A DFT channel estimation method for OFDM-PLC system

Abstract

This invention discloses a DFT channel estimation method for OFDM-PLC systems, comprising: obtaining channel estimates for pilot subcarriers using a least-squares channel estimation algorithm; converting the estimated channel response to the time domain using IDFT to obtain the channel impulse response values ​​of the sampling points; dividing the sampling points into sampling points within a cyclic prefix and noise points according to the length of the CP; setting the channel impulse response values ​​of the noise points to zero; setting the channel impulse response values ​​of the sampling points within the cyclic prefix whose signal-to-noise ratio is less than or equal to the average signal-to-noise ratio to zero, thus obtaining a first-order denoising result; setting the channel impulse response values ​​of the sampling points within the cyclic prefix whose energy is less than or equal to the average energy of the noise in the sampling points within the cyclic prefix to zero, thus obtaining a second-order denoising result; and obtaining an estimate of the channel frequency response through discrete Fourier transform. This invention's method, by performing second-order denoising on the channel impulse response, makes the estimated channel frequency response closer to the true value.

Description

Technical Field

[0001] This invention belongs to the field of channel technology, specifically, it relates to a DFT channel estimation method for OFDM-PLC systems. Background Technology

[0002] The application of power line carrier communication (PLC) in smart grids has significantly reduced the construction and maintenance costs of telecommunications functions within the power grid. Based on signal frequency bandwidth and data transmission rate, PLCs can be classified into narrowband PLCs (NB PLCs) and wideband PLCs (BB PLCs). PLC channels are not only affected by the random access and disconnection of electrical equipment in the system, but also face strong noise, frequency-selective fading, and multipath effects. In recent years, Orthogonal Frequency Division Multiplexing (OFDM) technology has been increasingly applied to PLCs due to its high spectral efficiency, resistance to multipath transmission, and resistance to frequency-selective fading. The purpose of channel estimation is to estimate the time-domain or frequency-domain response of the channel, correct and recover the received data, and obtain the performance gain of coherent detection. However, traditional OFDM channel estimation methods are not well-suited for PLC channels due to the presence of both background noise and impulse noise. Least Squares (LS) channel estimation is the most commonly used algorithm in OFDM systems, but it does not consider the impact of noise, resulting in a large mean square error when the signal-to-noise ratio is low. Although Minimum Mean Square Error (MMSE) is the optimal criterion in pilot-based channel estimation algorithms, its drawback is the need for prior information about the channel, which is difficult to obtain in burst noise communication systems. Furthermore, the algorithm's complexity is high due to the large number of matrix inversions involved. Discrete Fourier Transform (DFT)-based channel estimation techniques improve the performance of LS channel estimation by eliminating noise effects beyond the maximum channel delay. However, traditional DFT algorithms only remove noise outside the cyclic prefix (CP) and cannot filter out noise within the CP.

[0003] To suppress the influence of noise within the CP (Portable Component Analysis) system, some researchers have proposed threshold-based denoising methods. These methods filter noisy sampling points by setting an energy threshold within the CP. The threshold is typically chosen based on the energy values ​​and amplitude moduli of sampling points both inside and outside the CP, and is set as either the maximum or median value. The problem with these methods is that the threshold is sometimes too large to retain valid data, or sometimes too small to filter noise effectively. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a DFT channel estimation method for OFDM-PLC systems. By performing secondary noise reduction on the channel impulse response, the estimated channel frequency response is closer to the true value.

[0005] To achieve the above technical objectives, the present invention adopts the following technical solution: a DFT channel estimation method for OFDM-PLC systems, specifically including the following steps:

[0006] Step 1: Use the least squares channel estimation algorithm to obtain the channel estimate of the pilot subcarrier, and transform the channel estimate of the pilot subcarrier to the time domain through the inverse discrete Fourier transform to obtain the channel impulse response value of the sampling point;

[0007] Step 2: Divide the sampling points into sampling points within the cyclic prefix and noise points according to the length of the cyclic prefix. Calculate the average noise power of the noise points using the channel impulse response values ​​of the noise points, and set the channel impulse response values ​​of the noise points to zero.

