Signal processing method for power line carrier communication based on OFDM (Orthogonal Frequency Division Multiplexing) modulation

Abstract

The invention discloses a signal processing method for power line carrier communication based on OFDM (Orthogonal Frequency Division Multiplexing) modulation, which relates to the technical field of power line carrier communication, and comprises the steps of S1, designing and inserting a cyclic prefix and a guard interval, S2, improving a synchronization algorithm to realize frequency offset processing, S3, performing noise suppression optimization on a front-end signal processing module, and S4, designing a parallel processing architecture and allocating tasks. S5, receiving end channel estimation, S6, frequency domain equalization processing, S7, OFDM demodulation processing, and S8, symbol synchronization and data recovery. According to the invention, through the steps S1 and S2, multipath time delay of a power line channel is effectively resisted, and inter-symbol interference and inter-carrier interference are reduced; through the steps S3, S5 and S6, the effects of accurately identifying the pulse noise and Gaussian noise characteristics of the power line and dynamically suppressing the noise are achieved; through the steps S4 and S8, the parallel processing architecture is utilized to improve the signal processing efficiency and avoid the overload effect of the processor, and meanwhile, the data error correction capability is enhanced through convolutional coding or Turbo coding, so that the real-time performance and reliability of high-speed data transmission are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power line carrier communication, in particular to a signal processing method for power line carrier communication based on OFDM modulation. BACKGROUND

[0002] With the rapid development of smart grid and Internet of Things technology, power line carrier communication, as a technology for realizing data transmission by using existing power line infrastructure, has been widely applied in smart home control, power monitoring, remote meter reading and other fields due to its advantages of no need for additional laying of communication lines, low cost and convenient deployment; power line carrier communication based on orthogonal frequency division multiplexing technology can effectively resist multipath fading and improve spectrum utilization by dividing high-speed data streams into multiple low-speed sub-data streams and modulating them onto multiple orthogonal subcarriers for transmission, and has become the mainstream technical solution in the current power line carrier communication field.

[0003] However, the power line channel environment is complex and changeable, and there are problems such as serious noise interference, multipath effect and frequency selective fading, which bring many challenges to power line carrier communication based on OFDM modulation; on the one hand, the interference signals such as impulse noise and Gaussian noise existing in the power line will seriously affect the signal transmission quality, and the traditional filtering method is difficult to achieve accurate suppression of complex noise; on the other hand, the multipath time delay characteristics of the power line channel are easy to cause inter-symbol interference and inter-carrier interference, and the conventional cyclic prefix design and synchronization algorithm are difficult to meet the high-precision synchronization demand; in addition, the power line carrier communication signal processing process involves a large number of complex operations, and the traditional serial processing architecture has bottlenecks in real-time performance and processing efficiency, which is difficult to adapt to high-speed data transmission scenarios; it is necessary to design a signal processing method for power line carrier communication based on OFDM modulation to solve the above problems. SUMMARY

[0004] To achieve the above purpose, the present application is implemented by the following technical scheme: a signal processing method for power line carrier communication based on OFDM modulation, comprising the following implementation steps: S1, cyclic prefix and guard interval design and insertion: for the OFDM modulation process in power line carrier communication, based on the multipath time delay characteristics of the power line channel, the maximum time delay spread value Tmax of the power line channel is obtained by channel measurement, and a specific cyclic prefix length Lcp and guard interval Lgi are designed; S2, improved synchronization algorithm to realize frequency offset processing: an improved synchronization algorithm is adopted, and the frequency offset of the received signal is estimated at the receiving end by using the cyclic correlation characteristics based on the pilot symbol; The pilot symbol sequence in the received signal and the locally known pilot symbol sequence are subjected to cyclic correlation operation, the frequency offset ∆f is calculated by finding the correlation peak position, and frequency offset compensation is performed according to the calculation result where Ts is a sampling period, x(n) is an original signal, and y(n) is a compensated signal; S3, noise suppression optimization of the front-end signal processing module: noise samples of the power line under different time periods and different load conditions are collected to construct a data set containing 1000 groups of noise data; The time-frequency distribution characteristics of the power line noise are extracted by using the short-time Fourier transform, the time-frequency distribution characteristic image is input into the deep learning feature extraction unit, the cross-entropy loss function and the stochastic gradient descent algorithm are used for training, the pulse interference characteristics and the Gaussian noise characteristics of the power line noise are identified, and the classification weight is output; S4, parallel processing architecture design and task allocation: a parallel processing architecture is adopted, the signal processing tasks are allocated to multiple processor units, the parallel processing module includes at least three independent processor units, which respectively perform the functions of cyclic prefix insertion, frequency offset compensation and noise suppression; And the task priority of the processor unit is allocated through a dynamic load balancing strategy, the task queue length preset threshold is set to 50, and when the task queue length of any processor unit exceeds the preset threshold, the task migration mechanism is triggered, part of the tasks are migrated to the processor unit with lighter load, and the efficiency and timeliness of signal processing are ensured.

