CN105635021A - Pulse noise combined inhibition method in multicarrier communication system - Google Patents

Abstract

The invention belongs to the field of wireless communication, and relates to a pulse noise combined inhibition method in a multicarrier communication system. The method comprises the following steps: first of all, performing time-domain nonlinear preprocessing on signals to be processed z(k) to obtain y(k) signals; performing FFT to transform the y(k) signals to a frequency domain to obtain frequency-domain signals Y(k); then performing hard decision processing in the frequency domain, and performing IFFT on signals Z'(k) after a hard decision to obtain time-domain signals y'(k); then obtaining n'(k) by subtracting the y'(k) from the signals y(k), and obtaining negative pulse noise estimation i'(k) through calculation by use of the signals n'(k); and finally, subtracting the negative pulse noise estimation i'(k) signals from the y(k). According to the invention, the combined inhibition method is obtained by combining time-domain nonlinear processing with a frequency-domain feedback noise estimation method are combined; and simulation verification succeeds under two typical pulse noise models, i.e., Bernoulli-Gaussian model and a Middleton A-type model, and results show that the method can reduce the symbol error rate to a quite large degree and can realize inhibition with faster speed and high accuracy under high-pulse noise power.

Description

Combined suppression method for impulse noise in multi-carrier communication system

Technical Field

The invention belongs to the field of wireless communication, relates to an impulse noise suppression technology in a multi-carrier communication system, and particularly relates to a combined suppression method of impulse noise in the multi-carrier communication system.

Background

With the rapid development of wireless communication technology, the performance requirement on communication data transmission is higher and higher, and on the other hand, the VHF/UHF band wireless communication is subjected to the interference of impulse noise. In addition to additive white gaussian noise, impulse noise is another additive noise source that causes performance degradation of wireless communication systems, mainly caused by spark plug ignition, electrical switching devices, high voltage lines, lightning, etc. of automobiles. Compared with white gaussian noise, impulse noise is generated with characteristics of burstiness, short pulses, high energy and sharpness. Research shows that the influence of impulse noise on a wireless communication system is mainly distributed on VHF and lower UHF frequency bands, and the influence of the impulse noise in the frequency band range is more serious along with the modernization of cities and the popularization of electrical equipment.

In order to reduce the equalization complexity under the frequency selective fading channel and improve the robustness to impulse noise, OFDM (orthogonal frequency division multiplexing) is proposed and has been widely used in wireless systems such as DTMB (digital television terrestrial broadcasting), DVB (digital video broadcasting), WiMAX (worldwide interoperability for microwave access), and the like and PLC power line carrier communication. It is mentioned in the literature that OFDM is inherently robust to low energy impulse noise relative to single carrier systems, since the DFT (discrete fourier transform) processing at the receiver end of OFDM spreads the impulse noise in one OFDM symbol period in the time domain over the frequency domain. On one hand, because the impulse noise energy is averaged to all the sub-carriers, this weakens the influence of the impulse noise to some extent, but on the other hand, because of the energy spreading effect of DFT, the impulse noise with high energy deteriorates the carrier-to-energy-plus-noise ratio of each sub-carrier after frequency domain spreading, which results in more serious symbol decision errors.

Since the occurrence probability of impulse noise is small, and the influence of impulse noise on OFDM depends on the total energy of the impulse within a single OFDM symbol period, the OFDM system with a large number of carriers is more advantageous in reducing the influence of the impulse. For a wireless communication system in a low UHF band, an additional algorithm must be applied to suppress impulse noise in the physical layer due to the relative shortage of the code interleaving depth and the number of carriers.

Aiming at pulse noise commonly existing in VHF/UHF frequency bands, the current main suppression methods include time domain nonlinear processing, interleaving depth increasing, filter compensation methods and the like. The time domain nonlinear processing mainly comprises methods of zero clearing, amplitude limiting, combination of zero clearing and amplitude limiting and the like. The algorithms are simple and easy to implement, show good characteristics in practical application, and particularly play a good role in inhibiting high-energy impulse noise. However, it is not easy to find a proper zero clearing or amplitude limiting threshold value in the time domain processing, and meanwhile, negative noise is introduced through the time domain processing algorithm, so that the method cannot bring great performance improvement. The OFDM characteristic is utilized to adopt a block interleaving technology in a time domain, so that the influence of impulse noise can be further weakened; the method has good performance in resisting pulse noise of different degrees in a low-order modulation mode, but under high-order modulation, particularly under the environment of strong pulse, the increase of the interleaving depth only brings more serious bit error rate, and simultaneously the byte interleaving also increases the complexity of a system; meanwhile, the method has certain dependence on the actual noise environment, and simultaneously, the trade-off between the system performance and the complexity is required. The filter compensation method is a periodic impulse noise suppression algorithm, the algorithm carries out impulse noise detection of multiple OFDM symbols in a frequency domain, and the position of the impulse noise on a subcarrier is determined by a majority voting method; meanwhile, a multi-order self-adaptive IIR notch filter is designed before the system is synchronized to inhibit periodic noise, then a distorted received signal is compensated through the IIR notch filter, Forward Error Correction (FEC) processing is carried out at a decoding end, and a large error rate is obtained at a certain bandwidth cost and improved; however, the limitation of this method is also obvious, and only the periodic impulse noise can be suppressed, but the method has no suppression effect on the burst impulse noise.

