Improved adaptive echo cancellation system
By introducing a delay estimation module and a digital delay module in the adaptive echo cancellation system, the effective area of the echo time domain response is positioned and the DEC module is used for echo cancellation, which solves the problems of large hardware resources demand and slow convergence speed in the existing system, and efficient echo cancellation is achieved.
Patent Information
- Application Number
- CN202510462698.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-14
- Publication Date
- 2025-06-03
AI Technical Summary
When handling echo channel time domain response, existing adaptive echo cancellation systems require a large number of hardware resources and tap counts, resulting in large demand for hardware resources and slow coefficient convergence.
By introducing a delay estimation module and a digital delay module, the effective area of the echo time domain response is located, and the original DEC module is used to eliminate echoes, reducing hardware resource requirements and speeding up the convergence speed of FIR filter coefficients.
It realizes the effective echo elimination under the requirements of smaller hardware resources, improves the efficiency and performance of the echo cancellation system, saves computing resources and speeds up the convergence speed.
Smart Images
Figure CN120090660A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of communication technologies, and particularly relates to an improved adaptive echo cancellation system. Background Art
[0002] In the application of single-pair Ethernet transceivers, single-pair Ethernet needs to perform bidirectional data transmission on the same pair of wires, that is, full-duplex communication. Since signals are transmitted on the same pair of wires, the transmitted and received signals will interfere with each other to generate echoes, and the echoes will reduce the signal quality and signal-to-noise ratio, resulting in an increase in the bit error rate.
[0003] The reasons for echo generation are as Figure 1 , when the signal at the transmitting end (TX) is transmitted through the cable, reflections will be caused due to factors such as impedance mismatch and cable coupling, or crosstalk will be caused due to factors such as the insufficient isolation degree of the hybrid circuit, and part of the energy is reflected back to the receiving end (RX).
[0004] Echo cancellation algorithms can significantly reduce the interference of echoes on received signals, improve signal quality, and ensure the reliability of data transmission. It is of great significance for realizing full-duplex communication, improving signal quality, saving bandwidth, reducing hardware costs, adapting to complex environments, supporting high-speed transmission, and simplifying network deployment, and is a key technology to ensure the efficient and stable operation of single-pair Ethernet.
[0005] The core of the existing digital echo cancellation algorithm (Digital Echo Canceler, DEC) is to dynamically model the leakage path from the local transmitted signal to the receiving end, generate a "predicted echo" and subtract the echo from the received signal in real time.
[0006] Figure 2 Shown is a basic digital echo cancellation system based on a DEC module. The signal transmitted by the transmitting end is known data, and the echo can be regarded as the transmitted signal crosstalking to the receiving end through a channel; the DEC module adaptively converges through an LMS (Least Mean Square algorithm) module to obtain a set of FIR filter coefficients that can simulate the "channel", and the transmitted signal passes through the FIR filter to obtain the predicted echo signal; subtracting the predicted echo signal from the received signal can effectively cancel the echo crosstalk.
[0007] One of the key technical points of DEC is to quickly track changes in cable impedance (such as characteristic changes caused by temperature and vibration) to ensure the accuracy of echo estimation. However, in an actual system, the leakage path through which the echo crosstalks from the transmitting end to the receiving end is a sparse system in the time domain, such as Figure 3As shown, it can be seen that the main energy of its time-domain response is relatively concentrated. Since there is a certain time delay for the signal to crosstalk from the transmitting end to the receiving end, it is manifested as a long section of zero response before the concentrated area of the time-domain response energy of the echo channel. If an adaptive filter is used to completely converge the time-domain response of the echo channel, an FIR filter with a large number of taps is required, which has a very high demand for hardware resources. Moreover, most of the taps are in a wasted state. Not only that, if the number of taps is too large, it will also greatly slow down the coefficient convergence speed. Summary of the Invention
[0008] In view of the defects existing in the prior art, the present invention provides an improved adaptive echo cancellation system. Based on the traditional adaptive echo cancellation system, the method of the present invention locates the effective region of the echo time-domain response through a delay estimation module and a digital delay module, and then uses the original DEC module to cancel the echo of the received signal, which can reduce the demand for hardware resources and accelerate the convergence speed of the FIR filter coefficients.
[0009] The technical solution adopted by the present invention is as follows:
[0010] An improved adaptive echo cancellation system includes a transmitting end, a receiving end, and a DEC module arranged between the transmitting end and the receiving end;
[0011] Characterized in that it further includes a delay estimation module and a digital delay module;
[0012] The delay estimation module is arranged between the transmitting end and the receiving end, and is used to estimate the delay in the time domain of the echo channel and send the estimated delay information to the digital delay module;
[0013] The digital delay module digitally delays the transmitted signal according to the received delay information to locate the effective region of the echo time-domain response, and sends the delayed transmitted signal to the DEC module;
[0014] The DEC module cancels the crosstalk noise component in the received signal according to the delayed transmitted signal.
