CN1809044A - Circuit and method of peak-to-average ratio suppression in OFDM system - Google Patents

Abstract

This invention relates to digital communication technique field and to the circuit and method to suppress signal peak value power and to lower low peak proportion circuit in the crossing frequency multiple communication system, which comprises the following operations: sampling the input data and for IDFT converting; amplifying the converted data and then for limiting operations; removing band noise of the DFT conversion data after limitation for zooming out by times; outputting data after modulation on data; the said limiting operation amplifying times and the zoon out times are same.

Description

The circuit and the method that suppress peak-to-average ratio in the ofdm system

Technical field

The present invention relates to digital communication technology field, relate in particular to the inhibition signal peak power in a kind of orthogonal FDM communication system, the circuit and the method for reduction peak-to-average ratio.

Background information

OFDM (Orthogonal Frequency Division Multiplexing, OFDM) be the very high modulation system of a kind of channel utilization, has good anti fading performance, can realize the parallel convey of data, extensive use along with digital signal processing technology rapid development and large scale integrated circuit, OFDM is subjected to paying close attention to widely day by day, especially in the high-speed digital communication field.

In ofdm communication system, use has synchronized relation on frequency a plurality of carrier waves come modulation signal, because the envelope value of each carrier wave statistics is independent, increase along with carrier number, the ratio of the peak to average power ratio of stack back signal, be that (Peak-to-Average power ratio, numerical value PAPR) is bigger for peak-to-average ratio.Therefore the dynamic range of modulation signal is quite big, and this has higher linear dynamic range with regard to requiring the power amplifier in the system, with frequency spectrum diffusion and the nonlinear distortion of avoiding transmission signals.Simultaneously also require follow-up D/A converter to have bigger switching bandwith, so just increased system cost and realized difficulty.

At present, the way that is used for reducing PAPR has a lot, and we can roughly be divided into two classes with these technology: a class is before the OFDM multiplexer input code flow suitably to be handled.For example the input data of OFDM are used the way of block encoding, can be reduced the independence of each road signal of multiplexer, thereby reduce the probability of signal peak peak value stack, and then reduce PAPR; Perhaps use the way of partial transmission sequence, the number of signal when superposeing by reducing, thus lower PAPR; Also have a kind of way of shining upon selected, promptly by mapping, a kind of assembled scheme is selected in all possible combination that produces each road code stream then therein, makes the PAPR minimum of the signal after IFFT conversion stack.By above introduction as seen, the disadvantage of these class methods be exactly need be bigger amount of calculation.

Another kind of is after the 0FDM multiplexer signal to be handled.Direct, the most effective way is carried out amplitude limit to analog signal exactly.Yet amplitude limit is a non-linear process, and it will cause disturbing and out-of-band noise in the serious band, thereby reduces the performance of BER and the spectrum efficiency of whole system.

A kind of existing method of after the OFDM multiplexer signal being handled is: Ochiai, H., Imai, H. at civilian Performance analysis of deliberately clipped OFDMsignals (Communications, IEEE Transactionson, Volume:50, Issue:1, Jan.2002 Pages:89-101) a kind of over-sampling limit filtration of speaking of reduces the method for PAPR, its structure as shown in Figure 1, wave tail zero filling in input, carry out IDFT (the Inverse Discrete Fourier Transform inverse discrete Fourier transformer inverse-discrete) conversion (J is an oversample factor) that JN is ordered then, the data of output are sent into and are carried out the amplitude limit operation in the amplitude limiter, directly send into DFT (Discrete Fourier Transform discrete Fourier transform (DFT)) module after the data output from amplitude limiter and carry out JN point DFT conversion, remove out-of-band noise after the reduction of data and become N data and carry out N point IDFT conversion, what the PAPR of dateout was more original improves.

We find that this method that reduces PAPR by the over-sampling limit filtration makes Data bit width More effectively utilizedThe outer interference greatly of the band that amplitude limit produces reduced, but there are the following problems: with 64_QAM (Quadrature Amplitude Modulation quadrature amplitude modulation), A/D is 8 and is example: suppose in the 64_QAM coding, amplitude peak is 127, obtain less PAPR, Amax is to the maximum about 10, and bit wide is not used more effectively, and because the general receiving circuit of the too little employing of amplitude limit value will produce serious error code.

