CN101018071A - Method for removing the peak power - Google Patents

Abstract

The method for eliminating peak power comprises: (1) after a certain time slot, according to carrier number, rotating phase for training sequences; (2) modulating and spreading frequency, filtering with L-time interpolation to obtain different data, and linear superpositioning all data; and (3) taking M-time interpolation filter, eliminating peak, then taking 1/M-time extract filtering. Wherein, the (L*M)/2 data near the peak power are interpolated by same-length sequences. This invention is fast, and has no nonlinear distortion.

Description

The method of cutting peak power

Technical field

The present invention relates to the communications field, relate in particular to the method for cutting peak power in a kind of TD-SCDMA (TD SDMA) multifrequency point communication system and the communication ends of realization cutting peak power.

Background technology

In the sub-district of TD-SCDMA multifrequency point communication system, owing to need the multifrequency point signal of emission will merge at digital intermediate frequency, so that a shared cover transmitter is launched in follow-up radio-frequency channel, therefore will occur because the peak-to-peak superposition produces stronger signal peak-to-average ratio PAPR (peak toaverage power raio, the ratio of peak power and average power) at the antenna transmitting terminal.Peak power is too big, is easy to make radio-frequency transmissions to the inelastic region, thereby produces stronger ACLR (Adjacent Channel Leakage powerRatio, ACLR, the side channel leakage power ratio of transmitter), and then reduces systematic function.If do not want to cause nonlinear distortion, the power that transmits must be less than the 1dB compression point, this falls very lowly with regard to the average power that requires signal, but in this case, the efficient of power amplifier will reduce, the signal power of radio frequency transmission simultaneously can not reach the DB number that physical layer requires, thereby makes the coverage of base station dwindle, and user's signal power is impaired.And high peak-to-average force ratio can cause the dynamic range of D/A converter to diminish, if use the D/A converter of high exponent number, cost will improve greatly, if use the low D/A converter of exponent number, quantizing noise can strengthen.

Three kinds of methods that reduce peak-to-average force ratio are proposed: slicing method, sequence selection method and phase amplitude converter technique in the prior art.In TD-SCDMA standard multiple frequency point system, the sequence selection method implements complexity, will not adopt usually.The slicing method is that the too high peak value of power is eliminated, and is easy to generate nonlinear distortion.A kind of time domain peak clipping frequency domain filtering method is wherein arranged, and this kind method is at first used and is declared method firmly, will cut firmly above the data of power threshold, uses filter to eliminate then and cuts the nonlinear distortion that causes firmly.But this method nonlinear distortion can not eliminate fully, still can cause very big spectrum leakage.

Though the phase amplitude converter technique does not produce nonlinear distortion, but if feasible rotation each time back peak-to-average force ratio optimum, must carry out a large amount of calculating and need the transmission side information, time-delay is big and handle complexity, transmits side information and also is easy to generate the error code diffusion.At TD-SCDMA standard multiple frequency point system, prior art discloses improved method again on the basis of phse conversion method:

Adopt different midamble sign indicating numbers to be shifted by different frequent points and reduce the probability that peak value occurs.Though this kind method has effectively overcome the power of the midamble sign indicating number problem too high with respect to data division power, the power that eliminates is very limited, and still can not solve the too high problem of data division peak-to-average force ratio in the time slot.

Summary of the invention

The object of the present invention is to provide the method for cutting peak power, to solve the too high technical problem of peak-to-average force ratio that reduces the antenna transmitting terminal in the existing TD-SCDMA multifrequency point communication system.

For addressing the above problem, the present invention discloses a kind of method of cutting peak power, comprising:

(1) each circuit-switched data behind the modulation spread spectrum is carried out L times of interpolation filtering, and frequency conversion becomes the different frequent points data;

(2) all frequency point data are in the enterprising line linearity stack of time domain;

(3) superimposed data is carried out M times of interpolation filtering earlier, carries out peak clipping again, after finish 1/M times of filtering extraction, wherein, L, M are the interpolation multiple.

Optimally, step (1) is also comprising before the L times of interpolation filtering: insert training sequence code on each time slot of each circuit-switched data behind the modulation spread spectrum, the training sequence code of different frequent points carries out different phase place deflection.And the training sequence code of different frequent points carries out different phase place deflection and is specially: the Midamble code data is multiplied by exp (j* (n-1) * π), and j is an imaginary symbols, and n is a carrier index.