[0005] Preferably, the following implementation steps are further included: S5, channel estimation at the receiving end: the signal after frequency offset compensation and noise suppression is subjected to channel estimation at the receiving end, the least square method is used to estimate the channel frequency response H(k) by using the pilot symbol, and the linear interpolation method is used to interpolate the channel response at the data subcarriers to obtain the complete channel frequency response estimation value S6, frequency domain equalization processing: according to the channel frequency response estimation value obtained in step S5, the received signal is equalized by using the zero-forcing equalization algorithm or the minimum mean square error equalization algorithm; Equalizer coefficients where H*(k) is the conjugate of the channel frequency response, σ² is the noise variance, and N0 is the noise power spectral density; S7, OFDM demodulation processing: the equalized signal is subjected to OFDM demodulation, the cyclic prefix and the guard interval are removed, the signal is converted from the frequency domain to the time domain, and the original modulation symbol is recovered through the inverse fast Fourier transform; S8, symbol synchronization and data recovery: the demodulated symbol is subjected to symbol synchronization and timing recovery, the timing synchronization Gardner algorithm is used to accurately detect the symbol boundary, and the original data sequence is recovered.

[0006] Preferably, the S1 step is pre-encoding the orthogonal frequency division multiplexing symbol before inserting the cyclic prefix and guard interval, using Gray coding to encode the symbol, and reducing the error probability between symbols. The cyclic prefix length Lcp is set to be greater than or equal to , fs is the sampling frequency, and Lgi is set to 0.5*Lcp. The cyclic prefix and guard interval are inserted at the starting position of the orthogonal frequency division multiplexing symbol. In the S2 step, when performing frequency offset estimation, the received signal is divided into multiple subsequences with a length of L using a segmented correlation method, and a cyclic correlation operation is performed on each subsequence. Then, the average of the frequency offset estimation values of the subsequences is taken as the final frequency offset ∆f, thereby improving the accuracy of frequency offset estimation.

[0007] Preferably, in the S3 step, the deep learning feature extraction unit uses data enhancement techniques to rotate, translate, and scale the noise time-frequency distribution feature image during the training process, expands the training data set, and improves the generalization ability of the model. The parameter adjustment unit dynamically adjusts the filter coefficients according to the time-frequency distribution characteristics to achieve accurate suppression of noise. The convolutional neural network layer uses GPU for accelerated operation, and the parameter adjustment unit is composed of a control chip and a digital-to-analog converter.

[0008] Preferably, in the S4 step, the dynamic load balancing strategy calculates the load value of each processor unit based on multi-dimensional indicators such as CPU usage, memory occupancy, and task queue length, and allocates task priorities in order of load value from small to large.

[0009] Preferably, in the S5 step, when performing channel estimation, a channel estimation method based on compressed sensing is introduced. In the case of a small number of pilots, the sparse characteristics of the power line channel are used to reconstruct the channel frequency response, thereby improving the efficiency and accuracy of channel estimation.

[0010] Preferably, in the S6 step, after frequency domain equalization processing, the equalized signal is subjected to bit error rate detection. When the bit error rate exceeds the preset threshold value 0.001, the equalization algorithm is automatically switched from ZF equalization algorithm to MMSE equalization algorithm, or a fusion strategy of the two algorithms is used for secondary equalization.

[0011] Preferably, in the S8 step, after symbol synchronization and timing recovery, the recovered original data sequence is subjected to error correction encoding processing. Convolutional encoding or Turbo encoding error correction encoding is used to further improve the reliability of data transmission.

