A method of suppressing low frequency noise
By using a root-raised cosine filter and spectrum inversion technology in combination at the transmitting and receiving ends, the problem of increased cost and complexity of high-pass filters in existing technologies is solved, achieving low-cost and efficient low-frequency noise suppression.
Patent Information
- Application Number
- CN202411528509.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-30
- Publication Date
- 2025-11-21
- Estimated Expiration
- 2044-10-30
AI Technical Summary
In existing technologies, the introduction of high-pass filter devices increases costs and alters the existing architecture. High-pass filters damage the original service signals, DC balanced coding increases the complexity of the DSPs at the transmitting and receiving ends, and filtering algorithms are complex and lack flexibility.
At the transmitting end, the original service signal is converted into a digital signal, and upsampling and shaping filtering are performed using a root-raised cosine filter to perform spectrum inversion operation; at the receiving end, matched filtering and downsampling are performed to achieve spectrum inversion and RRC filtering operation.
It reduces the cost of communication systems without changing the existing system architecture, minimizes damage to original service signals, has low algorithm complexity, high flexibility, and effectively suppresses low-frequency noise.
Smart Images
Figure CN119382802B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of optical communication technology, and particularly to a method for suppressing low-frequency noise. BACKGROUND
[0002] With the introduction of diversified multimedia and data services such as Internet of Things, cloud computing, remote surgery, and the rapid development of 5G communication networks, the global network data traffic has experienced an explosive growth. The explosive growth of data traffic naturally poses a major challenge to data center optical interconnection. Intensity modulation direct detection (IMDD) is widely used in high-speed short-range optical interconnection scenarios such as data center optical interconnection due to its low cost, low power consumption, simple configuration, high reliability, etc. In addition, the pulse amplitude modulation (PAM) format has been studied for use in improving the transmission capacity of intensity modulation direct detection (IMDD) systems. However, the damage caused by multipath interference to the IMDD transmission system based on pulse amplitude modulation cannot be ignored; among them, the multipath interference is caused by multiple reflections of optical signals when passing through contaminated fiber connectors, and the noise generated by multipath interference belongs to low-frequency noise (MHz level).
[0003] In recent years, various solutions have been proposed at the sending end and the receiving end to alleviate the low-frequency noise caused by multipath interference in the IMDD transmission system, such as:
[0004] Since multipath interference noise is low-frequency noise, ranging from megahertz, conventional digital signal processing (DSP) filters require thousands of taps, leading to excessive DSP power consumption. To address this, YJ Wen, Y. Cui, and Y. Bai, in their paper "Mitigation of optical multipath interference impact for directly detected PAMn system," proposed adding an analog high-pass filter (a first-order RC analog high-pass filter, consisting of a resistor and a capacitor) after the photodetector (PD). This eliminates the need for complex DSP filtering algorithms and reduces DSP power consumption. However, because multipath interference noise overlaps with the low-frequency components of the signal, directly filtering out the low-frequency components with a high-pass filter can also filter out the low frequencies of the original service signal. N. Cheng, in his paper "Optical Multipath Interference Mitigation for PAM4 Transmission Using Line Coding and High-pass," further addressed this issue. In the "Filtering" section, to mitigate the signal impairment caused by high-pass filtering, DC balanced coding is proposed at the transmitting end. DC balanced coding divides the signal sequence into blocks and encodes each block so that the DC is zero. In terms of spectrum, after DC balanced coding, the zero-frequency component of the signal is eliminated, and the energy of the low-frequency part of the signal is transferred to other frequency ranges. In this way, the overlap between the low-frequency part of the original service signal and the low-frequency noise generated by multipath interference is reduced, thus reducing the impairment caused by high-pass filtering.
[0005] In their paper "Optical Multipath Interference Mitigation for High-Speed PAM4 IMDD Transmission System", C. Huang et al. proposed two filtering algorithms: Algorithm 1 removes the amplitude shift caused by low-frequency noise through sliding window filtering; Algorithm 2 reconstructs multipath interference noise using decision values through a sliding window and subtracts it, thus mitigating the low-frequency noise caused by multipath interference in the IMDD transmission system.
