Power delay spectrum estimation method based on frequency domain noise whitening

Abstract

The invention discloses a power delay spectrum estimation method based on frequency domain noise whitening, which is applied to a power line communication system. According to the method, in order to solve the problems that power line channel frequency selective noise is high and a special silence period is lacked, the noise power spectral density is directly estimated from a received signal through frequency domain smoothing processing, a frequency domain whitening filter is designed according to the noise power spectral density to conduct whitening and normalization processing on channel response, the frequency selective noise is converted into flat noise, and the frequency selective noise is obtained. And a power delay spectrum with a substantially reduced noise floor is generated. On the basis, multipath components are accurately recognized through peak detection and parameter estimation, signals are reconstructed to separate and calculate signal and noise power, and finally high-precision signal-to-noise ratio estimation is achieved. According to the method, a silent period is not needed, frequency selective noise can be effectively resisted, and the accuracy of multipath parameter and signal-to-noise ratio estimation is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of power line communication technology, and in particular to a power delay spectrum estimation method based on frequency domain noise whitening in power line communication systems. Background Technology

[0002] Power line communication (PLC) is a communication method that uses existing power line networks as the information transmission medium. Because power line networks are widely distributed, using power lines as the communication medium eliminates the need to drill holes and rewire indoors to rebuild the communication network. It offers advantages such as low cost and convenient connection, and its application in the smart Internet of Things (IoT) is becoming increasingly widespread.

[0003] Unlike wireless communication, the performance of power line communication (PLC) is primarily constrained by the PLC channel. The complex characteristics of PLC channels necessitate various signal processing techniques to combat the adverse effects of harsh PLC channels on PLC communication and improve communication performance. Accurate estimation of the channel's power delay spectrum (PDP) and signal-to-noise ratio (SNR) is crucial for system performance. However, the unique channel characteristics of PLC present numerous technical challenges for PDP estimation.

[0004] The noise environment of power line communication channels is complex, exhibiting significant non-Gaussian and non-stationary characteristics. Noise includes frequency-selective background noise, burst impulse noise, narrowband interference, and periodic noise. Power line networks have complex topologies and numerous impedance mismatch points, resulting in dense multipath propagation. Their channel propagation characteristics are characterized by strong frequency-selective fading and rapid time-varying properties. Furthermore, power line communication lacks a dedicated quiet period, making it difficult to obtain clean noise sampling timeframes in continuous power line communication systems.

[0005] Existing PDP estimation schemes all have their own shortcomings. For example, traditional methods based on quiet periods insert a dedicated quiet period between data transmission intervals to measure noise characteristics. This method consumes transmission resources during the quiet period, reducing system throughput and spectral efficiency. Furthermore, due to the rapid time-varying nature of power line noise, its timeliness is insufficient. For example, frequency domain methods based on empty subcarriers estimate noise PSD using empty subcarriers not used for data transmission. This method ignores frequency selectivity factors, and under strong frequency-selective fading, uneven distribution of empty subcarriers can easily lead to estimation errors. For example, time-domain impulse noise consistency methods detect and suppress impulse noise in the time domain before PDP estimation. This method has high computational complexity in filtering and detection in the time domain and only considers impulse noise, without taking into account the frequency selectivity of background noise. For example, channel estimation methods based on frequency domain smoothing use the smoothness characteristics of the frequency domain channel response to distinguish noise and signal. This method requires accurate signal-to-noise ratio and channel statistical characteristics. The power line channel model is complex, so this premise is not valid, and the matrix operation complexity is high.

[0006] Based on the above analysis, power line communication requires a PDP estimation technique to address the needs of eliminating the need for a silent period, resisting frequency-selective noise, achieving high resolution under dense multipath conditions, and being highly adaptable to the time-varying and non-stationary characteristics of power line channels.