Can adopt custom-designed receiving circuit to solve error code problem? Hangjun Chen, Haimovich, A. be An iterative method to restore the performanceof clipped and filtered OFDM signals in the text.A kind of receiver is proposed in (Communications, 2003.ICC ' 03.IEEE International Conference on, Volume:5,11-15 May 2003Pages:3438-3442 vol.5), as shown in Figure 2, wherein $a=1-e^{{-\mathrm{\γ}}^2}+\frac{\sqrt{\mathrm{\π}}}{2}\mathrm{erfc}\left(\mathrm{\γ}\right),$ We find that this receiver requirement receiving unit need contain the information that sends part, and the requirement and the receiving circuit that can not adapt to multiple emission system have fully also used the structure the same with transtation mission circuit, cause circuit too complicated.

Summary of the invention

The present invention is primarily aimed at the deficiency of above method, and a kind of method for inhibiting peak-to-average ratio of limit filtration is provided, and can improve the PAPR value of signal greatly.

The circuit that suppresses peak-to-average ratio in a kind of ofdm system disclosed by the invention comprises following be linked in sequence, the module of handling the input data: the input data are carried out over-sampling adds 0 information and carries out IDFT conversion (Inverse Discrete Fourier Transform inverse discrete Fourier transformer inverse-discrete) after data over-sampling module; Over-sampling data after the conversion are carried out the amplitude limit module of amplitude limit operation; Data after the amplitude limit are carried out the denoising module that out-of-band noise is removed in DFT conversion (Discrete Fourier Transform discrete Fourier transform (DFT)) afterwards; With the output module that data is carried out dateout after the IDFT conversion; It is characterized in that the function of described amplitude limit module is the data that come from described over-sampling module to be amplified earlier carry out afterwards the amplitude limit operation again; The function of described denoising module is that the data DFT conversion that comes from the amplitude limit module is removed out-of-band noise afterwards simultaneously, also will carry out multiple again and dwindle; The multiple that dwindles in amplification multiple and the described denoising module in the described amplitude limit module is identical.

The method that suppresses peak-to-average ratio in a kind of ofdm system disclosed by the invention comprises the step of following sequential processes input data: first step over-sampling, carry out over-sampling, add 0 and carry out the IDFT conversion after data the input data; The second step amplitude limit carries out the amplitude limit operation to the data that come from first step over-sampling; The 3rd step denoising is carried out the data after the amplitude limit to remove out-of-band noise after the DFT conversion; The 4th step output is carried out dateout after the IDFT conversion to the data of exporting after the 3rd step denoising; It is characterized in that, go on foot in the amplitude limit, carry out the amplitude limit operation after the data that come from described first step over-sampling are amplified earlier again described second; In described the 3rd step denoising, the data DFT conversion that will come from the amplitude limit module is removed out-of-band noise afterwards, also will carry out multiple again and dwindle simultaneously; Amplification multiple in the described second step amplitude limit is identical with the described the 3rd multiple that dwindles that goes on foot in the denoising.

Suppress the circuit and the method for peak-to-average ratio in the ofdm system disclosed by the invention and before the input data are done the amplitude limit operation data are amplified, parameter is n1; With reduction of data, parameter is 1/n1 after through JN point DFT conversion, and average power is increased, and makes data bit width reduce the error rate when more effectively being utilized; Peak value in the amplitude limiter is not the peak value of system, but the value that process emulation is set, the sphere of action of amplitude limit value is component (real part and an imaginary part), chooses equilibrium valve between the error rate and inhibition PAPR effect.Amplify before the data output after N point IDFT, guarantee the maximum utilization of data area.The error rate of the present invention is little, can use general receiving circuit.

Description of drawings

For further specifying technology contents of the present invention, below in conjunction with embodiment and accompanying drawing describes in detail as after, wherein:

Fig. 1 is the theory diagram of over-sampling limit filtration.