[0012] In summary, the application provides a signal processing method for power carrier communication based on OFDM modulation, which has the following beneficial effects: The frequency offset processing step is realized by S1 cyclic prefix and guard interval design and insertion and S2 improved synchronization algorithm, which effectively resists power line channel multipath delay, reduces inter-symbol interference and inter-carrier interference, accurately compensates frequency offset by using pilot symbol cyclic correlation characteristics, restores OFDM subcarrier orthogonality, and improves signal demodulation accuracy; meanwhile, S1 cyclic prefix length Lcp is set to ≥ , and Lgi is set to 0.5*Lcp, which isolates OFDM symbol boundaries from the time domain and effectively eliminates symbol overlap caused by multipath reflection; S2 reduces the frequency offset estimation error to ≤0.001*fs by pilot symbol segmentation and cyclic correlation operation, and , so that the inter-carrier interference suppression ratio is improved to more than 25dB, and the QAM demodulation error rate is stably below 0.0001 under 16QAM modulation.

[0013] Through the noise suppression optimization of S3 front-end signal processing module, S5 receiving end channel estimation and S6 frequency domain equalization processing step, the characteristics of power line impulse noise and Gaussian noise are accurately identified and dynamically suppressed, and the signal-to-noise ratio and integrity of signal transmission are improved by combining least square channel estimation and frequency domain equalization algorithm to eliminate channel fading and inter-symbol interference; the 3-layer convolutional neural network trained based on 1000 noise samples in S3 improves the accuracy of identification of impulse noise and classification of Gaussian noise, and the dynamically adjusted filter coefficient reduces the in-band noise power by 15-20dB; S5 introduces compressed sensing channel estimation, and the channel frequency response reconstruction error is ≤0.05 under the condition of reducing the number of pilots; S6 step improves the signal integrity and overall signal-to-noise ratio of the system by MMSE equalization algorithm when the channel fading depth is ≥10dB.

[0014] Through S4 parallel processing architecture design and task allocation and S8 symbol synchronization and data recovery step, the signal processing efficiency is improved by using parallel processing architecture to avoid processor overload, and the data error correction capability is enhanced by convolutional coding or Turbo coding to ensure the real-time performance and reliability of high-speed data transmission. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The flow architecture schematic diagram of the signal processing method for power carrier communication based on OFDM modulation of the application. DETAILED DESCRIPTION

[0016] The following will be described in detail in combination with the Figure 1 The application will be further described in detail.

[0017] EMBODIMENT

[0018] Referring to Figure 1 The application provides a technical solution: a signal processing method for power line carrier communication based on OFDM modulation, comprising the following implementation steps: S1, cyclic prefix and guard interval design and insertion: for the OFDM modulation process in power line carrier communication, based on the multipath time delay characteristics of the power line channel, the maximum time delay spread value Tmax of the power line channel is obtained through channel measurement, a specific cyclic prefix length Lcp and guard interval Lgi are designed, and by setting the cyclic prefix length and guard interval, the multipath time delay of the power line channel is effectively resisted, the inter-symbol interference and inter-carrier interference are reduced, and the anti-multipath capability of the OFDM system is improved; S2, improved synchronization algorithm to realize frequency offset processing: an improved synchronization algorithm is adopted, the frequency offset of the received signal is estimated at the receiving end by using the cyclic correlation characteristics based on the pilot symbol, the accuracy of the frequency offset estimation is improved through the cyclic correlation operation of the pilot symbol, and the damage of the carrier frequency offset to the orthogonality of the OFDM subcarriers is reduced; The pilot symbol sequence in the received signal is cyclically correlated with the locally known pilot symbol sequence, the frequency offset ∆f is calculated by finding the correlation peak position, and frequency offset compensation is performed according to the calculation result Where Ts is the sampling period, x(n) is the original signal, and y(n) is the compensated signal. By accurately compensating the frequency offset through y(n), the subcarrier orthogonality is restored, ICI is reduced, and the accuracy of signal demodulation is improved; S3, noise suppression optimization of the front-end signal processing module: noise samples of the power line under different time periods and different load conditions are collected, a data set containing 1000 groups of noise data is constructed, and through multi-scene noise sample collection, the complex noise environment of the power line is covered to provide diversified training data for the deep learning model; The time-frequency distribution characteristics of the power line noise are extracted by using short-time Fourier transform, the time-frequency distribution characteristic image is input into the deep learning feature extraction unit, the cross-entropy loss function and the stochastic gradient descent algorithm are used for training, the pulse interference characteristics and Gaussian noise characteristics of the power line noise are identified and the classification weight is output, the noise time-frequency characteristics are automatically extracted through the deep learning network, the pulse noise and Gaussian noise are accurately classified, and the decision basis for dynamic filtering is provided; S4, parallel processing architecture design and task allocation: a parallel processing architecture is adopted, the signal processing tasks are allocated to multiple processor units, the parallel processing module includes at least three independent processor units, which respectively perform the functions of cyclic prefix insertion, frequency offset compensation and noise suppression, through task splitting to independent processor units, the parallelization of the signal processing flow is realized, and the overall processing efficiency is improved; And through the dynamic load balancing strategy distribution processor unit task priority, set the task queue length preset threshold is 50, when any processor unit task queue length exceeds the preset threshold, trigger task migration mechanism, part of the task migration to the load lighter processor unit, ensure the efficiency and timeliness of signal processing, through dynamic load balancing and task migration, avoid single processor unit overload, ensure the real-time performance and system throughput of signal processing.