[0006] In summary, the shortcomings of the existing technology are as follows:
[0007] (1) The scheme of adding an analog filter after the photodetector introduces additional components, increases the cost of components, and changes the existing architecture; in addition, the scheme of using a high-pass filter alone will cause significant damage to the original service signal while filtering out low-frequency noise caused by multipath interference.
[0008] (2) The use of DC balanced encoding at the transmitting end increases the complexity of the DSP at both the transmitting and receiving ends;
[0009] (3) The size of the sliding window in the filtering algorithm is often hundreds or thousands, which makes the algorithm complex. It also requires decision values and optimization weighting factors, and has poor flexibility for randomly changing noise. Summary of the Invention
[0010] Therefore, the technical problem to be solved by the present invention is to overcome the problems of the introduction of high-pass filter devices in the prior art, which increases the cost of devices and changes the existing architecture; at the same time, the use of high-pass filter alone will damage the original service signal; the use of DC balanced coding at the transmitting end will increase the complexity of the DSP at the transmitting and receiving ends; and the use of filtering algorithms will result in high algorithm complexity and poor flexibility in the face of randomly changing noise.
[0011] To address the aforementioned technical problems, this invention provides a method for suppressing low-frequency noise, comprising:
[0012] At the transmitting end, the original service signal is input and converted into a digital signal to obtain a digital signal sequence. Based on a preset upsampling factor, the digital signal sequence is upsampled to obtain an upsampled digital signal sequence. A root-raised cosine filter is used to shape and filter the upsampled digital signal sequence to obtain a filtered digital signal sequence. A spectrum inversion operation is performed on the filtered digital signal sequence, wherein the spectrum inversion operation involves inverting all odd or even bits in the filtered digital signal sequence to obtain an inverted digital signal sequence.
[0013] In the channel, an inverted digital signal sequence is transmitted, while low-frequency noise signals are introduced simultaneously;
[0014] At the receiving end, after performing a spectrum inversion operation on the digital signal sequence that has introduced low-frequency noise, a matched filtering operation is performed using the same root-raised cosine filter as at the transmitting end to obtain a new digital signal sequence. After downsampling the new digital signal sequence, the final digital signal is obtained and output.
[0015] Preferably, the step of performing a spectrum inversion operation on the filtered digital signal sequence includes inverting all odd or even bits in the filtered digital signal sequence to obtain an inverted digital signal sequence.
[0016] Select the filtered digital signal sequence Invert the odd-numbered bits; for the filtered digital signal sequence Reverse the odd-numbered positions to obtain the reversed sequence. Its expression is:
[0017] ;
[0018] in, express Odd number of digits; This indicates the sequence number in the filtered digital signal sequence.
[0019] Preferably, the step of using a root-raised cosine filter to perform a shaping filter operation on the upsampled digital signal sequence to obtain a filtered digital signal sequence includes:
[0020] By using convolution and a root-raised cosine filter, each symbol in the upsampled digital signal sequence is filtered to obtain the filtered digital signal sequence. Its expression is:
[0021] ;
[0022] in, This represents the upsampled digital signal sequence; This represents the impulse response of the root-raised cosine filter;
[0023] The bandwidth of the filtered digital signal sequence is limited to the bandwidth parameter value of the root-raised-cosine filter, that is:
[0024] The Nyquist bandwidth of a digital signal is obtained based on its symbol rate, and its expression is as follows:
[0025] ;
[0026] in, The Nyquist bandwidth of a digital signal; The symbol rate of a digital signal;
[0027] Based on the Nyquist bandwidth of the digital signal and the roll-off factor of the root-raised cosine filter, the bandwidth parameter value of the root-raised cosine filter is calculated, and its expression is as follows:
[0028] ;
[0029] in, This represents the limit value for the spectral bandwidth of a digital signal; The Nyquist bandwidth of a digital signal; This represents the roll-off factor of the root-raised cosine filter.
[0030] Preferably, the roll-off factor of the root-raised cosine filter is adjusted based on the frequency domain response of the root-raised cosine filter. The value of is used to ensure the required low-frequency noise filtering bandwidth.
[0031] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:
[0032] (1) The method for suppressing low-frequency noise described in this invention can adopt a joint equalization scheme of the transmitting end and the receiving end when there is low-frequency noise interference in the channel, namely spectrum inversion and RRC filtering operation, without the need to introduce additional devices, reducing the cost of the communication system and without changing the existing system architecture.