Fig. 2 is prior art Hangjun Chen, and Haimovich, A. be An iterativemethod to restore the performance of clipped and filtered OFDMsignals in the text.A kind of receiver that proposes in (Communications, 2003.ICC ' 03.IEEE InternationalConference on, Volume:5,11-15 May 2003 Pages:3438-3442 vol.5).

Fig. 3 is a functional block diagram of the present invention.

Fig. 4 is the simulated effect that carries out according to of the present invention.Ordinate is the probability (CCDF) that peak-to-average ratio surpasses a certain threshold value z, and abscissa is the PAPR value.

Embodiment

Technical scheme disclosed by the invention is as shown in Figure 3, a kind of attainable method for inhibiting peak-to-average ratio based on the over-sampling limit filtration, following feature is arranged: for each converter unit, wave tail zero filling in input obtains JN data, carry out IDFT (the Inverse DiscreteFourier Transform inverse discrete Fourier transformer inverse-discrete) conversion (J is an oversample factor) that JN is ordered then, the data based simulation result of output carries out certain multiple and amplifies, if multiplication factor is n1, send into then and carry out the amplitude limit operation in the amplitude limiter, directly send into DFT (DiscreteFourier Transform discrete Fourier transform (DFT)) module after the data output from amplitude limiter and carry out JN point DFT conversion, and with the data removal out-of-band noise of exporting, be reduced to N data after amplifying (1/n1) multiple again, carry out N point IDFT conversion again, amplify n2 more doubly, the more original PAPR of the data of output improves significantly.

Committed step of the present invention and realization circuit thereof are described in detail as follows:

1. over-sampling module: the input data are carried out over-sampling and carried out JN point IDFT conversion

When signal was sampled according to Nyquist sample rate, the theoretical formula of signal PAPR statistical property was at this moment released in separate distribution between the sampled point easily, has become the theoretical foundation of a lot of PAPR inhibition methods.Result of study is pointed out simultaneously, when discrete signal reverts to continuous signal by D/A, PAPR rise phenomenon can occur, the discrete signal that obtains by the Nyquist sample rate especially, and PAPR gos up the most serious.Therefore, the effect that adopts oversampled signals to carry out the PAPR inhibition will be more obvious.Our method also is to be based on this thought.

We add some 0 information (being over-sampling) to original data flow, can also can concentrate on data and add 0 information afterwards, and carry out the IDFT conversion that JN is ordered between each adjacent valid data.Describe below that 0 information of adding is example after the data to concentrate on.

As shown in Figure 3, use A={A ₀, A ₁, A ₂..., A _N-1The original signal sequence that is used to transmit in expression ofdm system (sub-carrier number is N), wherein Ak is the complex data on the subcarrier k.After over-sampling, modulation and phase shift (phase shift is not included within our method, can not play the effect that reduces PAPR, is the easy analysis for the back here), transmission signals becomes:

Wherein:

$A_k^{\&prime;}=\left\{\begin{array}{cc}{A}_k& k<N\\ 0& k\&GreaterEqual;N,\end{array}\right.$

2. amplitude limit module: carry out the amplitude limit operation after the over-sampling data after the conversion are amplified

As previously described, if directly data being carried out the amplitude limit operation can find to obtain lower PAPR, then the utilance of bit wide is lower, and make that the data after most of conversion are destroyed because amplitude limit value is too small, we use general receiving circuit will produce serious error code, so need carry out certain processing to data.

Our processing method is that data are carried out certain amplification, average power is increased, the raising amplitude limit value that we can be suitable like this, so that obtain to suppress preferably the effect of PAPR, for fear of when hardware circuit is realized, introducing evolution, our amplitude limit value to as if the real part and the imaginary part of output valve.