[0008] Step 3: Calculate the average power of the sampling points within the cyclic prefix, and combine it with the average noise power of the noise points to calculate the average signal-to-noise ratio of the sampling points within the cyclic prefix and the alternative signal-to-noise ratio of the sampling points within the cyclic prefix;

[0009] Step 4: Set the channel impulse response value of the sampling points in the cyclic prefix whose signal-to-noise ratio is less than or equal to the average signal-to-noise ratio to zero to obtain the first noise reduction result;

[0010] Step 5: Calculate the total energy of the sampling points within the cyclic prefix, and combine it with the average signal-to-noise ratio of the sampling points within the cyclic prefix to calculate the average energy of the noise in the sampling points within the cyclic prefix.

[0011] Step 6: If the energy of a certain sampling point within the cyclic prefix is ​​less than or equal to the average energy of the noise in the sampling points within the cyclic prefix, set the channel impulse response value of the sampling points excluding the cyclic prefix to zero to obtain the secondary noise reduction result. Then, obtain the estimate of the channel frequency response through discrete Fourier transform.

[0012] Furthermore, the channel estimation H of the pilot subcarrier in step 1 LS for:

[0013] H LS (k)=Y(k) / X(k)=H(k)+W(k) / X(k),0≤k≤N-1

[0014] Where k represents the subcarrier number, N represents the total number of subcarriers, Y(k) represents the received signal in the frequency domain, X(k) represents the transmitted signal in the frequency domain, H(k) represents the frequency response of the multipath channel, and W(k) represents the noise in the frequency domain.

[0015] Furthermore, the channel impulse response value of the sampling point in step 1 is:

[0016]

[0017] Among them, h LS (n) represents the channel impulse response value at the nth sampling point, IDFT[] represents the inverse of the discrete Fourier transform, j represents the imaginary unit, h(n) represents the impulse response of the multipath channel, w(n) represents the noise complex sample in the PLC system, and w(n) = IDFT[W(k) / X(k)].

[0018] Furthermore, the process of dividing the sampling points in step 2 is as follows:

[0019]

[0020] Among them, L CP This indicates the length of the cyclic prefix (CP).

[0021] Furthermore, the average noise power P at the noise point in step 2 w for:

[0022]

[0023] Furthermore, the average power P of the sampling points within the cyclic prefix described in step 3 s+w for:

[0024]

[0025] The average signal-to-noise ratio γ of the sampling points within the cyclic prefix av for:

[0026]

[0027] The alternative signal-to-noise ratio γ of the sampling points within the cyclic prefix n for:

[0028]

[0029] Furthermore, the noise reduction result h in step 4 DFT-1 (n) is:

[0030]

[0031] Furthermore, the total energy of the sampling points within the cyclic prefix mentioned in step 5 is:

[0032]

[0033] The total energy of the noise inside the cyclic prefix is:

[0034]

[0035] The average energy of the noise in the sampling points within the cyclic prefix is:

[0036]

[0037] Furthermore, the secondary noise reduction result h in step 6 DFT-2 (n) is:

[0038]

[0039] Furthermore, the estimation of the channel frequency response in step 6 is as follows:

[0040]

[0041] Compared with the prior art, the present invention has the following advantages: The DFT channel estimation method of the present invention for OFDM-PLC system is designed to address the widespread background noise and burst impulse noise in the PLC channel by using the alternative signal-to-noise ratio of each sampling point in the CP and the average energy of the useful signal at the sampling point in the CP. The method filters the impulse noise and background noise respectively. Compared with similar technologies, the present invention can obtain more accurate channel estimation with lower mean square error and lower computational complexity. Attached Figure Description

[0042] Figure 1 Block diagram of the transmitter and receiver of an OFDM-PLC system;

[0043] Figure 2 This is a flowchart of the DFT channel estimation method for OFDM-PLC systems according to the present invention;

[0044] Figure 3 A comparison chart of channel power estimated by different channel estimation methods;

[0045] Figure 4 This is a comparison chart of the mean square error of different channel estimation methods. Detailed Implementation

[0046] The technical solution of the present invention will be further explained and described below with reference to the accompanying drawings.