From the above background, in an actual noise environment, the conventional impulse noise suppression method faces the problems of high system complexity, limited performance improvement space, high dependence on the environment, and the like, so that the suppression performance is difficult to meet the requirement of high-speed accurate data stream. Because impulse noise has the characteristics of burstiness, short duration, high energy and sharpness, impulse noise has a high requirement on a physical layer design inhibition method due to the fact that the impulse noise environment is severe and the system complexity is considered.

Disclosure of Invention

The invention aims to provide a combined suppression method of impulse noise in a multi-carrier communication system, which combines the characteristics of impulse noise in a burst noise environment and combines time domain nonlinear processing and a frequency domain feedback noise estimation method; simulation verification is carried out under two typical pulse noise models of Bernoulli-Gaussian and Middletona, and results show that the method can reduce the symbol error rate to a great extent and achieve the suppression of higher speed and higher accuracy under the condition of high pulse noise power. The technical scheme adopted by the invention is as follows:

a method for jointly suppressing impulse noise in a multi-carrier communication system comprises the following steps:

step 1, performing time domain nonlinear preprocessing on a signal z (k) to be processed to obtain a signal y (k);

step 2, performing FFT on the y (k) signal subjected to time domain nonlinear preprocessing to transform the signal into a frequency domain to obtain a frequency domain signal Y (k);

step 3, carrying out hard decision processing in a frequency domain to obtain a signal Z' (k);

step 4, performing IFFT transformation on the signal Z '(k) after hard decision to obtain a time domain signal y' (k);

step 5, subtracting the signal y '(k) obtained in the step 4 from the signal y (k) obtained in the step 1 to obtain n' (k);

step 6, calculating to obtain a negative impulse noise estimation i '(k) by using the signal n' (k);

and 7, subtracting the signal obtained in the step 6 from the y (k) obtained in the step 1 to obtain a negative impulse noise estimation i '(k) signal, and obtaining an impulse noise combined suppression signal z' (k).

Further, the time domain nonlinear preprocessing in the step 1 adopts a zero setting method.

The specific steps of the step 6 are as follows:

first, the average power of the signal n' (k) is calculated as S,

$S=\sqrt{\underset{k=0}{\overset{N-1}{\Σ}}|n^{\′}\left(k\right){|}^2}$

the undershoot noise estimate is then:

$i^{\&prime;}\left(k\right)=\{\begin{array}{cc}{n}^{\&prime;}\left(k\right)& \left|n^{\&prime;}\left(k\right)\right|\&GreaterEqual;C\&CenterDot;S\\ 0& \left|n^{\&prime;}\left(k\right)\right|<C\&CenterDot;S\end{array},$ and C is an impulse noise estimation threshold value.

The invention has the following effects: the method combines the characteristics of impulse noise in a burst noise environment and combines time domain nonlinear processing and frequency domain feedback noise estimation methods; the influence of impulse noise on multi-carrier communication is effectively inhibited, and the robustness of a wireless communication system is greatly improved.

Drawings

FIG. 1 is a system flow diagram.

Fig. 2 is a block diagram of a time-frequency joint suppression process.

Fig. 3 shows the sampled values of Bernoulli-Gaussian noise.

Fig. 4 shows sampled values of MiddletonA type noise.

FIG. 5 is a diagram of a blank optimal threshold simulation result.

FIG. 6 is a graph of theoretical performance for different Blanking thresholds.

Fig. 7 is a schematic diagram of a hard decision process.

Fig. 8 shows the cancellation process in the joint suppression of impulse noise.

Fig. 9 is a performance curve of different suppression methods from the viewpoint of SNR under the Bernoulli-Gaussian noise model.