[0033] (2) The method for suppressing low-frequency noise described in this invention involves converting the original service signal into a digital signal at the transmitting end, and then using an RRC filter to perform filtering and shaping operations on the upsampled digital signal. The high-frequency components in the resulting filtered digital signal are suppressed. After performing a spectrum inversion operation, the low-frequency components are swapped with the high-frequency components, resulting in a very small low-frequency component in the final signal spectrum characteristics at the transmitting end. Since the spectrum characteristics of noise are more prominent in the low-frequency components, the overlap between the final signal and the noise spectrum at the transmitting end is small after passing through a channel with low-frequency noise interference. Therefore, it can be concluded that the overlap between the original service signal and the noise spectrum is small. At the receiving end, the inverted digital signal and noise are simultaneously subjected to spectrum inversion. Low-frequency noise is transformed into high-frequency noise, and then RRC matched filtering is performed again, which is equivalent to performing a low-pass filtering operation on the entire received sequence. Since the low-frequency noise has become high-frequency noise after spectrum inversion, low-frequency noise interference in the channel can be successfully filtered out. Because the spectrum overlap between the original service signal and the noise is small, the damage to the original service signal is relatively small during the process of filtering out low-frequency noise interference in the channel. The algorithm has low complexity and high flexibility in suppressing low-frequency noise. Even for randomly changing low-frequency noise, the method can still have a good suppression effect. Attached Figure Description
[0034] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein:
[0035] Figure 1 This is a schematic diagram of multipath interference generation;
[0036] Figure 2 This is a flowchart of a method for suppressing low-frequency noise provided by the present invention;
[0037] Figure 3 This is a schematic diagram of a spectrum inversion operation;
[0038] Figure 4 This is a schematic diagram showing the swapping of low-frequency and high-frequency components of a digital signal after a spectrum inversion operation.
[0039] Figure 5This is a schematic diagram of a method for suppressing low-frequency noise based on the present invention in suppressing multipath interference. Detailed Implementation
[0040] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.
[0041] like Figure 1 As shown, multipath interference is caused by multiple reflections of the optical signal when passing through a contaminated fiber optic connector; the reflected signal has random optical delay and random optical attenuation relative to the original signal. After square-law detection by a photodetector (PD), the phase noise is converted into intensity noise, resulting in a large bit error rate (BER). JL Gimlett and NK Cheung have demonstrated in their paper "Effects of phase-to-intensity noise conversion by multiple reflections on gigabit-per-second DFB laser transmission systems" that the noise generated by multipath interference is low-frequency noise (MHz level). Pulse amplitude modulation formats with different levels have different tolerances to multipath interference; the higher the level, the lower the tolerance to multipath interference.
[0042] To address the low-frequency noise generated by multipath interference, this invention provides a method for suppressing low-frequency noise, the specific process of which is described below. Figure 2 As shown, it includes:
[0043] S1: At the transmitting end, the original service signal is input and converted into a digital signal to obtain a digital signal sequence; in a specific embodiment of the present invention, the digital signal sequence can be a PAM4 sequence;
[0044] Based on a preset upsampling factor, an upsampling operation is performed on the digital signal sequence to obtain an upsampled digital signal sequence; wherein, the sampling rate of the digital signal sequence is increased to a multiple of the sampling rate through upsampling; in a specific embodiment of the present invention, the preset upsampling factor can be 2, that is, the sampling rate of the digital signal sequence is increased to twice the sampling rate;
[0045] Using a root-raised cosine filter, a shaping filter is applied to the upsampled digital signal sequence to obtain the filtered digital signal sequence, including:
[0046] By using convolution and a root-raised cosine filter, each symbol in the upsampled digital signal sequence is filtered to obtain the filtered digital signal sequence. Its expression is:
[0047] ;
[0048] in, This represents the upsampled digital signal sequence; This represents the impulse response of the root-raised cosine filter;