[0047] like Figure 1The diagram shows the transmitter and receiver block diagram of an OFDM-PLC system. The input bitstream to this OFDM-PLC system undergoes the following operations at the transmitter: serial-to-parallel conversion, modulation, pilot insertion, inverse fast Fourier transform (IFFT), addition of a cyclic prefix (CP), parallel-to-serial conversion, and digital-to-analog conversion. The signal undergoes the opposite operations at the receiver. The frequency domain data X(k) of N subcarriers is converted to time domain data x(n) using IFFT as follows:

[0048]

[0049] After adding the cyclic prefix (CP), the length of the time-domain data is usually greater than the maximum channel delay. The time-domain data is sent to the PLC channel after parallel-to-serial conversion and digital-to-analog conversion. The impulse response of the PLC channel can be expressed as:

[0050]

[0051] Where L is the number of paths in the multipath channel, a i τ is the magnitude of the i-th path. i It is the delay of the i-th path;

[0052] The time-domain signal y(n) obtained through the multipath fading channel and after removing the CP can be expressed as:

[0053] y(n)=x(n)*h(n)+w(n),0≤n≤N-1

[0054] Where w(n) represents the complex number of noise samples in the PLC system, including background noise and impulse noise;

[0055] The frequency domain signal obtained by FFT of y(n) is:

[0056] Y(k)=X(k)H(k)+W(k),0≤k≤N-1

[0057] Where X(k) is the transmitted signal in the frequency domain, Y(k) is the received signal in the frequency domain, H(k) represents the frequency response of the multipath channel, and W(k) represents the noise in the frequency domain.

[0058] like Figure 2 This invention provides a DFT channel estimation method for OFDM-PLC systems, specifically including the following steps:

[0059] Step 1: Use the least squares channel estimation algorithm (LS) to obtain the channel estimate H of the pilot subcarriers. LS And the channel estimation H of the pilot subcarriers. LSThe channel impulse response value h at the sampling point is obtained by converting the data to the time domain using the inverse discrete Fourier transform (IDFT). LS (n);

[0060] Channel estimation H of pilot subcarriers in this invention LS for:

[0061] H LS (k)=Y(k) / X(k)=H(k)+W(k) / X(k),0≤k≤N-1

[0062] Where k represents the subcarrier number, N represents the total number of subcarriers, Y(k) represents the received signal in the frequency domain, X(k) represents the transmitted signal in the frequency domain, H(k) represents the frequency response of the multipath channel, and W(k) represents the noise in the frequency domain.

[0063] The channel impulse response value of the sampling point in this invention is:

[0064]

[0065] Among them, h LS (n) represents the channel impulse response value at the nth sampling point, IDFT[] represents the inverse of the discrete Fourier transform, i represents the imaginary unit, h(n) represents the impulse response of the multipath channel, w(n) represents the noise complex sample in the PLC system, and w(n) = IDFT[W(k) / X(k)].

[0066] Step 2: Since the useful channel CIR is mainly concentrated in the sampling points within the cyclic prefix CP, while the sampling points outside the cyclic prefix CP only include noise, the sampling points are divided into sampling points within the cyclic prefix and noise points according to the length of the cyclic prefix CP. The average noise power of the noise points is calculated through the channel impulse response value of the noise points, and the channel impulse response value of the noise points is set to zero to eliminate the influence of noise as much as possible.

[0067] The process of dividing the sampling points in this invention is as follows:

[0068]

[0069] Among them, L CP This indicates the length of the cyclic prefix (CP).