Fig. 10 is a performance curve of different suppression methods from the viewpoint of SINR under the Bernoulli-Gaussian noise model.

FIG. 11 is a performance curve of the joint suppression algorithm under the Bernoulli-Gaussian noise model.

Fig. 12 is a performance curve of different suppression methods from the SNR perspective under the Middleton noise model.

Fig. 13 is a performance curve of different suppression methods from the SINR perspective under the Middleton noise model.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples.

The system flow chart of this embodiment is shown in fig. 1, which shows all the modules through which the data stream passes, and this is also a schematic block diagram of the OFDM system standard, in order to reproduce impulse noise, two typical impulse noise models of Bernoulli-Gaussian and MiddletonA are added to the channel, and ideal channel estimation and ideal timing synchronization are assumed.

The time-frequency joint suppression implementation block diagram is shown in fig. 2, and mainly comprises a time domain nonlinear processing module, an FFI and IFFT module, a hard decision device, and an impulse noise estimator. Z (k) represents a signal to be suppressed received by a receiving end, y (k) represents a signal subjected to time domain nonlinear preprocessing, y (k) represents a result of re-FFT of the preprocessed signal, Z ' (k) is a result of the signal subjected to the hard decision device, y ' (k) is a result of the IFFT of the hard decision signal returned to the time domain, y (k) subtracts the result of y ' (k) to obtain an estimated value n ' (k) to be estimated of impulse noise, i ' (k) represents a result of the signal subjected to the impulse noise estimator, and finally subtracts the estimated value i ' (k) of the impulse noise from y (k) to obtain Z ' (k).

First, two impulse noise models of the Bernoulli-Gaussian and MiddletonA classes, both of which are additive noise, are introduced. The Bernoulli-Gaussian pulse noise model is a Gaussian noise model, wherein pulse noise is distributed in all time domain symbols and is expressed as a product of an independent and identically distributed Bernoulli random process and a zero-mean Gaussian process, and main parameters of the model are as follows: is the probability P of occurrence of burst noise, the variance of additive white Gaussian noiseAnd variance of burst noise

Under the Bernoulli-Gaussian noise model, the probability of the transmitted data is P, and the power is POtherwise only subject to a variance ofIs interfered by gaussian white noise, the probability density function is thus expressed as:

$f\left(x\right)=\frac{P}{\sqrt{2\mathrm{\&pi;}({\mathrm{\&sigma;}}_g^2+{\mathrm{\&sigma;}}_i^2)}}\exp \&lsqb;-\frac{x^2}{2({\mathrm{\&sigma;}}_g^2+{\mathrm{\&sigma;}}_i^2)}\&rsqb;+\frac{1-P}{\sqrt{2{\mathrm{\&pi;\&sigma;}}_g^2}}\exp (-\frac{x^2}{2{\mathrm{\&sigma;}}_g^2})$

the MATLAB simulated time domain sequence is shown in fig. 3.

Middleton is a non-Gaussian narrow-band noise model with white Gaussian noise as the main parameterImpulse noiseThe probability density function is expressed as:

$p\left(x\right)=e^{-A}\underset{m=0}{\overset{\mathrm{\&infin;}}{\&Sigma;}}\frac{A^m}{m!\sqrt{2{\mathrm{\&pi;\&sigma;}}_m^2}}\exp (-\frac{|x{|}^2}{2{\mathrm{\&sigma;}}_m^2})$

wherein,

${\mathrm{\&sigma;}}_m^2=\frac{{\mathrm{\&sigma;}}^2(\frac{m}{A}+\mathrm{\&Gamma;})}{1+\mathrm{\&Gamma;}},\mathrm{\&Gamma;}={\mathrm{\&sigma;}}_G^2/{\mathrm{\&sigma;}}_I^2,{\mathrm{\&sigma;}}^2={\mathrm{\&sigma;}}_G^2+{\mathrm{\&sigma;}}_I^2$

the probability density function for class a noise derives from the assumption that the number of pulses affecting the receiver obeys a poisson distribution, the parameter a and the pulse condition that controls the noise. A is an impulse index which represents the average number of impulses affecting the normal demodulation of the receiver in one symbol duration and is the ratio of the power of impulse noise to the power of Gaussian noise; when a <1, the impulsiveness of the noise becomes stronger, and when a >1, the distribution of the noise approaches a gaussian distribution; when <1, the pulse is strong, and when >1, the Gaussian characteristic is more obvious; the simulated time domain sequence is shown in fig. 4.