[0007] It should be noted that the information disclosed in the background section above is only for understanding the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] The main objective of this invention is to overcome the deficiencies in the aforementioned background technology and propose a power delay spectrum estimation method based on frequency domain noise whitening in power line communication systems. Through frequency domain noise whitening processing, frequency-selective noise is transformed into flat noise, thereby obtaining a power delay spectrum with flat background noise and clear multipath components in the time domain. The core innovation lies in utilizing the channel's frequency domain smoothing characteristics to directly estimate noise characteristics from in-band data. By designing a whitening filter to compensate for the noise's frequency selectivity, frequency-selective noise is transformed into flat noise. Based on the noise basis of the whitened PDP, high-precision multipath parameter estimation is achieved. Signal components are reconstructed through multipath parameters, achieving accurate separation of signal and noise.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: A power delay spectrum estimation method based on frequency domain noise whitening includes the following steps: Step 1: Obtain the frequency domain channel response of the received signal; Step 2: Estimate the frequency domain noise power spectral density based on the frequency domain channel response; Step 3: Perform frequency domain smoothing on the noise power spectral density; Step 4: Design a frequency domain whitening filter based on the smoothed noise power spectral density; Step 5: Perform whitening filtering and energy normalization processing on the frequency domain channel response using the whitening filter; Step 6: Convert the whitened frequency domain channel response to the time domain to generate the whitened power delay spectrum; Step 7: Perform peak detection on the whitened power delay spectrum to identify multipath components and estimate their path parameters; Step 8: Reconstruct the signal components based on the identified path parameters, and calculate the signal power and noise power respectively in combination with the smoothed noise power spectral density to obtain the signal-to-noise ratio estimate.

[0010] A computer program product includes a computer program that, when executed by a processor, implements the power delay spectrum estimation method based on frequency domain noise whitening.

[0011] The present invention has the following beneficial effects: This invention provides a power delay spectrum estimation method based on frequency domain noise whitening in power line communication systems, enabling power delay spectrum estimation and signal-to-noise ratio (SNR) estimation in frequency-selective noise environments. This method, through steps including frequency domain noise power spectral density estimation, frequency domain whitening filter design, frequency domain signal whitening processing, PDP generation, path identification and parameter estimation, and SNR estimation, effectively combats frequency-selective noise without requiring a quiet period, generating high-quality power delay spectrum estimates and accurate SNR estimates. This invention is applicable to power line communication systems in complex noise environments, improving the accuracy of multipath parameter estimation and SNR estimation.

[0012] This invention demonstrates significant advantages in the field of power line communication, effectively solving core problems such as strong frequency-selective noise, lack of quiet periods, and dense multipath propagation in power line communication channels. Specifically: (1) No silent period is required, and noise characteristics can be accurately estimated in continuous transmission scenarios; (2) Anti-frequency selective noise, effectively compensates for the frequency selectivity of noise through whitening treatment; (3) Strong path resolution capability, multipath components are more clearly visible on whitened PDP; (4) Deep separation of signal and noise: Based on multipath parameters, the signal components are reconstructed to achieve accurate separation of signal and noise, avoiding the problem of mutual contamination between signal and noise in traditional methods, and achieving accurate estimation of signal-to-noise ratio; The main advantages of the embodiments of the present invention include: (1) Accurate noise estimation without quiescent period: Utilizing the frequency domain smoothing characteristics of the channel itself, noise PSD is estimated in real time and continuously during data transmission, thereby improving spectrum utilization. (2) Construct a frequency domain whitening processing framework: Design and apply a whitening filter to convert frequency-selective noise into flat white noise, laying the foundation for time domain analysis; (3) High-precision multipath detection: High-precision path parameter estimation is achieved on a flat noise substrate on a whitened PDP; (4) Accurate signal-to-noise ratio estimation: By reconstructing the signal components through multipath parameters, the signal and noise are accurately separated, thereby achieving accurate estimation of the signal power, noise power and signal-to-noise ratio.