[0070] The average noise power P at the noise point in this invention w for:

[0071]

[0072] Step 3: Calculate the average power of the sampling points within the cyclic prefix CP, and combine it with the average noise power of the noise points to calculate the average signal-to-noise ratio of the sampling points within the cyclic prefix CP and the alternative signal-to-noise ratio of the sampling points within the cyclic prefix CP;

[0073] In this invention, the average power P of the sampling points within the cyclic prefix CP s+w for:

[0074]

[0075] In this invention, the average signal-to-noise ratio γ of the sampling points within the cyclic prefix CP av for:

[0076]

[0077] In this invention, the alternative signal-to-noise ratio γ of the sampling points within the cyclic prefix CP n for:

[0078]

[0079] Step 4: To more effectively filter out the impulse noise inside the cyclic prefix CP, γ n ≤γ av This indicates that the energy of the noise component at the sampling point is greater than its average level. Since the energy of the impulse noise sample is much greater than that of the background noise sample, this sample can be considered as impulse noise. The channel impulse response value of the sampling points within the cyclic prefix CP, whose signal-to-noise ratio is less than the average signal-to-noise ratio, is set to zero, resulting in the first noise reduction result h. DFT-1 (n):

[0080]

[0081] Step 5: Calculate the total energy of the sampling points within the cyclic prefix CP, and combine it with the average signal-to-noise ratio of the sampling points within the cyclic prefix CP to calculate the average energy of the noise in the sampling points within the cyclic prefix CP;

[0082] The total energy of the sampling points within the cyclic prefix CP in this invention is:

[0083]

[0084] The total energy of the internal noise of the cyclic prefix CP in this invention is:

[0085]

[0086] In this invention, the average energy of the noise in the sampling points within the cyclic prefix CP is:

[0087]

[0088] Step 6: Since the CIR amplitude of a multipath signal is usually greater than the CIR amplitude of noise, during denoising, only sampling points with larger CIR amplitudes need to be retained. The relatively weak background noise within the cyclic prefix (CP) is filtered out. If the energy of the sampling points within the cyclic prefix is ​​less than or equal to the average energy of the noise within the sampling points, since the energy of both impulse noise and signal samples is much greater than the energy of background noise samples, |h DFT-1 (n)| 2 ≤E w-av This indicates that the sample point is background noise and should be filtered out. The channel impulse response value of the sample points excluding the cyclic prefix is ​​set to zero to obtain the secondary noise reduction result. The channel frequency response is then estimated using a discrete Fourier transform.

[0089] The secondary noise reduction result h in this invention DFT-2 (n) is:

[0090]

[0091] The channel frequency response is estimated in this invention as follows:

[0092]

[0093] Example

[0094] The DFT channel estimation method of this invention for OFDM-PLC systems was simulated, and the simulation parameters were set as shown in Table 1.

[0095] Table 1. Simulation Parameters

[0096] Parameter Value Channel Model Rayleigh multipath fading channel multipath number 3 Number of symbols 300 FFT / IFFT points 256 Cyclic prefix length 15 Insert pilot spacing 4 Channel noise model Middleton Class A noise Background noise to impulse noise power ratio 0.001 Modulation method DBPSK

[0097] like Figure 3 Comparison figures between the channel estimation results of the proposed method, the LS channel estimation method, the traditional DFT channel estimation method, the channel estimation method based on the root mean square delay spread approximation, and the channel estimation method based on DFT smoothing filtering, and the actual channel are presented. The results show that, except for the LS channel estimation method, the other methods can obtain effective channel estimates. Compared with the traditional DFT channel estimation method, the channel estimation method based on the root mean square delay spread approximation, and the channel estimation method based on DFT smoothing filtering, the proposed method, through two denoising steps on the sampled signal within the denoising cyclic prefix, can more effectively filter out impulse noise and background noise, thus obtaining a channel estimate that is closer to the actual channel.

[0098] like Figure 4By comparing the channel estimation results of the proposed channel estimation method with the mean square error of the actual channel, the LS channel estimation method, the traditional DFT channel estimation method, the channel estimation method based on the root mean square delay spread approximation, and the channel estimation method based on DFT smoothing filtering, it can be seen that the channel estimation method proposed in this invention has the smallest mean square error between the estimated channel and the actual channel. This indicates that the estimated channel obtained by this algorithm is closer to the actual channel.