Step (3) peak clipping comprises that the one-level peak value detects and eliminates process at least, and wherein each grade peak value detects and eliminates process and may further comprise the steps: a1: detection power surpasses the data symbol of the power threshold at the corresponding levels of preserving in advance, and obtains peak value; A2: interpolation method eliminates peak value.

When peak clipping comprised multistage peak value detection and eliminates process, the power threshold that step a1 is provided with was successively decreased step by step.

Step a1 is specially: b1: be docile and obedient the data cached symbol of preface, and find power to surpass the data symbol of predefined power threshold at the corresponding levels; B2: buffer memory data symbol subsequently, and whether the power of judging the data in buffer symbol less than the power of last data in buffer symbol, if then the power of last data cached symbol is peak value, otherwise, proceed b2; B3: obtain K data symbol before this data symbol of the data symbol that comprises the peak power correspondence, and further K data symbol after the data symbol of buffer memory peak power correspondence.

Step a2 is specially: if the I circuit-switched data that peak power corresponds to is greater than zero, then subtract each other the corresponding position with default peak clipping sequence of the I circuit-switched data of the 2*K of Bao Cuning data symbol, otherwise the corresponding position addition with default peak clipping sequence of the I circuit-switched data of 2*K the data symbol of preserving; If the Q circuit-switched data that peak power corresponds to is greater than zero, then subtract each other the corresponding position with default peak clipping sequence of the Q circuit-switched data of the 2*K of Bao Cuning data symbol, otherwise the corresponding position addition with default peak clipping sequence of the Q circuit-switched data of 2*K the data symbol of preserving.

Optimally, K is (L*M)/2.

A kind of communication ends that realizes cutting peak power comprises:

Some interpolation filters: be used for the circuit-switched data behind the ovennodulation spread spectrum is carried out L times of interpolation filtering;

Some frequency converters: be used for the data after the interpolation are converted to a corresponding frequency point data;

One superimposer: connect all converter units, be used for all frequency point data after the frequency conversion in the enterprising line linearity stack of time domain;

One interpolation despiker: described interpolation despiker comprises interpolation filtering unit, peak clipping unit and filtering extraction unit, wherein,

Interpolation filtering unit: be used for superimposed data is carried out M times of interpolation filtering;

Peak clipping unit: be used for that filtered data are carried out the peak value detection and eliminate with peak value;

Filtering extraction unit: be used for data after the peak clipping cell processing are carried out 1/M times of filtering extraction.

Communication ends also comprises: some phase place deflectors: after inserting training sequence code on each time slot of each circuit-switched data, and the phase place deflection that the training sequence code of different frequent points is correlated with.

Interpolation filter, frequency converter and superimposer are realized by digital up converter DUC.The interpolation filtering unit is that digital up converter DUC realizes that the filtering extraction unit is to be realized by digital down converter DDC.

Described peak clipping unit comprises peak detector, memory cell and adder at least, wherein,

Memory cell: be used for the power threshold of each grade of buffer memory peak value detection and the data symbol after the buffer memory sampling;

Peak detector: connected storage is used to seek peak power;

Adder: connect peak detector and memory cell, be used for the plus and minus calculation of peak signal and peak clipping sequence.

Compared with prior art, the present invention has the following advantages: because data are carried out L interpolation filtering doubly earlier, after carry out M times of interpolation filtering again, promptly a CHIP (chip) has finished L*M times of interpolation, be easy to find peak power, the individual data symbol of K (K is (L*M)/2 preferably) before and after the data symbol at peak power place uses the filter of equal length to carry out the interpolation peak clipping, because the interpolation method peak clipping is simple plus and minus calculation, so processing speed is fast, the frequency spectrum of peak clipping sequence and input signal mate simultaneously, so the peak clipping operation does not increase out-of-band interference, do not cause nonlinear distortion, therefore ACLR does not worsen, and then has improved the efficient that reduces peak-to-average force ratio.