[0019] Also includes the following implementation steps: S5, receiving end channel estimation: in the receiving end after the frequency offset compensation and noise suppression signal channel estimation, using pilot symbols least squares method to estimate the channel frequency response H(k), and through linear interpolation method on the data subcarrier channel response interpolation, get the complete channel frequency response estimate, through the least squares method and linear interpolation, get the complete channel frequency response estimate, for subsequent equalization processing provides accurate channel state information; S6, frequency domain equalization processing: according to the channel frequency response estimate value obtained in step S5, using zero forcing equalization algorithm or minimum mean square error equalization algorithm for equalization of received signal, through the frequency domain equalization algorithm to eliminate channel fading and intersymbol interference, restore the original amplitude and phase of signal; Equalizer coefficients , wherein H*(k) is the conjugate of channel frequency response, σ² is the noise variance, N0 is the noise power spectral density, through W(k) calculation equalizer coefficients, combined with H*(k) and σ² parameters, realize the accurate compensation of channel distortion; S7, OFDM demodulation processing: OFDM demodulation of the equalized signal, remove the cyclic prefix and guard interval, convert the signal from the frequency domain to the time domain, through the fast recovery of the original modulation symbol, through the OFDM demodulation process, the frequency domain signal is converted into time domain symbol, complete the inverse transform recovery of modulation symbol; S8, symbol synchronization and data recovery: symbol synchronization and timing recovery of demodulated symbol, using timing synchronization Gardner algorithm, accurate detection of symbol boundary, restore the original data sequence, through Gardner algorithm realizes the symbol level accurate synchronization, ensure the boundary of data sequence is correct, avoid the error caused by symbol misplacement.

[0020] S1 in step, before inserting the cyclic prefix and guard interval, pre coding processing of orthogonal frequency division multiplexing symbol, using Gray coding method for coding, reduce the error probability between symbols, through the Gray coding reduces the error probability between adjacent symbols, even if a single bit error also only affects the adjacent symbol, improve the coding fault tolerance; Cyclic prefix length Lcp is set to ≥ , fs is the sampling frequency, Lgi is set to 0.5*Lcp, the cyclic prefix and guard interval are inserted at the starting position of the orthogonal frequency division multiplexing symbol, the specific values of Lcp and Lgi are determined by quantization calculation, the cyclic prefix length covers the maximum multipath delay, the guard interval further isolates the symbol boundary, and the reliability of system transmission is improved; In the S2 step, when performing frequency offset estimation, the received signal is divided into multiple subsequences with a length of L by using the segmented correlation method, and the cyclic correlation operation is performed on each subsequence respectively, and then the average value of the frequency offset estimation values of each subsequence is taken as the final frequency offset ∆f, so as to improve the accuracy of frequency offset estimation, reduce the influence of random noise on frequency offset estimation by taking the average value through segmented correlation, and improve the stability and accuracy of the estimation result.