[0015] Further, the delay estimation module satisfies covering the time-domain response of the entire echo channel in the time domain.
[0016] Further, the delay estimation module respectively performs low-pass filtering and downsampling on the transmitted signal and the received signal, and then uses the least mean square algorithm to converge to obtain a set of coarse echo channel time-domain response coefficients h(0), h(1),...., h(n-1);
[0017] The one-dimensional discrete centroid calculation formula of the square response is used to calculate the delay clock beats Δh of the echo path at the downsampling rate:
[0018]
[0019] Among them, h(t) is the t-th filtering coefficient converged by the delay estimation module, where t = 0, 1,.... n - 1.
[0020] Furthermore, the digital delay module calculates the delay value according to the delay clock beats Δh and the downsampling ratio, and performs digital time delay on the transmitted signal according to the delay value.
[0021] Furthermore, the sampling rate of the delay estimation module is 1 / M, where M is an integer greater than 1.
[0022] Advantages of the present invention:
[0023] Based on the traditional adaptive echo cancellation system, the present invention adds a delay estimation module and a digital delay module. Among them, the delay estimation module can model the approximate path of echo leakage with very little hardware resources to preliminarily locate the effective region of the echo time-domain response; the digital delay module delays the transmitted signal at the transmitting end according to the estimated delay value and then inputs it into the FIR filter of the DEC module. At this time, the FIR filter only needs a few taps to cover only the effective region of the echo response, thereby saving computing resources and accelerating the convergence speed. Description of the Drawings
[0024] Figure 1 is the schematic diagram of echo generation;
[0025] Figure 2 is the traditional digital echo cancellation system based on the DEC module;
[0026] Figure 3 is the time-domain response diagram of the echo channel;
[0027] Figure 4 is the system framework diagram of the present invention;
[0028] Figure 5 is the detailed module diagram of the system in this embodiment;
[0029] Figure 6 is the typical echo channel response diagram;
[0030] Figure 7 is the eye diagram of the far-end signal without echo crosstalk;
[0031] Figure 8 is the eye diagram of the signal with echo crosstalk;
[0032] Figure 9 is the convergence coefficient diagram (a) of the DEC module in the traditional DEC method and the eye diagram (b) of the processed signal.
[0033] Figure 10 This is the convergence coefficient graph (a) of the DEC module and the eye diagram of the processed signal (b) in the method of the present invention.
[0034] Figure 11 It is the convergence coefficient graph in the delay estimation module. Specific implementation manner
[0035] Next, in combination with the drawings and embodiments, the technical solution of the present invention will be described.
[0036] The adaptive echo cancellation system in this embodiment, as Figure 4 shown, includes a transmitting end, a receiving end, a DEC module, a delay estimation module, and a digital delay module; the processing flow of the signal in the system is as Figure 5 shown.
[0037] The DEC module is arranged between the transmitting end and the receiving end, and includes a first FIR filter and a first LMS module.
[0038] The delay estimation module is arranged between the transmitting end and the receiving end, and is used to estimate the delay in the time domain of the echo channel and send the estimated delay value to the digital delay module.
[0039] Specifically, the delay estimation module includes two low-pass filters, two downsampling modules, a second LMS module, a second FIR filter, and a subtractor. Among them, the two low-pass filters and the two downsampling modules perform low-pass filtering and downsampling on the transmitted signal and the received signal respectively; the purpose of low-pass filtering is to prevent aliasing after downsampling.
[0040] The downsampled transmitted signal passes through n - 1 D flip-flop taps in the second FIR filter, is multiplied by the filter coefficient vector h(t) one by one, and the result is sent to the accumulator; the accumulator accumulates the received data and sends the obtained sum to the subtractor.
[0041] After the downsampled received signal is subtracted from the sum sent by the accumulator, the obtained data is sent to the second LMS module.
[0042] The second LMS module receives the data sent by the subtractor and the tap data sent by the D flip-flop, and updates the filter coefficient vector h(t) using the least mean square algorithm according to the received data. After the coefficients converge, a set of stable coarse echo channel time-domain response coefficients h(0), h(1),...., h(n - 1) are obtained and sent to the one-dimensional discrete centroid calculation module.