The present invention uses the method for phase place deflection to make that the peak clipping processing of back can be more at business datum power partly owing to inherited physical layer, the power that eliminates Midamble sign indicating number part is with respect to the excessive problem of other business datum part power, and make that the peak value fluctuating of Midamble sign indicating number is approaching, can be more accurate when carrying out channel estimating.

The present invention is provided with multistage PDE (peak value detects and eliminates process), the power threshold that each grade PDE is provided with reduces step by step, each grade PDE carries out the peak clipping operation, after the data symbol process N level PDE of input intermediate frequency, the data that surpass power threshold almost can be eliminated, and reach the requirement of system to PAPR.

Description of drawings

Fig. 1 is the flow chart of the method for cutting peak power of the present invention;

Fig. 2 is an example process schematic diagram of the cutting peak power method that adopts of the present invention;

Fig. 3 is the power schematic diagram of a time slot after the data of six frequencies superpose;

Fig. 4 is the power schematic diagram through a time slot after the data stack of six frequencies of phase place deflection;

Fig. 5 is that a peak value detects and the flow chart that eliminates process;

Fig. 6 is the flow chart that multistage PDE of the present invention (peak value detects and eliminates process) handles;

Fig. 7 is the power contrast figure of input data and peak clipping pulse;

Fig. 8 is input data and the power contrast figure that subtracts each other the back dateout through the peak clipping pulse;

Fig. 9 is the power spectral density contrast figure of peak clipping pulse, input signal;

Figure 10 is to use the power contrast figure before and after phase place rotation-multistage digital interpolation method peak clipping;

Figure 11 is frequency spectrum contrast figure before and after the peak clipping;

Figure 12 is CCDF contrast figure before and after the peak clipping.

Embodiment

Below in conjunction with accompanying drawing, specify the present invention.

See also Fig. 1, it is the flow chart of the method for cutting peak power of the present invention.It may further comprise the steps:

S110: each circuit-switched data behind the modulation spread spectrum is carried out L times of interpolation filtering, is converted to the different frequent points data;

S120: all frequency point data are in the enterprising line linearity stack of time domain;

S130: superimposed data is carried out M times of interpolation filtering earlier, carries out peak clipping again, after finish 1/M times of filtering extraction, wherein, L, M are the interpolation multiple.

Because data are carried out L interpolation filtering doubly earlier, after carry out M times of interpolation filtering again, promptly a CHIP (chip) has finished L*M times of interpolation, can find peak power than being easier to, the individual data symbol of K (K can be preferably (L*M)/2) before and after the data symbol at peak power place uses the filter of equal length to carry out the interpolation peak clipping, because the interpolation method peak clipping is simple plus and minus calculation, so processing speed is fast, the frequency spectrum of peak clipping sequence and input signal mate simultaneously, so the peak clipping operation does not increase out-of-band interference, do not cause nonlinear distortion, so ACLR does not worsen.And, before and after peak clipping, do not produce unnecessary redundant data, do not need the cutting redundant data, the impaired minimizing of initial data.

Below specify the method for cutting peak power of the present invention.See also Fig. 2, it is an example process schematic diagram of the cutting peak power method of the present invention's employing.

After the frequency point data modulation mapping and spectrum-spreading and scrambling of physical layer, speed reaches 1.28MPS.Along with the frequency number of TD-SCDMA multifrequency point communication system support increases, higher peak value can occur with bigger probability when the enterprising line linearity stack of time domain, must produce nonlinear distortion at radio-frequency front-end like this, and then cause the Power leakage of neighboring trace greatly.And, enter analog domain and can't carry out the peak clipping processing according to user's requirement later on, need change (digital-to-analogue conversion) for this reason at D/A and become analog signal to carry out the peak clipping processing before.

Concrete peak clipping processing procedure is:

Carry out step S11 earlier: on each time slot of each circuit-switched data behind the modulation spread spectrum, insert training sequence code, the phase place deflection that the training sequence code of different frequent points is correlated with.