[0099] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A DFT channel estimation method for OFDM-PLC systems, characterized in that, Specifically, the steps include the following: Step 1: Use the least squares channel estimation algorithm to obtain the channel estimate of the pilot subcarrier, and transform the channel estimate of the pilot subcarrier to the time domain through the inverse discrete Fourier transform to obtain the channel impulse response value of the sampling point; Step 2: Divide the sampling points into sampling points within the cyclic prefix and noise points according to the length of the cyclic prefix. Calculate the average noise power of the noise points using the channel impulse response values ​​of the noise points, and set the channel impulse response values ​​of the noise points to zero. Step 3: Calculate the average power of the sampling points within the cyclic prefix, and combine this with the average noise power of the noise points to calculate the average signal-to-noise ratio (SNR) of the sampling points within the cyclic prefix and the alternative SNR of the sampling points within the cyclic prefix. : in, This represents the average noise power at the noise point. This represents the channel impulse response value at the nth sampling point. Indicates a cyclic prefix CP Length; Step 4: Set the channel impulse response value of the sampling points in the cyclic prefix whose signal-to-noise ratio is less than or equal to the average signal-to-noise ratio to zero to obtain the first noise reduction result; Step 5: Calculate the total energy of the sampling points within the cyclic prefix. Combined with the average signal-to-noise ratio of the sampling points within the cyclic prefix, calculate the average energy of the noise in the sampling points within the cyclic prefix. : in, This represents the total energy of the sampling points within the cyclic prefix. This represents the average signal-to-noise ratio of the sampling points within the cyclic prefix; Step 6: If the energy of a certain sampling point within the cyclic prefix is ​​less than or equal to the average energy of the noise in the sampling points within the cyclic prefix, set the channel impulse response value of the sampling point whose energy is less than or equal to the average energy of the noise in the sampling points within the cyclic prefix to zero, and obtain the secondary noise reduction result. The channel frequency response is estimated by using discrete Fourier transform.

2. The DFT channel estimation method for an OFDM-PLC system according to claim 1, characterized in that, Channel estimation of pilot subcarriers in step 1 H LS for: in, k N represents the subcarrier sequence number, and N represents the total number of subcarriers. Y ( k ) represents the received signal in the frequency domain. X ( k () represents the transmitted signal in the frequency domain. H ( k This represents the frequency response of a multipath channel. W ( k ) represents noise in the frequency domain.

3. The DFT channel estimation method for an OFDM-PLC system according to claim 2, characterized in that, The channel impulse response value of the sampling point in step 1 is: in, This represents the channel impulse response value at the nth sampling point. This represents the inverse of the discrete Fourier transform. j Represents the imaginary unit. This represents the impulse response of a multipath channel. This represents the complex number of noise samples in the PLC system. .

4. The DFT channel estimation method for an OFDM-PLC system according to claim 3, characterized in that, The process of dividing the sampling points in step 2 is as follows: in, L CP Indicates a cyclic prefix CP The length.

5. The DFT channel estimation method for an OFDM-PLC system according to claim 4, characterized in that, The average noise power at the noise point in step 2 P w for: ， 。 6. The DFT channel estimation method for an OFDM-PLC system according to claim 5, characterized in that, The average power of the sampling points within the cyclic prefix in step 3 P s+w for: ， ； The average signal-to-noise ratio of the sampling points within the cyclic prefix for: 。 7. The DFT channel estimation method for an OFDM-PLC system according to claim 6, characterized in that, Noise reduction result in step 4 for: 。 8. The DFT channel estimation method for an OFDM-PLC system according to claim 7, characterized in that, The total energy of the sampling points within the cyclic prefix in step 5 is: ， ； The total energy of the noise inside the cyclic prefix is: 。 9. A DFT channel estimation method for an OFDM-PLC system according to claim 8, characterized in that, Secondary noise reduction results in step 6 for: 。 10. A DFT channel estimation method for an OFDM-PLC system according to claim 9, characterized in that, The channel frequency response estimation in step 6 is as follows: 。

Images