CN103200139A - OFDM signal bandwidth blind estimating method - Google Patents

Abstract

The invention discloses an OFDM signal bandwidth blind estimating method which includes the following steps. Firstly, a received OFDM signal undergoes Welch changing and a power spectrum is obtained. Secondly, the obtained power spectrum undergoes EMD decomposition and multiple IMF components are obtained. Thirdly, a threshold lambda of IMF variance contribution rate is set. IMF components of which the IMF variance contribution rates are smaller than lambda are removed. By means of reserved IMF components, filtering and noise abating are performed and an OFDM signal power spectrum is reconstructed. Fourthly, a differential of the reconstructed power spectrum is worked out. Positions of the maximum value and the minimum value of an impulse are abstracted and respectively serve as a starting point and a closing point of an OFMD signal passband and a differential value of the maximum value and the minimum value is the width of the passband so that the width of the OFDM signal is estimated. Fifthly, multiple circulation times are set, the previous four steps are repeated and an average statistics value of the bandwidth is estimated and serves as an estimated value of a precise bandwidth of the OFMD signal.

Description

A kind of ofdm signal bandwidth blind estimating method

Technical field

The invention belongs to communication technical field, be specifically related to the ofdm signal bandwidth blind estimating method of low complex degree under a kind of low signal-to-noise ratio multipath channel, can be used for ofdm system demodulation and spectrum monitoring in the non-cooperative communication.

Background technology

OFDM (OFDM) technology is a kind of multi-carrier modulation technology with operating factor of highband, and it has advantages such as antinoise, anti-multipath, suitable high speed data transfer, so the OFDM technology obtains application more and more widely in the communications field.In non-cooperative communication system, accurate ofdm signal bandwidth parameter is significant for spectrum monitoring and ofdm system reconstruct, so the present invention announces a kind of ofdm signal bandwidth blind estimating method.

At present less for the frequency domain bandwidth Estimation Study of ofdm signal, method the earliest is at first to obtain signal spectrum with FFT, simply signal bandwidth is estimated that the estimation of pure like this vision can not obtain accurate estimated result by people's vision then.2000, people such as Walter Akmouche have proposed a kind ofly based on method of wavelet the bandwidth of ofdm signal to be estimated, referring to Walter Akmouche, Eric Kerherve, et al., " OFDM spectral characterization:estimation of the bandwidth and the number of sub-carriers; " [J] .Statistical Signal and Array Processing2000, Proceedings of the Tenth IEEE Workshop on.Aug.14-162000, pp.48-52.This method only obtains ideal effect under the high s/n ratio situation.2005, people such as Liu Peng have proposed ofdm signal bandwidth estimation method under a kind of low signal-to-noise ratio multipath channel, referring to Peng Liu, Bing-bing Li, Zhao-yang Lu, Feng-kui Gong.An OFDM Bandwidth Estimation Scheme for Spectrum Monitoring[J] .WCNM05,2005:228-231.Though higher correct expectancy rate is arranged this method but amount of calculation is bigger.

Summary of the invention

The objective of the invention is to overcome the deficiency that ofdm signal bandwidth estimation accuracy rate is low under the low signal-to-noise ratio multipath channel, amount of calculation is big, provide a kind of ofdm signal bandwidth blind estimating method of low complex degree, to realize the accurate estimation of ofdm signal bandwidth under the low signal-to-noise ratio multipath channel.

Realize the technical scheme of the object of the invention, comprise the steps:

(1) ofdm signal that receives is carried out the Welch conversion, try to achieve power spectrum;

(2) power spectrum that obtains is carried out EMD and decompose, obtain a plurality of IMF components;

(3) IMF variance contribution ratio threshold value λ is set, rejects the IMF variance contribution ratio less than the IMF component of λ, and utilize the IMF component that keeps to carry out filtering de-noising and reconstruct ofdm signal power spectrum;

(4) power spectrum of reconstruct is differentiated, extract the maximum of pulse and the position at minimum value place, respectively as starting point and the cut-off point of ofdm signal passband, its difference is come the bandwidth of estimating OFDM signal as the width of passband with this with it.