According to existing TD-SCDMA multifrequency point scheme, a plurality of frequencies of supporting for same sub-district have following agreement: scrambler that dominant frequency point is identical with assisting the frequency use and basic midamble (training sequence) sign indicating number, therefore the basic midamble code position of each frequency is a complete matching, and the basic midamble sign indicating number of each frequency all is identical binary sequence vector simultaneously.Will occur owing to the peak-to-peak superposition produces bigger peak power in the basic midamble sign indicating number part of transmitting terminal, be far longer than the data power in the time slot.Fig. 3 is the power schematic diagram of a time slot after the data of six frequencies superpose.As can be seen from Figure 3, the power of stack back midamble sign indicating number is far longer than the power of other data in the time slot.

The power interface of Midamble sign indicating number is bordering on the power of data in the time slot in the middle of if the phase place that the Midamble sign indicating number of different frequent points is correlated with is rotated, and the peak-to-average force ratio of a time slot when peak clipping subsequently will be even lower in this time, and the Midamble sign indicating number is impaired few.

For this reason, only need to carry out different phase place deflection at the Midamble of different frequent points sign indicating number: the Midamble code data is multiplied by exp (j* (n-1) * π), and j is an imaginary symbols, and n is a carrier index, n:1～N _Always, N _AlwaysCarrier wave sum for each sub-district in the TD-SCDMA multi-carrier communications systems.Accordingly, only needing that at each frequency of receiving terminal the Midamble code data is multiplied by exp (j* (n-1) * π) gets final product.

Such as, support the TD-SCDMA communication system of three frequencies, can carry out the deflection of 180 degree at the Midamble of second frequency sign indicating number, and the Midamble sign indicating number of first frequency and the 3rd frequency carries out 0 degree and 360 degree rotations, as there not being deflection, so this kind phase place spinning solution is the most simple, and effect is obvious.

For another example, supporting the TD-SCDMA communication system of six frequencies, is that the Midamble sign indicating number of second frequency, the 4th frequency, the 6th frequency carries out the deflection of 180 degree.

By that analogy, for the TD-SCDMA communication system of multifrequency point, the Midamble code data that is numbered the even number frequency carries out the deflection of 180 degree, and the Midamble code data that is numbered the odd number frequency is not rotated.Certainly also to carry out corresponding processing at receiving terminal.

The method that the Midamble sign indicating number of different frequent points carries out different phase place deflection also has a lot, just no longer gives an example at this.See also Fig. 4, it is the power schematic diagram of a time slot after the data stack of six frequencies of overshoot.Fig. 4 and Fig. 3 are compared as can be known, the power that has eliminated Midamble sign indicating number part is with respect to the excessive problem of other business datum part power, use the method for phase place deflection to make that the peak clipping processing of back can be more at business datum power partly owing to inherited physical layer, and make the peak value of Midamble sign indicating number rise and fall approaching, can be more accurate when carrying out channel estimating, and the Midamble sign indicating number is impaired little, makes channel estimating more accurate.

Consult Fig. 2 again, the peak clipping process is carried out step S12 again: each circuit-switched data after step S11 is handled is carried out filtering after the interpolation, and frequency conversion is the data of corresponding frequency then.

Carry out step S13 then: all frequency point data are superposeed at the enterprising line linearity of time domain.

The physical layer data of TD-SCDMA is earlier through ovennodulation, and spectrum-spreading and scrambling is formed frame data, and transmission rate is 1.28Mps.Carry out L times of interpolation at digital intermediate frequency, make into high-speed data.Such as L is 60, and the speed after the interpolation reaches 76.8Mps.Simultaneously multichannel physical layer data after the interpolation is carried out frequency at frequency domain and move, carry out linear superposition in time domain.So the mathematic(al) representation of multi-frequency point TD-SCDMA signal before entering peak clipping is:

$\underset{-}{x}\left(t\right)=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{\mathrm{\λ}}^{(m,k)}\·{\underset{-}{d}}_n^{(m,k)}{s}_{(n-1)Q+q}^{(m,k)}\·C{r}_0\left(t\right)\·\exp (j{\mathrm{\ω}}_{0}t+m{\mathrm{\ω}}_1).---(4-1)$

n＝1ΛN，q＝1ΛQ

S in the formula _p ^{(m, k)}Be the distinctive spreading code of frequency m: spreading code c ^(k)Can be regarded as the distinctive spreading code of a user and sub-district with the combination of cell scrambling v, different frequencies uses different scramblers:

$s_p^{(m,k)}={\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="A20061000741100191.png,A20061000741100281.png"\; state="\{\{state\}\}">c</figure-callout>}}_{1+\left[(p-1)\mathrm{mod}Q\right]}^{\left(k\right)}\&CenterDot;{\mathrm{<figure-callout\; id="1"\; label="v"\; filenames="A20061000741100191.png,A20061000741100281.png"\; state="\{\{state\}\}">v</figure-callout>}}_{1+\left[(p-1)\mathrm{mod}Q\right]}^m,k=1\mathrm{\&Lambda;K},p=1\mathrm{\&Lambda;NQ}---(4-2)$

Q is the later spreading code length of normalization, and length is 16 for TD-SCDMA, and A is a later symbol weight that code channel takies of chnnel coding modulation, and the length after the normalization is 22, λ ^kBeing the amplitude adjusted factor, is to obtain according to power conversion.Cr ₀(t) be at t moment root raised cosine filter coefficient.ω ₀Be the center frequency point of intermediate frequency, ω ₁Be the shared band bandwidth of each frequency.

Defined peak-to-average power ratio this moment and be the ratio of peak power and its average power in a period of time, according to definition:

$\mathrm{PAPR}=\frac{\underset{0\&le;m\&le;\mathrm{NQ}}{\max }\left\{{\left|x\right[m\left]\right|}^2\right\}}{\mathrm{mean}\left\{{\left|x\right[m\left]\right|}^2\right\}},0\&le;m\&le;(2\mathrm{NQ}+L_m)\&CenterDot;N_{\mathrm{TS}}\&CenterDot;N_{\mathrm{interpolate}}---(4-3)$

L _mThe length that is the midamble data is 144chip, N _TSBe to participate in PAPR to calculate a timeslot number, N _InterpolateBe the factor of carrying out interpolation at intermediate frequency, the TD-SCDMA communication system is carried out L (L=60) times interpolation at intermediate frequency in the present embodiment, so N _Interpolate=L.

From formula 4-1 as can be known, the number of the peak power of data symbol and modulation system, logical subscriber number, frequency all has relation, and along with being on the increase of frequency number and number of users, peak-to-average force ratio will be big more.

Also see also Fig. 2, carry out step S14 at last: superimposed data is carried out M times of interpolation filtering earlier, carries out peak clipping again, after finish 1/M times of filtering extraction.

The intermediate frequency data peak clipping still will be carried out M times of interpolation later before entering the D/A converter, such as M desirable 4.When the data of physical layer are carried out interpolation, the big more D/A conversion analog signal afterwards that just approaches more of interpolation multiple.In order to make the data amplitude that is eliminated more approach to import the analog signal of power amplifier, therefore the present invention to the intermediate frequency data of preparing peak clipping carry out again M (as, M=4) times interpolation, so just be easy to find the actual location of peak power, the real initial data that participates in peak clipping simultaneously can reduce, also with regard to indirect minimizing the damage of initial data, so a chip (chip) data can interpolation L*M before peak clipping (as, 60*4=240) doubly, the fluctuating of front and back symbol power just has correlation.

Not all data all need to carry out peak clipping to be handled, and the present invention is provided with a power threshold, just carries out the peak clipping processing if having above the data of this thresholding.How the data that surpass threshold power carry out peak clipping, and when peak clipping all is to need to consider.Finding suitable peak clipping point also is a key of the present invention.Though, be in no hurry to eliminate, after occurring, the current by the time power peak that surpasses thresholding carries out the peak clipping operation again when the data symbol of sampling surpasses threshold value.Be after the present invention finds data symbol above power threshold earlier, at the data symbol that continues to seek peak power point, after carry out peak clipping and handle.

Peak clipping is handled and is comprised the detection of one-level peak value at least and eliminate process, and promptly comprise: a1: detection power surpasses the data symbol of the power threshold at the corresponding levels of preserving in advance, and obtains peak value; A2: interpolation method eliminates peak value.

Step a1 is specially:

B1: be docile and obedient the data cached symbol of preface, and find power to surpass the data symbol of predefined power threshold at the corresponding levels;

B2: buffer memory data symbol subsequently, and whether the power of judging the data in buffer symbol less than the power of last data in buffer symbol, if then the power of last data cached symbol is peak value, otherwise, proceed b2;

B3: obtain K data symbol before this data symbol of the data symbol that comprises the peak power correspondence, and further K data symbol after the data symbol of buffer memory peak power correspondence.