The time domain nonlinear processing module adopts a zero setting (Blanking) method, and the impulse noise exceeding a specific threshold value T directly carries out zero clearing processing on the time point:

$y_k=\{\begin{array}{cc}{r}_k,& \left|r_k\right|\&le;T\\ 0,& \left|r_k\right|>T\end{array},k=0,1,\mathrm{...},N-1$

wherein, N is the subcarrier number, T is the threshold value, its size of choosing has two kinds of methods: firstly, through simulation acquisition, it is known that when a threshold value is too high, a lot of impulse noise may not be detected, the probability of missed detection may increase, and when the threshold value is too low, the transmitted signal data may be treated as impulse noise, which may increase the probability of false alarm; under the simulation result of a large amount of data, an optimal threshold value is ensured to exist, so that the false alarm probability and the missed detection probability are optimal under the threshold value of the point, and the symbol error rate of the system is the lowest at the moment; as shown in fig. 5, a relation curve (using a modulation scheme of 16 QAM) between the threshold and the bit error rate under different simulated SINRs (i.e., signal-to-interference-and-noise ratio, where the power of the signal is divided by the sum of the impulse noise and the background noise power) is shown, and it can be seen from the graph that the position of the optimal threshold appears at about T ═ 3.1. Secondly, the maximum signal-to-noise ratio formula is output in the Blanking process, which is obtained by theoretical derivation:

$\mathrm{\&gamma;}={(\frac{E\&lsqb;|y_k{|}^2\&rsqb;}{K_0^2}-1)}^{-1}$

wherein,

$K_0=1-\underset{m=0}{\overset{1}{\&Sigma;}}{p}_m(1+\frac{T^2}{1+{\mathrm{\&sigma;}}_m^2})e\frac{T^2}{1+{\mathrm{\&sigma;}}_m^2}$

$E\&lsqb;|y_k{|}^2\&rsqb;=1+\underset{m=0}{\overset{1}{\&Sigma;}}{p}_m({\mathrm{\&sigma;}}_m^2-\&lsqb;T^2+(1+{\mathrm{\&sigma;}}_m^2)\&rsqb;e^{-\frac{T^2}{1+{\mathrm{\&sigma;}}_m^2}})$

by plotting the output SNR against SIR (i.e. signal to interference ratio, signal power to impulse noise power ratio) at different threshold values T as shown in fig. 6, it can be seen that T ═ 3.1 is a folding value of the output high SNR at the impulse noise of different powers.

The FFT and IFFT modules are the most basic modules in OFDM systems as well, the FFT is a fast algorithm to implement DFT (discrete fourier transform), the transform formulas of DFT and IDFT are as follows:

$X\left(k\right)=\underset{n=0}{\overset{N-1}{\&Sigma;}}x\left(n\right)e^{-j\frac{2\mathrm{\&pi;}nk}{N}}$

$x\left(n\right)=\frac{1}{N}\underset{k=0}{\overset{N-1}{\&Sigma;}}X\left(k\right)e^{j\frac{2\mathrm{\&pi;}nk}{N}}$

hard decision is also a non-linear processing procedure, and the module is based on constellation mapping of the OFDM system; the hard decision device can help the system to compensate the negative noise introduced during the time domain nonlinear processing, and the hard decision process of 16QAM in this embodiment is shown in fig. 7.

After processes of FFT, hard decision, IFFT and the like, the power of negative impulse noise introduced by time domain nonlinear processing is equally divided into each subcarrier, and the power is subtracted from y (k) to obtain estimated values of impulse noise and background noise; because the negative noise is introduced by time domain nonlinear preprocessing, the power of the negative noise is not very large, after FFT conversion is carried out, the power of the negative impulse noise is divided into the average power of each subcarrier according to the inherent low-power impulse noise resistance characteristic of OFDM, and the process of hard decision cannot cause large symbol decision errors; the existence of negative impulse noise raises only slightly the noise plane of y ' (k), and the subtraction of y (k) and y ' (k) extracts the useful data component in y (k), which theoretically illustrates the feasibility of estimating n ' (k).