(5) set repeatedly cycle-index and repeating step (1)～step (4), obtain the assembly average of estimated bandwidth as the accurate bandwidth estimation value of ofdm signal.

The advantage that the present invention has compared with prior art:

Empirical modal is decomposed (EMD) in the present invention and variance contribution ratio combines, and carries out de-noising by casting out the IMF component that variance contribution ratio is lower than threshold value, compares with traditional wavelet threshold de-noising, and de-noising effect is better; Simultaneously do not need to extract a plurality of maximum covariance values and omitted histogrammic statistics, computation complexity has reduced.Experiment simulation shows that under the condition of multipath and SNR=0dB, correct expectancy rate has reached 99.85%, and not only performance is more excellent and complexity is lower in visible the present invention.

Description of drawings

Fig. 1 is ofdm signal bandwidth blind estimating method flow chart of the present invention;

Fig. 2 is based on the noise-eliminating method of variance contribution ratio and the performance comparison figure of traditional noise-eliminating method among the present invention;

Fig. 3 is the ofdm signal power spectrum of trying to achieve with the Welch method among the present invention, and it is carried out EMD decompose each IMF component obtain and the oscillogram of residual volume R7;

Fig. 4 is the power spectrum restructuring graph after the EMD de-noising of using among the present invention;

Fig. 5 is to based on the figure as a result behind the power spectrum differential after the EMD de-noising reconstruct among the present invention;

Fig. 6 is the method and the performance comparison figure of conventional method under different signal to noise ratios among the present invention.

Embodiment

Please refer to Fig. 1, the ofdm signal that uses among the present invention is DVB-T ofdm signal 2K FFT pattern.

Suppose that the baseband OFDM signal that receives is by after the multipath channel:

$r\left(t\right)=\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{h}_l\left(t\right)s(t-{\mathrm{\τ}}_l)+\mathrm{\μ}\left(t\right)---1)$

Wherein μ (t) is additive white Gaussian noise, h _l(t) be the complex gain of multipath, τ _lBe the path delay of multipath, L is the path number, and s (t) has protection OFDM time domain baseband signal at interval, and its expression formula is:

$s\left(t\right)=\sqrt{\frac{P_0}{{\mathrm{<figure-callout\; id="0"\; label="N"\; filenames="HDA00003035230100021.png,HDA00003035230100031.png"\; state="\{\{state\}\}">N</figure-callout>}}_0}}\underset{k}{\mathrm{\&Sigma;}}\underset{n=0}{\overset{N_0-1}{\mathrm{\&Sigma;}}}{c}_{n,k}*e^{j(\mathrm{\&delta;\&phi;}+2\mathrm{\&pi;}(\mathrm{\&delta;}{\mathrm{<figure-callout\; id="0"\; label="f"\; filenames="HDA00003035230100021.png,HDA00003035230100031.png"\; state="\{\{state\}\}">f</figure-callout>}}_0+\mathrm{n\&Delta;f})(t-k{T}_s))}*g(t-{\mathrm{kT}}_s)---2)$

N wherein ₀Be the subcarrier number, P ₀Be the power of signal, c _{N, k}Be data symbol, it is in k sub-channel of n OFDM symbol, and δ φ is phase error, δ f ₀Be frequency shift (FS), Δ f is subcarrier spacing, and g (t) is impulse function, T _sIt is the OFDM symbol lengths.

Its specific implementation step is as follows:

Step 1 is tried to achieve power spectrum with the Welch conversion to the received signal.Its step is as follows:

1.1) be the data x (n) of N with length, n=0,1 ..., N-1 is divided into the L section, and every section has M data, and the i segment data is expressed as:

x _i(n)＝x(n+iM-M),0≤n≤M,1≤i≤L 3)

1.2) window function w (n) is added on each data segment, obtain the periodogram of each section, the periodogram of i section is:

$I_i\left(\mathrm{\&omega;}\right)=\frac{1}{U}{\left|\underset{n=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{x}_i\left(n\right)w\left(n\right)e^{-\mathrm{j\&omega;n}}\right|}^2,i=\mathrm{1,2}\&CenterDot;\&CenterDot;\&CenterDot;,M-1---4)$

In the formula, ω is the frequency domain variable, and U is called normalization factor, and its expression formula is:

$U=\frac{1}{M}\underset{n=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{w}^2\left(n\right)---5)$

1.3) regard uncorrelated mutually as with approximate between the periodogram of each section, last power spectrum estimation is:

$P_{\mathrm{xx}}\left(e^{\mathrm{j\&omega;}}\right)=\frac{1}{L}\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{I}_i\left(\mathrm{\&omega;}\right)---6)$

Formula (6) is asked statistical average, obtains:

$E\left[P_{\mathrm{xx}}\left(e^{\mathrm{j\&omega;}}\right)\right]=\frac{1}{2\mathrm{\&pi;}}{\&Integral;}_{-\mathrm{\&pi;}}^{\mathrm{\&pi;}}{P}_{\mathrm{xx}}\left(e^{\mathrm{j\&theta;}}\right)W\left(e^{j(w-\mathrm{\&theta;})}\right)\mathrm{d\&theta;}---7)$

In the formula,

$W\left(e^{\mathrm{j\&omega;}}\right)=\frac{1}{\mathrm{MU}}{\left|\underset{n=0}{\overset{M-1}{\mathrm{\&Sigma;}}}w\left(n\right)e^{\mathrm{j\&omega;}}\right|}^2---8)$

Step 2 is carried out EMD to the power spectrum that obtains and is decomposed, and obtains a plurality of IMF components.

To the ofdm signal x (t) that receives, the step that resolves into some rank IMF component with EMD is as follows:

2.1) determine all maximum points and the minimum point of signal x (t).Then all maximum points and all minimum points are coupled together with the cubic spline interpolation curve respectively, so just determine the coenvelope line x of signal _Up(t) and lower envelope line x _Low(t), the average of remembering upper and lower envelope is m ₁₁(t), then:

m ₁₁(t)＝(x _up(t)+x _low(t))/2 9)

2.2) deduct the average m of upper and lower envelope with signal x (t) ₁₁(t), obtain s ₁₁(t), then:

s ₁₁(t)＝x(t)-m ₁₁(t) 10)

Check s ₁₁(t) whether satisfy two conditions of IMF, satisfied then s ₁₁(t) compose to c ₁(t); Otherwise with s ₁₁(t) be used as primary signal and repeat above process, the s that after k circulation, obtains _1k(t) satisfy two conditions of IMF, and remember: c ₁(t)=s _1k(t).So far, obtain first IMF component, i.e. c ₁(t);

2.3) note residual signal r ₁(t)=x (t)-c ₁(t), as new sequence, step above repeating is until extracting all IMF components with it. and primary signal x this moment (t) finally is decomposed into n IMF and a residual components r _n(t):

$x\left(t\right)=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{c}_i\left(t\right)+r_n\left(t\right)---11)$

Each the IMF component that obtains behind EMD has constituted the basic function of signal, and this basic function directly comes from signal, is adaptive.

Step 3 arranges IMF variance contribution ratio threshold value λ, rejects the IMF variance contribution ratio less than the IMF component of λ, and utilizes the IMF component that keeps to carry out filtering de-noising and reconstruct ofdm signal power spectrum.