[0013] Other beneficial effects of the embodiments of the present invention will be further described below. Attached Figure Description

[0014] Figure 1 This is a flowchart of a power delay spectrum estimation method based on frequency domain noise whitening for use in a power line communication system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the frequency domain channel response of a power delay spectrum estimation method based on frequency domain noise whitening in a power line communication system according to an embodiment of the present invention. Figure 3 This is a schematic diagram illustrating the noise PSD estimation principle of a power delay spectrum estimation method based on frequency domain noise whitening in a power line communication system according to an embodiment of the present invention. Figure 4 This is a whitening effect diagram of a power delay spectrum estimation method based on frequency domain noise whitening used in a power line communication system according to an embodiment of the present invention. Figure 5 This is a diagram showing the peak detection result of a whitened PDP using a power delay spectrum estimation method based on frequency domain noise whitening in a power line communication system according to an embodiment of the present invention.

[0015] Figure 6 This is a block diagram illustrating the technical roadmap of a power delay spectrum estimation method based on frequency domain noise whitening for use in power line communication systems, according to an embodiment of the present invention. Detailed Implementation

[0016] The embodiments of the present invention will be described in detail below. It should be emphasized that the following description is merely exemplary and not intended to limit the scope and application of the present invention.

[0017] This invention proposes a power delay spectrum estimation method based on frequency domain noise whitening in power line communication systems. By performing frequency domain noise whitening, frequency-selective noise is transformed into flat noise, thereby obtaining a power delay spectrum with flat background noise and clear multipath components in the time domain. The core innovation lies in utilizing the channel's frequency domain smoothing characteristics to directly estimate noise characteristics from in-band data. A whitening filter is designed to compensate for the noise's frequency selectivity, transforming it into flat noise. Based on the noise floor of the whitened power line communication device (PDP), high-precision multipath parameter estimation is achieved. Signal components are reconstructed using these multipath parameters, enabling accurate separation of signal and noise.

[0018] See Figure 1 This invention provides a power delay spectrum estimation method based on frequency domain noise whitening, comprising the following steps: Step 1: Obtain the frequency domain channel response of the received signal; Step 2: Calculate the estimated value of the frequency domain noise power spectral density (PSD) based on the frequency domain channel response of the received signal; Step 3: Perform frequency domain smoothing on the calculated noise PSD estimate to obtain the smoothed noise PSD; Step 4: Design a frequency domain whitening filter based on the smoothed PSD estimate; Step 5: Perform whitening filtering on the frequency domain channel response of the received signal according to the designed frequency domain whitening filter to obtain the filtered frequency domain channel response, and then normalize the filtered frequency domain channel response. Step 6: Transform the whitened filtered frequency domain channel response into the time domain using IFFT, and then process it to generate the whitened power delay spectrum (PDP). Step 7: Perform peak detection on the whitened PDP results to identify multipath components and estimate path parameters; Step 8: Reconstruct the signal components and calculate the signal power based on the identified path parameter information, calculate the noise power based on the calculated frequency domain noise PSD, and finally obtain the accurate signal-to-noise ratio.

[0019] The implementation of specific embodiments of the present invention will be described in further detail below.

[0020] See Figures 1 to 6 This invention provides a power delay spectrum estimation method based on frequency domain noise whitening in a power line communication system, comprising the following steps: Step 1: Obtain the frequency domain channel response of the received signal, specifically: Consider a power line communication system following the IEEE 1901 standard protocol. The received frequency domain channel response on the k-th subcarrier of the m-th OFDM symbol is: .

[0021] Step 2: Calculate the estimated value of the frequency domain noise power spectral density (PSD) based on the received signal's frequency domain channel response. Specifically: The smoothing process of the frequency domain channel response is represented as follows:

[0022] in This is represented as the smoothed frequency domain channel response. It is a length of The filtering window satisfies , The filter window value is used here; a low-pass filter is employed, and the window length is... Take 25. The value of should be chosen such that the equivalent noise bandwidth of the smoothing filter is greater than the coherence bandwidth of the noise, but less than the coherence bandwidth of the channel. , For the estimated channel delay spread, For subcarrier spacing, This indicates a rounding operation. This smoothing operation is based on the prior knowledge that the envelope change of the frequency domain channel response is slower than noise, and can be viewed as a linear time-invariant filtering process. It suppresses high-frequency fluctuations (noise) in the channel frequency response while preserving its low-frequency variation trend (signal principal components). The effects before and after smoothing are as follows: Figure 2 As shown, Figure 2 The comparison between the original frequency domain channel response and the smoothed frequency domain channel response is shown. The original channel response exhibits obvious frequency-selective fading characteristics. By smoothing the original frequency domain channel response, a relatively smooth signal component estimate is obtained.