[0021] In the S3 step, the deep learning feature extraction unit adopts the data enhancement technology to perform rotation, translation and scaling operations on the noise time-frequency distribution feature image during the training process, expands the training data set, and improves the generalization ability of the model, so as to expand the training data set through data enhancement, improve the generalization ability of the deep learning model to different noise forms, and enhance the reliability of the model. The parameter adjustment unit dynamically adjusts the filter coefficients according to the time-frequency distribution characteristics, realizes accurate suppression of noise, the convolutional neural network layer adopts GPU for accelerated operation, the parameter adjustment unit is composed of a control chip and a digital-to-analog converter, and through dynamic adjustment of the filter coefficients, different types of noise are adaptively suppressed, the operation efficiency is improved by combining GPU acceleration, and real-time noise reduction processing is ensured.

[0022] In the S4 step, the dynamic load balancing strategy calculates the load value of the processor unit according to the CPU usage, memory occupancy and task queue length multi-dimensional indexes of each processor unit, and allocates the task priority in the order from small to large according to the load value, so as to dynamically evaluate the processor load through multi-dimensional indexes, avoid uneven task allocation caused by single index deviation, and improve the scientificity of load balancing.

[0023] In the S5 step, when performing channel estimation, the channel estimation method based on compressed sensing is introduced, the sparse characteristics of the power line channel are utilized to reconstruct the channel frequency response under the condition that the number of pilots is small, the efficiency and accuracy of channel estimation are improved, the compressed sensing technology is used to reduce the demand for the number of pilots, and the estimation efficiency is improved, and the estimation accuracy is improved by utilizing the channel sparsity.

[0024] In the step S6, after the equalization processing in the frequency domain, the equalized signal is subjected to the bit error rate detection, and when the bit error rate exceeds the preset threshold 0.001, the equalization algorithm is automatically switched from the ZF equalization algorithm to the MMSE equalization algorithm, or the fusion strategy of the two algorithms is adopted for the secondary equalization, the algorithm adaptive switching is triggered through the real-time bit error rate detection, the equalization performance is dynamically improved when the channel condition deteriorates, and the system transmission quality is ensured.

[0025] In the step S8, after the symbol synchronization and timing recovery, the original data sequence recovered is subjected to the error correction coding processing, the convolution coding or Turbo coding error correction coding mode is adopted, the reliability of the data transmission is further improved, the redundant check bits are introduced through the error correction coding, the detection and correction of the error bits in the transmission process are realized, and the reliability of the data transmission is further improved.

[0026] The embodiments of the specific implementation are the preferred embodiments of the present application, and are not limited to the protection scope of the present application, wherein the same parts are indicated by the same reference numerals. Therefore, any equivalent changes made according to the structure, shape, principle of the present application should be covered in the protection scope of the present application.

Claims

1. A signal processing method for power line carrier communication based on OFDM modulation, characterized in that: The implementation steps include the following: S1. Cyclic Prefix and Guard Interval Design and Insertion: For the OFDM modulation process in power line carrier communication, based on the multipath delay characteristics of the power line channel, the maximum delay spread value Tmax of the power line channel is obtained through channel measurement, and a specific cyclic prefix length Lcp and guard interval Lgi are designed. S2. Improved synchronization algorithm for frequency offset processing: An improved synchronization algorithm is adopted, which utilizes the cyclic correlation characteristics based on pilot symbols to estimate the frequency offset of the received signal at the receiving end. The pilot symbol sequence in the received signal is cyclically correlated with the locally known pilot symbol sequence. The frequency offset Δf is calculated by finding the position of the correlation peak, and frequency offset compensation is performed based on the calculation result. Where Ts is the sampling period, x(n) is the original signal, and y(n) is the compensated signal; S3. Noise suppression optimization of the front-end signal processing module: By collecting noise samples from the power line at different times and under different load conditions, a dataset containing 1000 sets of noise data is constructed. The time-frequency distribution features of power line noise are extracted using short-time Fourier transform. The time-frequency distribution feature image is input into a deep learning feature extraction unit, and training is performed using cross-entropy loss function and stochastic gradient descent algorithm. The impulse interference features and Gaussian noise features of power line noise are identified and classification weights are output. S4. Parallel processing architecture design and task allocation: A parallel processing architecture is adopted to allocate signal processing tasks to multiple processor units. The parallel processing module contains at least three independent processor units, which respectively perform cyclic prefix insertion, frequency offset compensation and noise suppression functions. Furthermore, a dynamic load balancing strategy is used to allocate task priorities to processor units. A preset threshold for task queue length is set to 50. When the task queue length of any processor unit exceeds the preset threshold, a task migration mechanism is triggered to migrate some tasks to processor units with lighter loads, ensuring the efficiency and timeliness of signal processing.