[0043] The one-dimensional discrete centroid calculation module calculates the delay clock beats Δh of the echo path at the downsampling rate using the one-dimensional discrete centroid calculation formula of the square response:
[0044]
[0045] Among them, h(t) is the t-th filtering coefficient converged by the delay estimation module, where t = 0, 1,.... n - 1. In this embodiment, n is taken as 25; the delay clock beats Δh are sent to the digital delay module as delay information.
[0046] The digital delay module calculates the delay value according to the delay clock beats Δh and the downsampling ratio (M is taken as 10 in this embodiment), digitally delays the transmitted signal according to the delay value, and sends the delayed transmitted signal to the DEC module.
[0047] The DEC module performs echo cancellation on the crosstalk noise component in the received signal according to the delayed transmitted signal.
[0048] To better illustrate the advantages of the present invention, a typical echo channel response as Figure 6 shown is constructed; a random PAM3 level signal is generated and Gaussian noise is added to it to simulate the signal transmitted from the far end, with an SNR = 26.2391 dB, and the eye diagram as shown in Figure 7 can be obtained.
[0049] Another group of random PAM3 level signals uncorrelated with the far-end signal is generated, and after convolution with the constructed echo channel response, it is superimposed on the simulated far-end signal to simulate echo crosstalk. The SNR of the signal after superimposing the echo crosstalk is -4.9264 dB, and the eye diagram as Figure 8 shown is obtained. It can be seen that after the signal passes through the echo crosstalk, the eye diagram can no longer distinguish the three levels at all.
[0050] The traditional DEC algorithm is used to process the echo of this signal. After the DEC module converges stably, the filtering coefficient and the eye diagram of the processed signal as Figure 9 shown are obtained. It can be seen that after the DEC converges stably, its filtering coefficient basically models the echo channel, and the two basically coincide numerically. The eye diagram of the signal is also basically restored, the SNR is increased from -4.9264 dB to 21.7064 dB, and the bit error rate is 0.
[0051] However, if the traditional DEC algorithm is to cover the entire time-domain response and adapt to the system changes caused by different environmental transformations, a large number of taps are required (at least 250 tap numbers are required to maintain a good effect); and since jitter also occurs in the silent region during the coefficient convergence process of the DEC, new noise will be generated during the process of filtering and eliminating the echo.
[0052] The system of the present invention is used to process the echo of the above signal; among them, the delay estimation module uses 25 taps for ten-fold downsampling filtering, roughly modeling the shape of the echo channel.Figure 11 This is the convergence coefficient graph obtained by the delay estimation module of the present invention. As can be seen from the graph, the delay estimation module accurately locates the position of the effective region of the echo channel.
[0053] Figure 10 Shown are the filter coefficients and the eye diagram of the processed signal obtained after the DEC module of the present invention converges stably. After estimating the delay of the echo silent region, the DEC module can model the effective region of the echo with only a few taps (only 40 taps are used for filtering), greatly saving the hardware cost, and not introducing unnecessary jitter to the silent region. The echo cancellation performance is also better, and the SNR of the signal after algorithm processing reaches 23.1448 dB.
Claims
1. An improved adaptive echo cancellation system, comprising a transmitting end, a receiving end, and a DEC module arranged between the transmitting end and the receiving end; It is characterized in that It also includes a delay estimation module and a digital delay module; The delay estimation module is arranged between the transmitting end and the receiving end, and is used to estimate the delay in the time domain of the echo channel, and send the estimated delay information to the digital delay module; The digital delay module performs digital delay on the transmission signal according to the received delay information to locate the effective area of the echo time domain response, and sends the delayed transmission signal to the DEC module; The DEC module performs echo cancellation on the crosstalk noise component in the received signal according to the delayed transmitted signal.
2. An improved adaptive echo cancellation system as claimed in claim 1, characterized in that: The delay estimation module meets the time domain response of covering the entire echo channel in the time domain.
3. An improved adaptive echo cancellation system as claimed in claim 2, characterized in that: The delay estimation module performs low-pass filtering and downsampling on the transmission signal and the reception signal respectively, and then uses the least mean square algorithm to converge to obtain a set of coarse echo channel time domain response coefficients h(0), h(1), ..., h(n-1); The one-dimensional discrete centroid calculation formula of the square response is used to calculate the delayed clock beat number Δh of the echo path under the downsampling rate: Wherein, h(t) is the t-th filter coefficient converged by the delay estimation module, t=0, 1, ....n-1.
4. An improved adaptive echo cancellation system as claimed in claim 3, characterized in that: The digital delay module calculates a delay value according to the delay clock beat number Δh and the downsampling multiple, and performs digital delay on the transmitted signal according to the delay value.
5. An improved adaptive echo cancellation system as claimed in claim 4, characterized in that: The sampling rate of the delay estimation module is 1 / M, where M is an integer greater than 1.