Because the frequency spectrum of desirable ofdm signal is rectangle, at the starting point of passband and cut-off point place because saltus step takes place, so variance is very big. this paper selects suitable component to carry out denoising and reconstruct power spectrum according to the variance contribution ratio of each IMF component.With the variance contribution ratio of the IMF component standard as the filtering de-noising, its process is as follows:

3.1) at first determine the population variance of all IMF components, calculate the variance contribution ratio of each IMF component more respectively, be designated as μ _i, 0＜μ _i＜1, i=1,2 ..., n;

3.2) ask threshold value λ=max (μ _i) η, 0＜η＜1;

3.3) if μ _i〉=λ then keeps this IMF component, otherwise removes this IMF component;

3.4) the IMF component reconstruct power spectrum that utilize to keep.

The IMF component that is higher than thresholding has comprised the saltus step information of passband in the multi-OFDM signal more, has kept the starting point of ofdm signal passband and the saltus step information at cut-off point place well with the power spectrum of its reconstruct.When casting out the IMF component that is lower than decision threshold, removed noise, reached the purpose of de-noising.Though part signal information also is removed, make waveform after the reconstruct produce to a certain degree distortion inevitably, do not cause the distortion at the trip point place of passband.As seen, the method announced of the present invention is not only simple but also can not influence estimation effect to the ofdm signal bandwidth.

Step 4 is differentiated to the power spectrum of reconstruct, extracts the maximum of pulse and the position at minimum value place, and respectively as starting point and the cut-off point of ofdm signal passband, its difference is come the bandwidth of estimating OFDM signal as the width of passband with this with it.

Step 5 is set repeatedly cycle-index and repeating step 1～step 4, obtains the assembly average of estimated bandwidth as the accurate bandwidth estimation value of ofdm signal.

Effect of the present invention can further specify by emulation:

1. simulated environment

The present invention is when emulation, and the ofdm signal of use is DVB-T2K FFT pattern, and transmission mode is non-layered, and modulation system is 64QAM, and protection is spaced apart 1/4, and interior encoder bit rate is 2/3, the valid data length T _u=224 μ s, protection gap length T _g=0.25T _u, symbol total length T _s=280 μ s, and adopt GSMTU6 footpath channel model, Doppler frequency shift 40Hz, frequency shift (FS) 10kHz, phase error is π/6.

In order to verify the validity of method provided by the present invention, use Welch conversion employing segmentation aliasing rate is 128 hamming windows of 50% in the emulation, after the Welch conversion, obtain the power spectrum of ofdm signal, decompose through EMD again and obtain a plurality of IMF components, select suitable parameters η to determine threshold value, select suitable component reconstruct power spectrum based on variance contribution ratio at last, thereby the bandwidth to its differential estimating OFDM signal, carry out covering for 200 times the test of Taka sieve, ask statistical average, with the accurate estimated bandwidth of its result as ofdm signal.

2. emulation content and result:

The selection of parameter η affects the selection of IMF component in the filtering de-noising process, therefore selects suitable η value most important.Simultaneously, in order to verify the validity based on the denoising method of each IMF component variance contribution rate, people's such as the method and Jiang Li the wavelet threshold denoising method based on EMD is compared.Fig. 2 is that η gets different values respectively under different signal to noise ratios, and when using the wavelet threshold denoising method, to the evaluated error of ofdm signal bandwidth.As can be seen from Figure 2, compare with people's such as Jiang Li the wavelet threshold denoising based on EMD, not only simple but also can estimate signal bandwidth more accurately based on the denoising method of variance contribution ratio, and when η=0.05, bandwidth estimation error minimum, so in this emulation, the value of η elects 0.05 as.

Fig. 3 has showed the ofdm signal power spectrum of trying to achieve with the Welch method, and it is carried out EMD decompose each IMF component obtain and the oscillogram of residual volume R7.

Fig. 4 is the power spectrum restructuring graph after the EMD de-noising used of the present invention.As can be seen from Figure 4, though in denoising, removed the part useful information based on the method for EMD de-noising, cause distortion to a certain extent, fully kept the saltus step information at ofdm signal passband starting point and cut-off point place.