The input signal n ' (k) of the impulse noise estimator is the result of subtracting the hard decision feedback signal y ' (k) from the signal y (k) which is subjected to the nonlinear preprocessing, and the y ' (k) actually estimates the negative impulse noise and the background noise which are introduced by the time domain nonlinear preprocessing. Since the energy of the undershoot noise is more than ten and several dB higher than the energy of the background noise, the undershoot noise can be easily detected in the background noise, the average power of the n' (k) signal is obtained as S,

$S=\sqrt{\underset{k=0}{\overset{N-1}{\&Sigma;}}|n^{\&prime;}\left(k\right){|}^2}$

only the impulse noise estimation threshold C needs to be set, the estimation of the negative impulse noise is:

$i^{\&prime;}\left(k\right)=\left\{\begin{array}{cc}{n}^{\&prime;}\left(k\right)& \left|n^{\&prime;}\left(k\right)\right|\&GreaterEqual;C\&CenterDot;S\\ 0& \left|n^{\&prime;}\left(k\right)\right|<C\&CenterDot;S\end{array}\right.$

finally, negative pulse noise is eliminated at the time domain end:

z'(k)＝y(k)-i'(k),k＝0,...,N-1

the cancellation process of impulse noise in joint suppression is shown in fig. 8.

An OFDM simulation platform is constructed, the number of subcarriers N is 4096, a 16QAM modulation mode is adopted, a non-linear preprocessing zero setting threshold T is 3.1, an impulse noise estimation threshold value C is 4.5, signals are transmitted in two paths of I, Q, and finally a simulation result is given.

The method for jointly suppressing the impulse noise comprises the following specific steps:

step 1, carrying out zero setting nonlinear preprocessing on a signal z (k) to be processed, carrying out zero clearing on sampling points larger than a threshold value T, and not processing the sampling points lower than the threshold value to obtain a signal y (k);

step 2, performing FFT on the y (k) signal subjected to time domain nonlinear preprocessing to transform the signal into a frequency domain to obtain a frequency domain signal Y (k);

step 3, carrying out hard decision processing in a frequency domain to obtain a signal Z' (k);

step 4, performing IFFT on the signal after hard decision to obtain a time domain signal y' (k);

step 5, subtracting the signal y '(k) obtained in the step 4 from the signal y (k) obtained in the step 1 to obtain n' (k);

step 6, calculating to obtain a negative impulse noise estimation i '(k) by using the signal n' (k);

and 7, subtracting the signal of the negative impulse noise estimation i '(k) obtained in the step 6 from the signal y (k) obtained in the step 1 to obtain a final signal z' (k) after the impulse noise combined suppression.

Fig. 9 to 11 show the simulation performance of the present invention under the Bernoulli-Gaussion noise model, and fig. 12 and 13 show the simulation performance under the Middleton noise model. A series of simulations are carried out on different background noise powers, different impulse noise powers and two noise models, and simulation results show that the method combining time domain Blanking nonlinear processing and hard decision feedback reprocessing has excellent suppression performance, particularly the suppression performance on impulse noise with different powers is quite excellent and is very close to a theoretical curve.

While the invention has been described with reference to specific embodiments, any feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise; all of the disclosed features, or all of the method or process steps, may be combined in any combination, except mutually exclusive features and/or steps.

Claims (3)

1. A method for jointly suppressing impulse noise in a multi-carrier communication system comprises the following steps:

step 1, performing time domain nonlinear preprocessing on a signal z (k) to be processed to obtain a signal y (k);

step 2, performing FFT on the y (k) signal subjected to time domain nonlinear preprocessing to transform the signal into a frequency domain to obtain a frequency domain signal Y (k);

step 3, carrying out hard decision processing in a frequency domain to obtain a signal Z' (k);

step 4, performing IFFT transformation on the signal Z '(k) after hard decision to obtain a time domain signal y' (k);

step 5, subtracting the signal y '(k) obtained in the step 4 from the signal y (k) obtained in the step 1 to obtain n' (k);

step 6, calculating to obtain a negative impulse noise estimation i '(k) by using the signal n' (k);

and 7, subtracting the signal obtained in the step 6 from the y (k) obtained in the step 1 to obtain a negative impulse noise estimation i '(k) signal, and obtaining an impulse noise combined suppression signal z' (k).

2. A method for joint suppression of impulse noise in a multi-carrier communication system as defined in claim 1, wherein said time-domain non-linear preprocessing in step 1 employs a nulling method.

3. The method for joint suppression of impulse noise in a multi-carrier communication system as set forth in claim 1, wherein said step 6 comprises the steps of:

first, the average power of the signal n' (k) is calculated as S,

$S=\sqrt{\underset{k=0}{\overset{N-1}{\&Sigma;}}|n^{\&prime;}\left(k\right){|}^2};$

second, the undershoot noise estimate is: