two innovative schemes, namely multi-symbol phase consistency decoding and multi-frequency point consistency feature decoding, which are used in cascade with an AlignTrack scheme. And accurate affiliation judgment of the peak value of the multi-user signal is realized through the phase difference value of adjacent symbols, the peak amplitude relation of the symbols under the sliding window and the three-spectral-line structural characteristics. According to the method, under the extremely low signal-to-noise ratio, the bit error rate can be remarkably reduced, the communication efficiency is improved, and the defect of data retransmission during LoRa multi-user collision is avoided.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things LoRa communication technology and multi-user detection technology, and specifically to a method for multi-user collision decoding using the multi-dimensional characteristics of the LoRa signal demodulation spectrum (including multi-symbol spectrum phase relationships and multi-frequency amplitude structures). Background Art

[0002] At present, some IoT scenarios still need to adopt random access methods (such as direct satellite-to-ground connection), which will inevitably face the problem of collision of user data packets, resulting in the need for data retransmission, which in turn causes a serious decrease in communication efficiency. In order to avoid data retransmission, collision decoding technology at the receiving end is currently an effective way. The main problem with other existing collision decoding technology solutions is the serious performance loss under low signal-to-noise ratio. Compared with other solutions, the AlignTrack collision decoding solution proposed by Tsinghua University is currently the best performing under low signal-to-noise ratio, but its bit error rate performance under low signal-to-noise ratio is still poor (the signal-to-noise ratio is SNR-18dB / -20dB, and the bit error rate has reached 10 -1 Compared with this solution, the LoRa multi-user collision decoding method based on spectral phase and amplitude structure proposed in the present invention can reduce the overall bit error rate by one order of magnitude under the same signal-to-noise ratio. Summary of the Invention

[0003] When the receiving end receives two or more LoRa signals with the same spreading factor at the same time, a collision will occur, resulting in multiple peaks in the frequency domain. At this time, the peak attribution problem needs to be solved. This paper mainly proposes two innovative collision decoding schemes: 1. Multi-symbol phase consistency decoding scheme; 2. Multi-symbol multi-frequency consistency feature decoding scheme. In addition, we also use the AlignTrack scheme [1] that uses amplitude characteristics in cascade with these two original schemes to enhance the decoding reliability of the overall system.

[0004] The specific processing steps for each module are as follows:

[0005] a. Perform phase consistency detection:

[0006] (1) De-slope and discrete Fourier transform the received LoRa signal to obtain discrete spectrum lines;

[0007] (2) Identify the locations of multiple peak frequencies generated by collisions;

[0008] (3) Extract the peak phases of adjacent symbols and calculate the phase difference combination;

[0009] (4) Based on the characteristic that the phase difference of adjacent symbols of the same user is the same, the peak ownership is determined in combination with the noise fluctuation threshold;

[0010] b. If the phase consistency detection cannot uniquely determine the target peak, execute the AlignTrack module:

[0011] (5) Generate the spectra of the left, middle, and right windows by sliding the demodulation window;

[0012] (6) Screen using the feature that the target peak satisfies the amplitude relationship HL < H < HR, where H is the peak amplitude of the current window, and HL / HR is the amplitude of the sliding window;

[0013] c. If the AlignTrack module still cannot uniquely determine the target peak, perform multi-frequency point consistency detection:

[0014] (7) Extract the main lobe amplitude of the candidate peak and the amplitudes of the left and right adjacent Fourier interval points;

[0015] (8) Calculate the frequency offset value of each peak according to the relationship between the amplitudes of the three spectral lines and the frequency deviation;

[0016] (9) Select the peak with the smallest frequency offset as the target data packet peak.

[0017] Preferably, the calculation method of the phase difference combination in step (3) is:

[0018] For the two-user collision scenario, when each symbol contains two peaks, calculate all 4 groups of phase difference combinations:

[0019] Δφ i (1,1) = angle(F i (k i,1 )) - angle(F i+1 (k i+1,1 ))

[0020] Δφ i (1,2) = angle(F i (k i,1 )) - angle(F i+1 (k i+1,2 ))

[0021] Δφ i (2,1) = angle(F i (k i,2 )) - angle(F i+1 (k i+1,1 ))

[0022] Δφ i (2,2) = angle(F i (k i,2 )) - angle(F i+1 (k i+1,2 ))

[0023] Among them F i (k i,1 ), F i (k i,2 ) is the current symbol peak, F i+1 (k i+1,1 ), F i+1 (k i+1,2 ) is the peak value of the next symbol.

[0024] Preferably, the setting of the phase difference determination threshold in step (4) satisfies:

[0025] The relationship between the noise fluctuation δ and the signal-to-noise ratio SNR is Δδ=0.09·e -0.085·SNR ;

[0026] The actual judgment condition is |Δφ i -Δφ i+1 |<Δδ.

[0027] Preferably, the screening conditions for the target peak in step (6) are:

[0028] If and only if a single peak satisfies H L <H<H R , it is determined to be the target data packet.

[0029] Preferably, the frequency deviation value in step (8) is calculated using a three-spectrum estimation formula:

[0030]

[0031] Preferably, the target peak value in step (9) is determined based on:

[0032] The frequency deviation of the target data packet approaches zero due to the synchronization alignment operation, so the peak with the smallest Δf is selected.

[0033] The present invention provides a LoRa multi-user collision decoding system based on combined spectral phase and amplitude characteristics, comprising three modules executed in cascade:

[0034] A phase consistency detection module is configured to determine peak attribution based on the phase difference of adjacent symbols;

[0035] AlignTrack module, configured to filter peaks using a sliding window amplitude relationship;

[0036] The multi-frequency point consistency feature detection module is configured to ultimately determine the target peak through three-line frequency deviation analysis.

[0037] The beneficial effects of the present invention are as follows: combining the above three modules can comprehensively improve the bit error rate performance of the collision decoding system, which is applicable to collision scenarios with a signal-to-noise ratio as low as -20dB. The bit error rate is reduced by an order of magnitude compared to using the AlignTrack solution alone. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is the overall flow chart of LoRa multi-user collision decoding of the present invention;

[0039] Figure 2 Phase consistency detection flow chart;

[0040] Figure 3 AlignTrack module detection flow chart;

[0041] Figure 4 Multi-frequency consistency test flow chart;

[0042] Figure 5 It is the cascade diagram of the scheme;

[0043] Figure 6 It is a schematic diagram of collision amplitude frequency;

[0044] Figure 7 This is a schematic diagram of AlignTrack peak change;

[0045] Figure 8 Schematic diagram of the amplitude of three spectral lines;

[0046] Figure 9 This is a comparison chart of the bit error rate of this solution and the current best collision resolution solution AlignTrack as the signal-to-noise ratio changes. DETAILED DESCRIPTION

[0047] The following describes the embodiments of the present invention through specific examples. Those skilled in the art will readily understand the other advantages and benefits of the present invention from the disclosure herein. The present invention may also be implemented or applied through various other specific embodiments, and the details in this specification may be modified or altered based on different viewpoints and applications without departing from the spirit of the present invention.

[0048] Please refer to the accompanying drawings. It should be noted that the illustrations provided in this embodiment are merely schematic illustrations of the basic concept of the present invention. Therefore, the drawings only show components relevant to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be varied arbitrarily, and the component layout may also be more complex.

[0049] The present invention will be further described in detail below with reference to the embodiments and drawings, but the present invention is not limited thereto.

[0050] Example

[0051] Figure 1 This is the overall flow chart of the present invention, Figure 5 Shown is a schematic diagram of three-module cascade.

[0052] Assume that Packet A is the synchronized data packet, that is, the target data packet, and Packet B is the collision data packet:

[0053] like Figure 2 , 6, 7, execute the phase consistency detection module:

[0054] Execute the phase consistency detection module:

[0055] (1) Figure 1 For example, after the slope of the LoRa data packet received by the i-th demodulation window is removed, the discrete spectrum line F can be obtained by discrete Fourier transform. i (k);

[0056] (2) Search and determine the peak frequency point position k of the signal discrete spectrum i,1 and k i,2 ; (Here we take two users as an example, this method can be expanded to more users)

[0057] (3) Extract the phase of the adjacent symbol peaks and make a difference to get the phase difference value of the adjacent symbols, and obtain all the phase difference combinations Δφ i , that is (if each symbol contains two peaks, four phase difference combinations need to be calculated)

[0058] Δφ i (1,1)=angle(F i (k i,1 ))-angle(F i+1 (k i+1,1 ))

[0059] Δφ i (1,2)=angle(F i (k i,1 ))-angle(F i+1 (k i+1,2 ))

[0060] Δφ i (2,1)=angle(F i (k i,2 ))-angle(F i+1 (k i+1,1 ))

[0061] Δφ i (2,2)=angle(F i (ki,2 ))-angle(F i+1 (k i+1,2 ))

[0062] (4) Since the phase difference between any two adjacent symbols of the same user should be the same, the user peak attribution can be determined based on this. First, considering the noise-free background, we define the phase difference value of a certain user as δ, when Δφ i =Δφ i+1 =δ, it can be determined that these peaks participating in the phase difference calculation belong to the same user. Otherwise, they are determined to be peaks of different users.

[0063] (5) However, δ will fluctuate due to the influence of noise, that is, Δφ i ∈(δ-Δδ,δ+Δδ), where Δδ is the fluctuation of δ due to noise. The relationship between the fluctuation and the signal-to-noise ratio SNR is given as Δδ=0.09·e -0.08·5SNR In actual use, the SNR is measured in advance, and then the corresponding Δδ is used as the phase difference consistency judgment threshold to determine whether the phase difference of three adjacent symbols meets Δφ i ∈(δ-Δδ,δ+Δδ), that is, determine |Δφ i -Δφ i+1 |<Δδ.

[0064] (6) Accumulate the number of peak pairs that meet the judgment conditions P;

[0065] (7) If there is only one |Δφ i -Δφ i+1 If the condition is met (i.e., if P = 1), the peak pair corresponding to the phase difference will be determined as the peak of the target data packet;

[0066] (8) If the phase difference is not 1, that is, P≠1 or P=0 (neither is within the phase consistency interval), the peak attribution is determined by moving the sliding window, that is, executing the AlignTrack module.

[0067] like Figure 3 As shown, execute the AlignTrack module:

[0068] (9) Here is a brief description of the AlignTrack solution [1]: Figure 7 As shown, window w represents the demodulation window (sampling window) for DFT and dechirp, and window w is obtained by sliding window w to the left and right. L , w R , after dechirp and FFT, the multi-peak spectrum corresponding to the three windows can be output.

[0069] (10) For the target packet peak, it will satisfy H L<H<H R (H is the frequency domain peak amplitude), and the non-target packet peak does not meet the judgment condition. For a synchronization window, if and only if there is only one peak that meets the constraint condition, it is considered to match the target packet. (For detailed operation, please refer to [1])

[0070] If multiple peaks appear and satisfy H L <H<H R , or if none of them are satisfied, then enter the multi-frequency point consistency feature detection module.

[0071] like Figure 4 As shown, execute the multi-frequency point consistency feature detection module:

[0072] Since the relationship between the peak spectrum line and the adjacent spectrum lines of the same user is basically the same, such as Figure 4 As shown. We will use the relationship between the peak spectrum line amplitude and the amplitude of adjacent spectrum lines as a way to determine user peak attribution. To facilitate the use of this law, we need to convert this relationship into an indicator. Given that the commonly used frequency estimation method can use three spectrum lines to map frequency values, this invention applies the three-spectral line frequency estimation technology to multi-user collision attribution determination. The specific user peak attribution determination scheme using the three-spectral line method is as follows:

[0073] (11) When there are multiple peaks satisfying H L <H<H R Condition, or when all peaks do not meet this condition, for all candidate peaks, such as Figure 8 As shown: Extract the main lobe peak point |F i (k i,1 )|(Take the peak value of the target data packet as an example) and the amplitude values at the adjacent Fourier interval points|F i (k i,1 -1)| and|F i (k i,1 +1)|, according to the relationship between the amplitude of the three Rife spectral lines and the frequency deviation λ, the peak amplitude with the larger amplitude among the adjacent Fourier interval points on the left and right sides of the peak will be extracted, that is, max(|F i (k i,1 -1)|,|F i (k i,1 +1)|), through the peak amplitude |F i (k i,1 )| and max(|F i (k i,1 -1)|,|F i (k i,1 The relationship between +1)|) and frequency deviation λ can be used to calculate the frequency deviation corresponding to the peak structure:

[0074]

[0075] (12) For the target data packet, due to the pilot synchronization processing and window alignment operation, the peak offset of each symbol is close to 0. At this time, the peak with the smallest frequency offset is found and it is determined that this peak belongs to the target data packet. On the contrary, since the interference data packet is not completely aligned with the demodulation window, there will inevitably be additional fractional frequency offset.

[0076] (13) Combining the above three modules can improve the performance of the collision decoding system, such as Figure 9 The bit error rate is reduced by an order of magnitude compared to using the AlignTrack solution alone.

[0077] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A LoRa multi-user collision decoding method based on combined spectral phase and amplitude characteristics, characterized in that: It includes the following steps: a. Perform phase consistency detection: (1) Perform de-slope and discrete Fourier transform on the received LoRa signal to obtain discrete spectral lines; (2) Identify the positions of multiple peak frequency points generated by collisions; (3) Extract the peak phases of adjacent symbols and calculate the phase difference combinations; (4) Based on the characteristic that the phase differences of adjacent symbols of the same user are the same, combine with the noise fluctuation threshold to determine the peak attribution; b. If the phase consistency detection cannot uniquely determine the target peak, execute the AlignTrack module: (5) Generate the spectra of the left, middle, and right three windows by sliding the demodulation window; (6) Use the characteristic that the target peak satisfies the amplitude relationship HL < H < HR for screening, where H is the peak amplitude of the current window, and HL / HR is the amplitude of the sliding window; c. If the AlignTrack module still cannot uniquely determine the target peak, execute multi-frequency point consistency detection: (7) Extract the main lobe amplitude of the candidate peak and the amplitudes of the left and right adjacent Fourier interval points; (8) Calculate the frequency offset values of each peak according to the relationship between the amplitudes of the three spectral lines and the frequency deviation; (9) Select the peak with the smallest frequency offset as the target data packet peak.

2. The method according to claim 1, characterized in that The calculation method of the phase difference combination described in step (3) is: For the two-user collision scenario, when each symbol contains two peaks, calculate all 4 groups of phase difference combinations: Δφ i (1,1)=angle(F i (k i,1 ))-angle(F i+1 (k i+1,1 )) Δφ i (1,2)=angle(F i (k i,1 ))-angle(F i+1 (k i+1,2 )) Δφ i (2,1)=angle(F i (k i,2 ))-angle(F i+1 (k i+1,1 )) Δφ i (2,2)=angle(F i (k i,2 ))-angle(F i+1 (k i+1,2 )) Among them F i (k i,1 ), F i (k i,2 ) is the current symbol peak, F i+1 (k i+1,1 ), F i+1 (k i+1,2 ) is the peak value of the next symbol.

3. The method according to claim 1, characterized in that The setting of the phase difference determination threshold in step (4) satisfies: The relationship between the noise fluctuation δ and the signal-to-noise ratio SNR is Δδ=0.09·e -0.085·SNR ; The actual judgment condition is |Δφ i -φ i+1 |<Δδ.

4. The method according to claim 1, wherein The screening condition of the target peak described in step (6) is: If and only if a single peak satisfies H L <H<H R , it is determined to be the target data packet.

5. The method according to claim 1, wherein The frequency offset value calculation in step (8) uses the three-spectral line estimation formula:

6. The method according to claim 1, characterized in that The determination basis of the target peak described in step (9) is: Since the frequency offset of the target data packet approaches zero due to the synchronization alignment operation, select the peak with the smallest Δf.

7. A LoRa multi-user collision decoding system based on combined spectral phase and amplitude characteristics, characterized in that: It includes three cascaded modules: Phase consistency detection module, configured to determine the peak attribution by the phase differences of adjacent symbols; AlignTrack module, configured to screen peaks by the amplitude relationship of the sliding window; Multi-frequency point consistency feature detection module, configured to finally determine the target peak through three-spectral line frequency offset analysis.

Images