As shown in Figure 3, we carry out following operation to signal:

Suppose that its n output sample is ${\tilde{s}}_n=\mathrm{Re}\left({\tilde{s}}_n\right)+\mathrm{jIm}\left({\tilde{s}}_n\right),$ Then have:

$\mathrm{Re}\left({\tilde{s}}_n\right)=\left\{\begin{array}{c}\mathrm{Re}\left(s_n^{\&prime;}\right),\mathrm{Re}\left(s_n^{\&prime;}\right)\&le;A_{\max }\\ A_{\max ,}\mathrm{Re}\left(s_n^{\&prime;}\right)>A_{\max }\end{array}\right.,$ $\mathrm{Im}\left({\tilde{s}}_n\right)=\left\{\begin{array}{c}\mathrm{Im}\left(s_n^{\&prime;}\right),\mathrm{Im}\left(s_n^{\&prime;}\right)\&le;A_{\max }\\ A_{\max },\mathrm{Im}\left(s_n^{\&prime;}\right)>A_{\max }\end{array}\right.$

A wherein _MaxMaximum for signal amplitude.

Definition amplitude limit rate $\mathrm{\&gamma;}=\frac{A_{\max }}{\sqrt{P_{\mathrm{in}}}},$ $\sqrt{P_{\mathrm{in}}}=\sqrt{\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}|x\left(n\right){|}^2}$

We find: 1) signal behind the JN point IDFT is amplified n1 doubly, can improve the Pin value of signal, and then reduce amplitude limit rate γ value, thereby effectively reduce the PAPR value.Therefore, consider, wish that the n1 value is the bigger the better from reducing PAPR value angle.2) principle chosen of Amax value is, guarantees that most data can not lost, and help the correct transmission of data like this, therefore wish that the n1 value is the smaller the better through after the amplitude limit operation.Comprehensive above factor, choosing of n1 value is contradiction as can be seen, so in actual applications, we need take the scheme of compromising.

3. denoising module: the data after the amplitude limit are carried out removing out-of-band noise after the JN point DFT conversion, carry out multiple again and dwindle

The limit filtration method of general inhibition PAPR is a non-linear process, and its shortcoming is to introduce serious out-of-band noise, thereby reduces the performance of BER and the spectrum efficiency of whole system.

We have reached the effect that suppresses PAPR through after the over-sampling of front, JN point IDFT conversion, multiple amplification, amplitude limit, the JN point DFT operation, and this step is that the out-of-band noise that amplitude limit is introduced is removed.As shown in Figure 3, shown in the following process of the method for enforcement:

DFT handles through the JN point, and the JN point data sequence that produces distortion is:

${\tilde{A}}^{\&prime;}=\{{\tilde{A}}_0^{\&prime;},{\tilde{A}}_1^{\&prime;},\&CenterDot;\&CenterDot;\&CenterDot;,{\tilde{A}}_{N-1}^{\&prime;},{\tilde{A}}_N^{\&prime;},\&CenterDot;\&CenterDot;\&CenterDot;,{\tilde{A}}_{\mathrm{JN}-1}^{\&prime;}\}$

Cast out the outer interference of band, then the length after the initial data distortion is that the sequence of N is:

${\tilde{A}}^{\&prime;}=\{{\tilde{A}}_0^{\&prime;},{\tilde{A}}_1^{\&prime;},\&CenterDot;\&CenterDot;\&CenterDot;,{\tilde{A}}_{N-1}^{\&prime;}\}$

4. output module: data are carried out OFDM modulation (N point IDFT) and suitably amplified dateout

Through the processing of front, the PAPR value of the data of this moment reduces greatly, and we carry out real OFDM modulation to data, and as shown in Figure 2, this sequence is carried out common OFDM modulation back with N point IDFT and amplified n2 output,

It is in order to guarantee the maximum utilization of data area that last dateout multiply by a coefficient n2.

Below by specific embodiment technical scheme of the present invention is further described.Embodiment is for having used WLAN (wireless local area network) 802.11a protocol system of the present invention, and this system adopts the 64QAM modulation system.Data bit width is 8, oversample factor J=2, and the multiple of scaling is n1=8 after the JN point IFFT conversion, and amplitude limit value is 70, and amplification multiple is n2=4 times after the N point IDFT.In order to observe the result that improves of our PAPR, the curve that we surpass probability (CCDF)-PAPR value of a certain threshold value z with peak-to-average ratio is used as the test effect of our method.Specific implementation method is as follows:

1. over-sampling module: the input data are carried out over-sampling and carried out JN point IDFT conversion

Analyze through us, the factor for over-sampling, J=2 is identical to suppressing the PAPR effect with J=16, but select realization (16 * 64 the IDFT conversion) difficulty of latter's hardware bigger, so we select oversample factor is 2, promptly after adding 64 0 afterwards, each IDFT converter unit (64) sends in 128 IDFT converter units.