[0023] The frequency domain estimate of the noise is expressed as:

[0024] in This represents the noise estimate of the m-th OFDM symbol on the k-th subcarrier.

[0025] The noise power spectral density (PSD) estimate is expressed as:

[0026] in This represents the PSD estimate of the m-th OFDM symbol on the k-th subcarrier.

[0027] Step 3: Perform frequency domain smoothing on the calculated noise PSD estimate to obtain the smoothed noise PSD estimate, specifically expressed as follows: =

[0028] in This represents the smoothed PSD estimate of the m-th OFDM symbol on the k-th subcarrier. Represents circular convolution. It is a length of The filtering window satisfies , The value represents the filter window; its size depends on the window type. Here, a moving average filter is used, and the window length is... The window length is set to 15. Smaller than the window length of the first smoothing used to estimate signal components , The range of values ​​for can be: to The odd number between these values. This smoothing process reduces the variance of the noise PSD estimate, resulting in a smooth and stable noise PSD estimate. The smoothed PSD effect is as follows: Figure 3 As shown, Figure 3The comparison between the original noise PSD and the smoothed noise PSD is shown. After smoothing, the noise PSD is smoother and more stable.

[0029] Step 4: Design a frequency domain whitening filter based on the smoothed PSD estimate, specifically as follows:

[0030] in This is the smoothed PSD estimate. A positive constant less than 1 is called the regularization parameter. The value of and the average noise power Related, can be set ,in for The average value here Values The filter is essentially a frequency-domain pre-whitening filter, and its frequency-domain response is inversely proportional to the square root of the noise PSD. The value of and the average noise power Related, For a constant much smaller than 1, experience shows exist to Good results can be achieved within the specified range. Preferably, the following can be selected. = As the default value, the whitening filter is designed based on the maximum likelihood estimation theory under colored noise background. It is a frequency domain pre-whitening filter whose frequency domain response is inversely proportional to the square root of the noise PSD, thus providing a smaller gain on subcarriers with high noise power and a larger gain on subcarriers with low noise power, making the equivalent noise spectrum tend to flatten (i.e., whitening) across the entire frequency band. Regularization parameter The introduction of this parameter is for the sake of numerical stability during engineering implementation, in order to avoid excessive filter gain when the PSD estimate approaches zero.

[0031] Step 5: Perform whitening filtering on the frequency domain channel response of the received signal according to the designed frequency domain whitening filter to obtain the filtered frequency domain channel response, specifically expressed as:

[0032] in The frequency domain channel response of the m-th OFDM symbol on the k-th subcarrier after whitening filtering.

[0033] The filtered frequency domain channel response is normalized, and the energy is normalized as follows: Calculate the signal power before and after whitening filtering:

[0034] in It is the original signal power before whitening filtering. This represents the signal power after whitening filtering, and N is the total number of subcarriers. Calculate the normalization factor:

[0035] in This is a normalization factor for positive constants. The normalized frequency domain channel response after filtering is expressed as:

[0036] in This is the normalized frequency domain channel response of the m-th OFDM symbol on the k-th subcarrier.

[0037] This normalization process avoids altering the total energy of the signal during whitening filtering, ensuring that the processing only changes the statistical properties of the noise.

[0038] Step 6: Transform the whitened filtered frequency domain channel response into the time domain using an IFFT, specifically as follows:

[0039] in (m) represents the time-domain impulse response value, and ifft is... Transformation; the process of generating a whitened PDP is specifically expressed as follows:

[0040] in This is the whitened PDP, and the whitened PDP is as follows: Figure 4 As shown, Figure 4 The comparison between whitened PDP and traditional PDP is shown. After whitening, the noise floor of the PDP is significantly reduced, while the relative intensity of multipath peaks is more prominent.

