Clock drift correction method based on maximum likelihood estimation and linear interpolation

Abstract

A clock drift correction method based on maximum likelihood estimation and linear interpolation, under line-of-sight transmission, periodically collects channel impulse response (CIR) data at a position with a known distance between transceivers to obtain sampling values ​​of clock drift, and then corrects the clock drift generated in the actual measurement process. The present invention takes into account the limited resolution of the actual measurement system, adopts maximum likelihood estimation to obtain clock drift sampling values, and uses linear interpolation method to perform piecewise linear modeling on clock drift, thereby realizing accurate correction of clock drift.

Description

Technical Field

[0001] The invention relates to a technology in the field of wireless communication, in particular to a clock drift correction method based on maximum likelihood estimation and linear interpolation. Background Art

[0002] In order to support wireless communication research, channel measurement is needed to obtain the original channel data. One of the important channel measurement methods is the time domain channel measurement method. In the time domain channel measurement, the transmitter and the receiver often use independent rubidium clocks to provide clock references. Due to hardware limitations, the frequencies of the rubidium clocks at the transmitter and receiver cannot be completely consistent. In addition, the phase noise introduced by many components in the measurement system leads to clock drift during the measurement process. In order to obtain accurate multipath component delay measurement results, an accurate and effective clock drift correction algorithm is urgently needed. Summary of the invention

[0003] In view of the fact that the recording of time information in the prior art depends on the clock source node, the resolution of the time information is limited by the resolution of the clock source node and cannot meet the ultra-high time resolution requirement in channel measurement, the present invention proposes a clock drift correction method based on maximum likelihood estimation and linear interpolation. The method takes into account the limited resolution of the actual measurement system, adopts maximum likelihood estimation to obtain the clock drift sampling value, and uses linear interpolation to perform piecewise linear modeling on the clock drift, thereby realizing accurate correction of the clock drift.

[0004] The present invention is achieved through the following technical solutions:

[0005] The present invention relates to a clock drift correction method based on maximum likelihood estimation and linear interpolation, comprising:

[0006] Step 1: Clock drift sampling: In the case of line-of-sight transmission, periodically collect channel impulse response (CIR) data at a known distance between transceivers to obtain a sampling value of clock drift, specifically:

[0007] 1.1 Based on CIR data, roughly estimate the delay of the line-of-sight transmission path Where: Δτ is the time domain resolution of the measurement system, and the index j corresponding to the maximum CIR value measured is argmax k |h[k]|, h[k] is the path gain at the kth CIR sampling point;

[0008] 1.2 Use maximum likelihood estimation to obtain an accurate estimate of the line-of-sight transmission path when sampling at the reference position for the i-th time Where: The measured frequency domain response is the discrete Fourier transform, the frequency domain sampling point f k=f1+(k-1)Δf, f1 is the starting frequency of the measured frequency band, Δf is the frequency domain resolution of the measurement system, is the search interval.

[0009] 1.3 Based on the accurate estimation of the measured line-of-sight transmission path delay and the theoretical value of the line-of-sight transmission path delay calculated according to the actual transceiver distance, the sampling value of the clock drift is obtained Where: τ Theo is the theoretical value of the line-of-sight transmission path delay.

[0010] Step 2: Correct clock drift: Correct the clock drift generated during the measurement process, specifically:

[0011] 2.1 Based on the linear interpolation method, for the channel impulse response measured at any time t, predict the clock drift involved in: t i and t j is the time between the i-th and j-th sampling at the reference position, Δτ i and Δτ j is the clock drift between the i-th and j-th sampling times, and For floor and ceiling.

[0012] 2.2 Correction of the channel impulse response measured at time t Where: ⊙ is the Hadamard product between matrices of the same size, that is, the element-by-element product, F = [f1, f1 + Δf, f1 + 2Δf, ..., f1 + (k-1) Δf, ..., f1 + (K-1) Δf] is a vector including the frequencies corresponding to each frequency domain sampling point, K is the measured CIR length, is the inverse discrete Fourier transform.

[0013] Technical Effects

[0014] The present invention breaks through the time resolution limitation brought by system hardware by clock drift sampling based on maximum likelihood estimation and clock drift correction based on linear interpolation, and can accurately estimate the system clock drift value; through clock drift correction based on linear interpolation, CIR data with accurate absolute delay measurement is obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the system of the present invention;

[0016] Figure 2 It is a flow chart of the present invention;

[0017] Figure 3 It is a schematic diagram of the effect of the embodiment. DETAILED DESCRIPTION

[0018] like Figure 1 As shown, this embodiment relates to a clock drift correction system based on maximum likelihood estimation and linear interpolation for wireless channel measurement, including: a clock drift sampling unit and a clock drift correction unit, wherein: the clock drift sampling unit obtains the system clock drift value at a certain moment based on the maximum likelihood estimation method according to the CIR data collected at the fixed transceiver position, the clock drift correction unit obtains the clock drift prediction value of the system at any moment according to the clock drift value recorded by the clock drift sampling unit by using the linear interpolation method, and performs clock drift correction on the collected CIR data to obtain accurate channel measurement results.