2. amplitude limit module: carry out the amplitude limit operation after the over-sampling data after the conversion are amplified

We determine several groups of parameters after setting up model emulation

First group: n1=8 n2=4 Amax=70;

Second group: n1=8 n2=4 Amax=127;

The 3rd group: n1=16 n2=4 Amax=127;

The 4th group: n1=8 n2=4 Amax=40;

The effect of the inhibition PAPR of the 4th group of parameter is best, but we find that its error rate is higher, and irrecoverable.In the comparison of first group and the 3rd group, though first group inhibition PAPR effect is relatively poor than the 3rd group, the 64QAM planisphere effect of its reception is significantly better than the 3rd group, and our selected first group of parameter is our optimal parameter combination at last.Its simulation result as shown in Figure 4.

3. denoising module: the data after the amplitude limit are carried out carrying out the multiple reduction and removing out-of-band noise after the JN point DFT conversion

For the IDFT unit of per 128 data, directly remove 64 points of back, each processing unit becomes 64 data.

4. output: data are carried out OFDM modulation (N point IDFT) and suitably amplified dateout

Per 64 data of removing out-of-band noise are carried out 64 IDFT conversion, amplify 4 times of dateouts afterwards.

The system's peak-to-average ratio that suppresses the PAPR processing is approximately about 11db, is reduced to about 7db through PAPR after the said method, and as shown in Figure 4, be the simulated effect that carries out according to of the present invention.Ordinate is the probability (CCDF) that peak-to-average ratio surpasses a certain threshold value z, and abscissa is the PAPR value.As seen the present invention not only effectively suppresses PAPR, does not also need information specific for receiving circuit, is easy to realize.

Claims (10)

1, the circuit that suppresses peak-to-average ratio in a kind of ofdm system comprises following be linked in sequence, the module of handling the input data: the input data are carried out over-sampling adds 0 information and carries out IDFT conversion (Inverse Discrete Fourier Transform inverse discrete Fourier transformer inverse-discrete) after data over-sampling module; Over-sampling data after the conversion are carried out the amplitude limit module of amplitude limit operation; Data after the amplitude limit are carried out the denoising module that out-of-band noise is removed in DFT conversion (Discrete Fourier Transform discrete Fourier transform (DFT)) afterwards; With the output module that data is carried out dateout after the IDFT conversion; It is characterized in that the function of described amplitude limit module is the data that come from described over-sampling module to be amplified earlier carry out afterwards the amplitude limit operation again; The function of described denoising module is that the data DFT conversion that comes from the amplitude limit module is removed out-of-band noise afterwards simultaneously, also will carry out multiple again and dwindle; The multiple that dwindles in amplification multiple and the described denoising module in the described amplitude limit module is identical.

2, the circuit that suppresses peak-to-average ratio in the ofdm system according to claim 1, it is characterized in that, the multiple that dwindles in amplification multiple and the described denoising module in the described amplitude limit module all equals 8, after described output module carries out the IDFT conversion to data, suitably amplify the back dateout again.

3, suppress the circuit of peak-to-average ratio in the ofdm system according to claim 2, it is characterized in that, dateout after described output module carries out amplifying 4 times again after the IDFT conversion to data.