[0049] The bandwidth of the filtered digital signal sequence is limited to the bandwidth parameter value of the root-raised-cosine filter, that is:
[0050] The Nyquist bandwidth of a digital signal is obtained based on its symbol rate, and its expression is as follows:
[0051] ;
[0052] in, The Nyquist bandwidth of a digital signal; The symbol rate of a digital signal;
[0053] Based on the Nyquist bandwidth of the digital signal and the roll-off factor of the root-raised cosine filter, the bandwidth parameter value of the root-raised cosine filter is calculated, and its expression is as follows:
[0054] ;
[0055] in, This represents the limit value for the spectral bandwidth of a digital signal; The Nyquist bandwidth of a digital signal; This represents the roll-off factor of the root-raised cosine filter;
[0056] The roll-off factor of the root-raised cosine filter is 0.5; a suitable value is selected based on the bandwidth and transmission rate of the actual system. optimization, It should be as large as possible; in practical applications, the roll-off factor of the root-raised cosine filter is adjusted based on its frequency domain response. The value ensures the required low-frequency noise filtering bandwidth, with the relative intensity noise within the required bandwidth being less than -145 dB / Hz. Typically, this applies to a 56 GHz signal. At this point, the filtering bandwidth for low-frequency noise is approximately 4 GHz;
[0057] Performing a spectrum inversion operation on the filtered digital signal sequence, wherein the spectrum inversion operation involves inverting all odd or even bits in the filtered digital signal sequence to obtain an inverted digital signal sequence, including:
[0058] Preset selection of filtered digital signal sequence Invert the odd-numbered bits; for the filtered digital signal sequence Reverse the odd-numbered positions to obtain the reversed sequence. Its expression is:
[0059] ;
[0060] in, express Odd number of digits; This indicates the sequence number in the filtered digital signal sequence; for details, please refer to... Figure 3 As shown, Figure 3 All odd-numbered positions in the signal sequence are represented by their corresponding negative sign values;
[0061] S2: In the channel, the inverted digital signal sequence is transmitted, and low-frequency noise signals are introduced at the same time;
[0062] S3: At the receiving end, after performing a spectrum inversion operation on the digital signal sequence that has introduced low-frequency noise, a matched filtering operation is performed using the same root-raised cosine filter as at the transmitting end to obtain a new digital signal sequence; after downsampling the new digital signal sequence, the final digital signal is obtained and output.
[0063] In one specific embodiment of the present invention, the upsampling factor is selected as 2, the roll-off factor of the root-raised cosine filter is 0.5, the symbol rate of the digital signal is 56 Gbaud, and the Nyquist bandwidth is 28 GHz. Then the spectral bandwidth of the filtered digital signal will be limited to 28 * (1 + 0.5) = 42 GHz.
[0064] Figure 4 This visually demonstrates the effect of the spectrum inversion operation; that is, after the spectrum inversion operation, the overall spectrum is inverted, as shown below. Figure 4 As shown by the dark lines in the text, and in Figure 4 The results clearly show that the spectrum is also limited to 42 GHz.
[0065] based on Figure 4 It is known that using the spectrum inversion operation will swap the low-frequency and high-frequency components of the digital signal; after the root-raised cosine filter (RRC) shapes and filters the digital signal sequence, it will limit the spectral bandwidth of the digital signal to a certain range. Therefore, the root-raised cosine filter is equivalent to a low-pass filter.
[0066] In summary, based on the overall analysis, we can conclude that:
[0067] At the transmitting end, after performing RRC filtering and shaping, the high-frequency components of the digital signal are suppressed. After performing a spectrum inversion operation, the low-frequency and high-frequency components are swapped. This means that the final signal spectrum at the transmitting end has very low-frequency components, as shown in the image. Figure 4 The dark lines indicate that after passing through a channel with low-frequency noise interference, the low-frequency portion of the signal overlaps with the noise; however, due to the previous RRC filtering and spectrum inversion operation at the transmitting end, the original service signal has a smaller spectral overlap with the noise.
[0068] At the receiving end, a spectrum inversion is first performed, inverting the spectrum of both the signal and noise together, thus turning low-frequency noise into high-frequency noise. Then, an RRC matched filter is performed, which is equivalent to performing a low-pass filter operation on the entire received sequence. Moreover, the low-frequency noise has become high-frequency noise after the spectrum inversion, thus successfully filtering out low-frequency noise interference in the channel.
[0069] Taking the suppression of multipath interference as an example, the specific process is as follows: Figure 5 As shown, Figure 5 In the process, an additional fiber 2 is added to the fiber optic link, which is used only to simulate the transmission of reflected signals under multipath effects to verify the effectiveness of the proposed scheme. A delay device is added to simulate the random optical delay of the reflected signal, and an adjustable optical attenuator 1 is added to simulate the random optical attenuation of the reflected signal.
[0070] based on Figure 5 At the transmitting end, a four-level pulse amplitude modulation (PAM4) sequence is generated, upsampled to a multiple sampling rate, shaped and filtered by a root-raised cosine (RRC) filter, and then all odd / even bits of the signal sequence are inverted to perform a spectrum inversion operation; through a simulated multipath interference link, the low-frequency part of the signal overlaps with the multipath interference noise spectrum.
[0071] based on Figure 5 At the receiving end, the odd / even bits of the signal sequence are inverted again to repeat the spectrum inversion process, resulting in an overall spectrum inversion. The signal is then shaped and filtered by a root-raised cosine matched filter to suppress high-frequency noise due to its low-pass filtering characteristics. Finally, downsampling and bit error rate calculation are performed.
[0072] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.
Claims
1. A method for suppressing low-frequency noise, characterized in that, include: At the transmitting end, the original service signal is input and converted into a digital signal to obtain a digital signal sequence. Based on a preset upsampling factor, the digital signal sequence is upsampled to obtain an upsampled digital signal sequence. A root-raised cosine filter is used to shape and filter the upsampled digital signal sequence to obtain a filtered digital signal sequence. A spectrum inversion operation is performed on the filtered digital signal sequence, wherein the spectrum inversion operation involves inverting all odd or even bits in the filtered digital signal sequence to obtain an inverted digital signal sequence. In the channel, an inverted digital signal sequence is transmitted, while low-frequency noise signals are introduced simultaneously; At the receiving end, after performing a spectrum inversion operation on the digital signal sequence that has introduced low-frequency noise, a matched filtering operation is performed using the same root-raised cosine filter as at the transmitting end to obtain a new digital signal sequence. After downsampling the new digital signal sequence, the final digital signal is obtained and output.
2. The method for suppressing low-frequency noise according to claim 1, characterized in that, The step of performing a spectrum inversion operation on the filtered digital signal sequence, wherein the spectrum inversion operation involves inverting all odd or even bits in the filtered digital signal sequence to obtain an inverted digital signal sequence, includes: Select the filtered digital signal sequence Invert the odd-numbered bits; for the filtered digital signal sequence Reverse the odd-numbered positions to obtain the reversed sequence. Its expression is: ； in, express Odd number of digits; This indicates the sequence number in the filtered digital signal sequence.
3. The method for suppressing low-frequency noise according to claim 1, characterized in that, The step of using a root-raised cosine filter to perform a shaping filter operation on the upsampled digital signal sequence to obtain a filtered digital signal sequence includes: By using convolution and a root-raised cosine filter, each symbol in the upsampled digital signal sequence is filtered to obtain the filtered digital signal sequence. Its expression is: ； in, This represents the upsampled digital signal sequence; This represents the impulse response of the root-raised cosine filter; The bandwidth of the filtered digital signal sequence is limited to the bandwidth parameter value of the root-raised-cosine filter, that is: The Nyquist bandwidth of a digital signal is obtained based on its symbol rate, and its expression is as follows: ； in, The Nyquist bandwidth of a digital signal; The symbol rate of a digital signal; Based on the Nyquist bandwidth of the digital signal and the roll-off factor of the root-raised cosine filter, the bandwidth parameter value of the root-raised cosine filter is calculated, and its expression is as follows: ； in, This represents the limit value for the spectral bandwidth of a digital signal; The Nyquist bandwidth of a digital signal; This represents the roll-off factor of the root-raised cosine filter.
4. The method for suppressing low-frequency noise according to claim 3, characterized in that, Based on the frequency domain response of the root-raised cosine filter, adjust the roll-off factor of the root-raised cosine filter. The value of is used to ensure the required low-frequency noise filtering bandwidth.