Fig. 5 is to based on the figure as a result behind the power spectrum differential after the EMD de-noising reconstruct, because saltus step takes place, produces positive and negative two pulses clearly at the starting point of passband and cut-off point place, with the position at its place respectively as starting point and the cut-off point of passband.

Fig. 6 performance comparison figure of method under different signal to noise ratios that to be the inventive method propose with Liu Peng, as can be seen from the figure, the inventive method correct expectancy rate under the condition of multipath and SNR=0dB can reach 99.85%.Its performance obviously is better than the method for Liu Peng, and visible method of the present invention can estimate the bandwidth of ofdm signal more accurately under the low signal-to-noise ratio multipath channel.In addition, the method for Liu Peng needs altogether

Inferior complex multiplication; And side of the present invention needs altogether Inferior complex multiplication, N wherein is signal length, M is every segment signal number, M＜N, and method of the present invention need not extract the covariance value of a plurality of maximums and omit statistics with histogram, so the complexity of the inventive method decreases than conventional method.Take all factors into consideration the complexity of performance and calculating, visible the inventive method is better than conventional method.

Claims (3)

1. an ofdm signal bandwidth blind estimating method comprises the steps:

(1) ofdm signal that receives is carried out the Welch conversion, try to achieve power spectrum;

(2) power spectrum that obtains is carried out EMD and decompose, obtain a plurality of IMF components;

(3) IMF variance contribution ratio threshold value λ is set, rejects the IMF variance contribution ratio less than the IMF component of λ, and utilize the IMF component that keeps to carry out filtering de-noising and reconstruct ofdm signal power spectrum;

(4) power spectrum of reconstruct is differentiated, extract the maximum of pulse and the position at minimum value place, respectively as starting point and the cut-off point of ofdm signal passband, its difference is come the bandwidth of estimating OFDM signal as the width of passband with this with it.

(5) set repeatedly cycle-index and repeating step (1)～step (4), obtain the assembly average of estimated bandwidth as the accurate bandwidth estimation value of ofdm signal.

2. a kind of ofdm signal bandwidth blind estimating method as claimed in claim 1, wherein step (2) is described carries out EMD to the power spectrum that obtains and decomposes, and carries out as follows:

2.1) determine all maximum points and the minimum point of signal x (t).Then all maximum points and all minimum points are coupled together with the cubic spline interpolation curve respectively, so just determine the coenvelope line x of signal _Up(t) and lower envelope line x _Low(t), the average of remembering upper and lower envelope is m ₁₁(t), then:

m ₁₁(t)＝(x _up(t)+x _low(t))/2

2.2) deduct the average m of upper and lower envelope with signal x (t) ₁₁(t), obtain s ₁₁(t), then:

s ₁₁(t)＝x(t)-m ₁₁(t)

Check s ₁₁(t) whether satisfy two conditions of IMF, satisfied then s ₁₁(t) compose to c ₁(t); Otherwise with s ₁₁(t) be used as primary signal and repeat above process, the s that after k circulation, obtains _1k(t) satisfy two conditions of IMF, and remember: c ₁(t)=s _1k(t).So far, obtain first IMF component, i.e. c ₁(t);

2.3) note residual signal r ₁(t)=x (t)-c ₁(t), as new sequence, step above repeating is until extracting all IMF components with it. and primary signal x this moment (t) finally is decomposed into n IMF and a residual components r _n(t):

$x\left(t\right)=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{c}_i\left(t\right)+r_n\left(t\right).$

3. a kind of ofdm signal bandwidth blind estimating method as claimed in claim 1, it is characterized in that: step (3) is carried out as follows:

3.1) at first determine the population variance of all IMF components, calculate the variance contribution ratio of each IMF component more respectively, be designated as μ _i, 0＜μ _i＜1, i=1,2 ..., n;

3.2) ask threshold value λ=max (μ _i) η, 0＜η＜1;

3.3) if μ _i〉=λ then keeps this IMF component, otherwise removes this IMF component;

3.4) the IMF component reconstruct power spectrum that utilize to keep.

Images