[0019] After specific practical experiments, such as Figure 2 As shown in the figure, in a typical wireless channel measurement scenario, the channel impulse response data is obtained using a channel measurement system based on the correlation method. The measurement center frequency is 12 GHz, the bandwidth is 1.536 GHz, and the number of channel impulse response sampling points is 2048. To accurately measure the clock drift, the transmitter and receiver are directly connected via an RF cable. The channel impulse response is recorded once per second, and the delay of the strongest multipath component in the channel impulse response is used as a measurement. The difference between this value and the delay of the strongest multipath recorded for the first time is the true value of the clock drift.

[0020] In order to verify the accuracy of different methods, the clock drift records every 30 seconds are taken as reference values. The clock drift is predicted using the existing linear model method and linear interpolation method in the literature, as well as the method based on maximum likelihood estimation and linear interpolation proposed in this invention. The clock drift is compared with the true value, and the clock drift error is used as the evaluation criterion E. Δτ =|Δτ P -Δτ T |, where: Δτ P -Δτ T are the predicted value and true value of the clock drift error respectively.

[0021] As shown in Table 1, there is a comparison of the clock drift errors of the existing methods and the present invention. It can be clearly seen that the linear model method among the existing methods has the worst performance. This is because the clock drift mostly conforms to piecewise linearity, so the accuracy of the linear model is limited. The linear interpolation method has a better effect, but the accuracy is limited due to the limited time domain resolution of the measurement system (0.65ns in this embodiment). The method based on maximum likelihood estimation and linear interpolation proposed in this invention significantly improves the prediction accuracy of clock drift. Specifically, the average error of the linear model method is 0.41ns, the average error of the linear interpolation method is 0.08ns, and the average error of the method proposed in this invention is 0.01ns.

[0022] Table 1 The actual value of clock drift and the predicted value of different models

[0023]

[0024]

[0025] Compared with the existing technology, this method realizes accurate sampling of clock drift through the maximum likelihood estimation method based on CIR data. It has high estimation accuracy and is not limited by hardware. It then predicts and corrects the clock drift through linear interpolation to obtain channel measurement data with accurate delay measurement results. It can be fully applicable to clock drift correction in wireless channel measurement.

[0026] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.

Claims

1. A clock drift correction method based on maximum likelihood estimation and linear interpolation for wireless channel measurement, characterized in that: include: Step 1: Clock drift sampling: In the case of line-of-sight transmission, channel impulse response data is periodically collected at a location with a known distance between transceivers to obtain a sampling value of clock drift, specifically including: 1.1 Based on CIR data, roughly estimate the delay of the line-of-sight transmission path Where: Δτ is the time domain resolution of the measurement system, and the index j corresponding to the maximum CIR value measured is argmax k |h[k]|, h[k] is the path gain at the kth CIR sampling point; 1.2 Use maximum likelihood estimation to obtain an accurate estimate of the line-of-sight transmission path when sampling at the reference position for the i-th time Where: The measured frequency domain response is the discrete Fourier transform, the frequency domain sampling point f k =f1+(k-1)Δf, f1 is the starting frequency of the measured frequency band, Δf is the frequency domain resolution of the measurement system, is the search interval; 1.3 Based on the accurate estimation of the measured line-of-sight transmission path delay and the theoretical value of the line-of-sight transmission path delay calculated according to the actual transceiver distance, the sampling value of the clock drift is obtained Where: τ Theo is the theoretical value of the line-of-sight transmission path delay; Step 2: Correct clock drift: Correct the clock drift generated during the measurement process, including: 2.1 Based on the linear interpolation method, for the channel impulse response measured at any time t, predict the clock drift involved in: t i and t j is the time between the i-th and j-th sampling at the reference position, Δτ i and Δτ j is the clock drift between the i-th and j-th sampling times, and for floor and ceiling rounding; 2.2 Correction of the channel impulse response measured at time t Where: ⊙ is the Hadamard product between matrices of the same size, that is, the element-by-element product, F = [f1, f1 + Δf, f1 + 2Δf, ..., f1 + (k-1) Δf, ..., f1 + (K-1) Δf] is a vector including the frequencies corresponding to each frequency domain sampling point, K is the measured CIR length, is the inverse discrete Fourier transform.

2. A clock drift correction system for implementing the method of claim 1, characterized in that: include: A clock drift sampling unit and a clock drift correction unit, wherein: the clock drift sampling unit obtains the system clock drift value at a certain moment based on the CIR data collected at the fixed transceiver position based on the maximum likelihood estimation method; the clock drift correction unit obtains the clock drift prediction value of the system at any moment based on the clock drift value recorded by the clock drift sampling unit using the linear interpolation method, and performs clock drift correction on the collected CIR data to obtain accurate channel measurement results.

Images