[0041] Step 7: Perform peak detection on the whitened PDP results to identify multipath components and estimate path parameters, specifically as follows: The detection threshold is calculated based on the noise statistical characteristics of whitened PDPs. ,in and These are the estimated values ​​of the noise mean and standard deviation, respectively. Here, the threshold factor related to the expected false alarm probability is... The value is 3; in the power line channel, The value should be between 3 and 4.

[0042] Local maxima exceeding a threshold T are searched in the whitened PDP and selected as candidate multipath peaks. A distance criterion is used to merge adjacent peaks, eliminating false detections caused by sidelobes. The number of paths satisfying the criteria is L, and the path index is i. .

[0043] The estimated path parameters are specifically expressed as follows: Time delay estimation: The peak position is subsampled and estimated using parabolic interpolation to obtain the time delay. The specific process is as follows: For the peak point of the m-th OFDM symbol and its adjacent points , , The fitted parabola is expressed as:

[0044] The coefficients a, b, and c are determined by three points and expressed as follows:

[0045] Peak precise location Represented as:

[0046] For precise time delay Represented as:

[0047] in Sampling interval; amplitude estimation Represented as:

[0048] in [] indicates that the m-th OFDM symbol after whitening filtering is in the i-th path Location PDP results, phase estimation Represented as:

[0049] in This indicates that the m-th OFDM symbol after whitening filtering is in the i-th path. Temporal impulse response of location; complex gain Represented as: .

[0050] Figure 5The peak detection results of the whitened PDP are shown. Path identification and parameter estimation based on the whitened PDP are demonstrated. By setting an appropriate detection threshold, multipath components can be effectively identified and their time delay and power parameters estimated, such as... Figure 5 The annotations in the middle display parameter information.

[0051] Step 8: Reconstruct the signal components based on the identified path parameter information. Specifically, based on the multipath parameters detected in Step 7, reconstruct the noise-free frequency domain channel response. The reconstructed frequency domain channel response is expressed as:

[0052] in The frequency domain channel response reconstructed by the m-th OFDM symbol on the k-th carrier. Let k be the frequency of the kth subcarrier, specifically represented as:

[0053] in Given the subcarrier spacing, the reconstructed signal is represented in the frequency domain as an exponential linear combination, with weights equal to the complex gain of each path; the calculated signal power is specifically expressed as:

[0054] Where PS is the signal power and M is the total number of OFDM symbols; the noise power calculation based on the calculated frequency domain noise PSD is specifically expressed as follows:

[0055] Where PN represents noise power; the obtained precise signal-to-noise ratio is specifically expressed as follows: .

[0056] By reconstructing signal components based on accurately estimated multipath parameters, deep separation of signal and noise is achieved, thus providing a basis for calculating signal power and noise power.

[0057] The above description, in conjunction with specific / preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various substitutions or modifications can be made to these described embodiments without departing from the inventive concept, and all such substitutions or modifications should be considered within the scope of protection of the present invention.

Claims

1. A power delay spectrum estimation method based on frequency domain noise whitening, characterized in that, Includes the following steps: Step 1: Obtain the frequency domain channel response of the received signal; Step 2: Estimate the frequency domain noise power spectral density based on the frequency domain channel response; Step 3: Perform frequency domain smoothing on the noise power spectral density; Step 4: Design a frequency domain whitening filter based on the smoothed noise power spectral density; Step 5: Perform whitening filtering and energy normalization processing on the frequency domain channel response using the whitening filter; Step 6: Convert the whitened frequency domain channel response to the time domain to generate the whitened power delay spectrum; Step 7: Perform peak detection on the whitened power delay spectrum to identify multipath components and estimate their path parameters; Step 8: Reconstruct the signal components based on the identified path parameters, and calculate the signal power and noise power respectively in combination with the smoothed noise power spectral density to obtain the signal-to-noise ratio estimate.

2. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 1, characterized in that, In step 1, the frequency domain channel response of the received signal is represented as H(m,k), where m represents the OFDM symbol index and k represents the subcarrier index. That is, H(m,k) is the frequency domain channel response of the m-th OFDM symbol on the k-th subcarrier.

3. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 1, characterized in that, Step 2, the step of calculating the estimated value of the frequency domain noise power spectral density (PSD) based on the frequency domain channel response of the received signal, includes: The frequency domain channel response H(m,k) is smoothed in the frequency domain to estimate the signal components. The smoothing process is expressed as follows: in This is represented as the smoothed frequency domain channel response. Represents circular convolution. It is a length of The filtering window satisfies , Here, is the filter window value; , For the estimated channel delay spread, For subcarrier spacing, Indicates the rounding operation; Based on the smoothed frequency domain channel response and the original frequency domain channel response, calculate the frequency domain estimate of the noise: in This represents the noise estimate of the m-th OFDM symbol on the k-th subcarrier; Based on the frequency domain estimation of the noise, the noise power spectral density (PSD) is estimated: in This represents the noise PSD estimate of the m-th OFDM symbol on the k-th subcarrier.

4. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 3, characterized in that, In step 3, the estimated noise PSD is smoothed in the frequency domain to obtain a smoothed noise PSD. The smoothing process is described as follows: = in This represents the smoothed noise PSD estimate. It is a length of The second filtering window satisfies , Here, represents the filter window value; where is the length of the second filter window. Less than the length of the first filter window .

5. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 1, characterized in that, In step 4, the frequency domain whitening filter is designed based on the smoothed PSD estimate. The whitening filter is expressed as: in, The smoothed noise PSD estimate. This is the regularization parameter.

6. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 1 or 5, characterized in that, Step 5 includes: According to the frequency domain whitening filter The received signal frequency domain channel response H(m,k) is whitened and filtered to obtain the filtered frequency domain channel response. : Calculate the signal power before and after whitening filtering and : Where N is the total number of subcarriers; Calculate the normalization factor : The normalization factor is used to normalize the filtered frequency domain channel response to obtain the normalized frequency domain channel response. : 。 7. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 6, characterized in that, Step 6 includes: The normalized frequency domain channel response Perform an IFFT transform to obtain the time-domain impulse response. : The whitened power delay spectrum is generated based on the time-domain impulse response. : 。 8. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 1, characterized in that, Step 7 includes: Based on the whitened power delay spectrum Based on the noise statistical characteristics, the detection threshold T is calculated: in, and These are the mean estimate and standard deviation estimate of the noise component in the whitened power delay spectrum, respectively. Threshold factor; Search for the power delay spectrum after whitening Local maxima exceeding the threshold T are selected as candidate multipath peaks. The candidate multipath peaks are verified and merged to obtain the number of multipaths L that meet the conditions and the index of each path i. ; For the identified i-th path, estimate its path parameters: Delay The peak position was obtained by subsampling estimation using parabolic interpolation: in, The sampling interval is... This refers to the precise sample point location of the peak obtained through parabolic interpolation; Amplitude estimation Represented as: Phase estimation Represented as: in, For the time-domain impulse response at position The value at; Complex gain It consists of the amplitude estimate and the phase estimate: 。 9. The power delay spectrum estimation method based on frequency domain noise whitening according to claim 8, characterized in that, Step 8 includes: Based on the path parameters estimated in step 7 and Reconstruct the frequency domain channel response : in, Let k be the frequency of the kth subcarrier. , Subcarrier spacing; Calculate the signal power PS based on the reconstructed frequency domain channel response: Where M is the total number of OFDM symbols; Based on the smoothed noise power spectral density Calculate the noise power PN: Based on the signal power PS and noise power PN, the signal-to-noise ratio (SNR) is calculated: 。 10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the power delay spectrum estimation method based on frequency domain noise whitening as described in any one of claims 1 to 9.

Images