4, according to the circuit that suppresses peak-to-average ratio in claim 1 and the 3 described ofdm systems, it is characterized in that the amplitude limit computing of described amplitude limit module is: n sampling point supposing described amplitude limit module output is ${\tilde{s}}_n=\mathrm{Re}\left({\tilde{s}}_n\right)+\mathrm{jIm}\left({\tilde{s}}_n\right),$ This sampling point is Sn '=Re (Sn ')+jIm (Sn ') before the amplitude limit computing, then has:

$\mathrm{Re}\left({\tilde{s}}_n\right)=\left\{\begin{array}{c}\mathrm{Re}\left(s_n^{\&prime;}\right),\mathrm{Re}\left(s_n^{\&prime;}\right)\&le;A_{\max }\\ A_{\max },\mathrm{Re}\left(s_n^{\&prime;}\right)>A_{\max }\end{array}\right.$

$\mathrm{Im}\left({\tilde{s}}_n\right)=\left\{\begin{array}{c}\mathrm{Im}\left(s_n^{\&prime;}\right),\mathrm{Im}\left(s_n^{\&prime;}\right)\&le;A_{\max }\\ A_{\max },\mathrm{Im}\left(s_n^{\&prime;}\right)>A_{\mathrm{amx}}\end{array}\right.$

A wherein _MaxMaximum for signal amplitude.

5, suppress the circuit of peak-to-average ratio in the ofdm system according to claim 4, it is characterized in that the maximum of the clipping signal amplitudes of described amplitude limit module equals A _Max70.

6, the method that suppresses peak-to-average ratio in a kind of ofdm system comprises the step of following sequential processes input data: first step over-sampling, carry out over-sampling, add 0 and carry out the IDFT conversion after data the input data; The second step amplitude limit carries out the amplitude limit operation to the data that come from first step over-sampling; The 3rd step denoising is carried out the data after the amplitude limit to remove out-of-band noise after the DFT conversion; The 4th step output is carried out dateout after the IDFT conversion to the data of exporting after the 3rd step denoising; It is characterized in that, go on foot in the amplitude limit, carry out the amplitude limit operation after the data that come from described first step over-sampling are amplified earlier again described second; In described the 3rd step denoising, the data DFT conversion that will come from the amplitude limit module is removed out-of-band noise afterwards, also will carry out multiple again and dwindle simultaneously; Amplification multiple in the described second step amplitude limit is identical with the described the 3rd multiple that dwindles that goes on foot in the denoising.

7, the method that suppresses peak-to-average ratio in the ofdm system according to claim 6, it is characterized in that, amplification multiple in the described second step amplitude limit and the multiple that dwindles in described the 3rd step denoising equal 8, in described the 4th step output data are carried out also will suitably amplifying the back dateout again after the IDFT conversion.

8, suppress the method for peak-to-average ratio in the ofdm system according to claim 7, it is characterized in that, dateout after in described the 4th step output data being carried out amplifying 4 times again after the IDFT conversion.

According to the method that suppresses peak-to-average ratio in claim 6 and the 8 described ofdm systems, it is characterized in that 9, the amplitude limit computing in the described second step amplitude limit is: n sampling point supposing described amplitude limit module output is ${\tilde{s}}_n=\mathrm{Re}\left({\tilde{s}}_n\right)+\mathrm{jIm}\left({\tilde{s}}_n\right),$ This sampling point is Sn '=Re (Sn ')+jIm (Sn ') before the amplitude limit computing, then has:

$\mathrm{Re}\left({\tilde{s}}_n\right)=\left\{\begin{array}{c}\mathrm{Re}\left(s_n^{\&prime;}\right),\mathrm{Re}\left(s_n^{\&prime;}\right)\&le;A_{\max }\\ A_{\max },\mathrm{Re}\left(s_n^{\&prime;}\right)>A_{\max }\end{array}\right.,\mathrm{Im}\left({\tilde{s}}_n\right)=\left\{\begin{array}{c}\mathrm{Im}\left(s_n^{\&prime;}\right),\mathrm{Im}\left(s_n^{\&prime;}\right)\&le;A_{\max }\\ A_{\max },\mathrm{Im}\left(s_n^{\&prime;}\right)>A_{\max }\end{array}\right.$

A wherein _MaxMaximum for signal amplitude.

10, suppress the method for peak-to-average ratio in the ofdm system according to claim 11, it is characterized in that, the maximum A of signal amplitude in the described second step amplitude limit _MaxEqual 70.

Images