2. The signal processing method for power line carrier communication based on OFDM modulation according to claim 1, characterized in that: It also includes the following implementation steps: S5. Receiver Channel Estimation: At the receiver, channel estimation is performed on the signal after frequency offset compensation and noise suppression. The channel frequency response H(k) is estimated using the least squares method with pilot symbols, and the channel response at the data subcarrier is interpolated using the linear interpolation method to obtain the complete channel frequency response estimate. S6. Frequency domain equalization processing: Based on the channel frequency response estimate obtained in step S5, the received signal is equalized using the zero-forcing equalization algorithm or the minimum mean square error equalization algorithm. Equalizer coefficients σ² is the conjugate of the channel frequency response, σ² is the noise variance, and N0 is the noise power spectral density. S7, OFDM demodulation: The equalized signal is demodulated using OFDM to remove the cyclic prefix and guard interval, converting the signal from the frequency domain to the time domain, and recovering the original modulation symbol through inverse fast Fourier transform. S8. Symbol Synchronization and Data Recovery: The demodulated symbols are synchronized and recovered at a specific time. The Gardner algorithm for timing synchronization is used to accurately detect symbol boundaries and recover the original transmitted data sequence.

3. The signal processing method for power line carrier communication based on OFDM modulation according to claim 1, characterized in that: In step S1, before inserting the cyclic prefix and guard interval, the orthogonal frequency division multiplexing symbols are pre-coded using Gray coding to reduce the probability of bit errors between symbols. The cyclic prefix length Lcp is set to ≥ fs is the sampling frequency, Lgi is set to 0.5*Lcp, and the cyclic prefix and guard interval are inserted at the beginning of the orthogonal frequency division multiplexing symbol; In step S2, when estimating the frequency offset, a segmented correlation method is used to divide the received signal into multiple subsequences of length L. Cyclic correlation operations are performed on each subsequence, and then the average value of the frequency offset estimates of each subsequence is taken as the final frequency offset ∆f, thereby improving the accuracy of frequency offset estimation.

4. The signal processing method for power line carrier communication based on OFDM modulation according to claim 1, characterized in that: In step S3, the deep learning feature extraction unit uses data augmentation techniques during training to rotate, translate, and scale the noise time-frequency distribution feature image to expand the training dataset and improve the model's generalization ability. The parameter adjustment unit dynamically adjusts the filter coefficients according to the time-frequency distribution characteristics to achieve precise noise suppression; the convolutional neural network layer uses a GPU for accelerated computation, and the parameter adjustment unit consists of a control chip and a digital-to-analog converter.

5. The signal processing method for power line carrier communication based on OFDM modulation according to claim 1, characterized in that: In step S4, the dynamic load balancing strategy calculates the load value of each processor unit based on multiple dimensions of indicators such as CPU utilization, memory usage, and task queue length, and allocates task priorities in ascending order of load value.

6. The signal processing method for power line carrier communication based on OFDM modulation according to claim 2, characterized in that: In step S5, a channel estimation method based on compressed sensing is introduced when performing channel estimation. When the number of pilots is small, the channel frequency response is reconstructed by utilizing the sparsity characteristics of the power line channel, thereby improving the efficiency and accuracy of channel estimation.

7. The signal processing method for power line carrier communication based on OFDM modulation according to claim 2, characterized in that: In step S6, after frequency domain equalization, the bit error rate of the equalized signal is detected. When the bit error rate exceeds the preset threshold of 0.001, the equalization algorithm is automatically switched from ZF equalization algorithm to MMSE equalization algorithm, or a fusion strategy of the two algorithms is used for secondary equalization.

8. The signal processing method for power line carrier communication based on OFDM modulation according to claim 2, characterized in that: In step S8, after symbol synchronization and timed recovery, the recovered original data sequence is subjected to error correction coding processing. Convolutional coding or Turbo coding error correction coding methods are used to further improve the reliability of data transmission.

Images