The value of K can preestablish, because K data symbol before and after the data symbol at peak power place uses the filter of equal length to carry out the interpolation peak clipping, the length of filter is long more, the precision of peak clipping is high more, but its amount of calculation is also big more, and is also big more to needs data in buffer amount simultaneously.The inventor finds that after deliberation the preferable value of K is (L*M)/2.Below be that (L*M)/2 are example just with K, illustrate that peak value detects and eliminates process.See also Fig. 5, it is that a peak value detects and the flow chart that eliminates process.

At first, wait for new data symbol input (step S201), and the data symbol of input is cached to (step S202) in the buffer, judge whether to start peak clipping and handle operation (step S203), if then carry out step S213, otherwise carry out step S204.The present invention can preestablish a peak clipping and handle sign, if peak clipping is handled when being designated " 1 ", judges that having started peak clipping handles operation, does not handle operation otherwise start peak clipping.

Step S204, further judge whether the data symbol number of having preserved in the buffer is (L*M)/2, if, then this data symbol of buffer memory is to buffer, front data in buffer symbol all preceding mobile one (step S205) carries out step S206 again, otherwise, carry out step S201, wait for new data symbol input.

Step S206: further judgement be cached to the data symbol in the buffer at last power whether greater than threshold power, if, carry out step S207, otherwise carry out step S201, waits for the input of new data symbol.

Step S207: whether the data symbol power of judging this input is greater than last data in buffer symbol power, if, then the data symbol power of explanation input is continuing growth, carry out step S201, otherwise, sign buffer this time data in buffer symbol power has reached power peak point (step S208), at this moment, the power of input data symbol has reached peak point and has identified last (L*M)/2 a data symbol of preserving in the startup peak clipping processing operation handlebar buffer this moment and all moved (L*M)/2 (step S210) forward, and sign startup peak clipping processing operation (step S211), and then carry out step S201.

Step S213: handle operation as if having started peak clipping, judge further that then the data number of having preserved reaches (L*M) in buffer, if then carry out step S214, otherwise this grade of buffer memory data symbol (S212) by the time carries out step S201 again.

Step S214: empty and start peak clipping and handle operation mark, use interpolation method to carry out peak clipping to (L*M) of buffer memory individual data symbol then and handle (step S215), export (L*M) the individual data symbol after the peak clipping at last again, buffer empties (step S216).

The peak clipping of step S215 is treated to: if the I circuit-switched data that peak power corresponds to is greater than zero, then subtract each other the corresponding position with default peak clipping sequence of the I circuit-switched data of the 2*K of Bao Cuning data symbol, otherwise the corresponding position addition with default peak clipping sequence of the I circuit-switched data of 2*K the data symbol of preserving; If the Q circuit-switched data that peak power corresponds to is greater than zero, then subtract each other the corresponding position with default peak clipping sequence of the Q circuit-switched data of the 2*K of Bao Cuning data symbol, otherwise the corresponding position addition with default peak clipping sequence of the Q circuit-switched data of 2*K the data symbol of preserving.In above-mentioned example, K gets (L*M)/2.

The method of obtaining peak clipping sequence is a lot, and peak clipping sequence comes down to filter coefficient, the frequency spectrum of filter and the frequency spectrum of input signal coupling.Filter is the carrier wave number according to input signal, the bandwidth that each carrier wave is shared, and the speed of signal transmission and the length of filter coefficient design for foundation.

If eliminate a peak-data, each (L*M)/2 data of front and back all will participate in peak clipping and handle.So not only the data of peak power point are by peak clipping, and contiguous data are equally also eliminated, and weaken gradually but eliminate intensity, and this selection with filter is relevant.The present invention uses the interpolation peak clipping sequence to come cutting peak power, and this method is referred to as the digital interpolation method.Its implication is added and subtracted with the peak clipping impulse ratio mutually in the detected peak-data of time domain exactly, and the digital signal that cans be compared to input has been inserted a peak clipping sequence.

Specifically, if for the I circuit-switched data, if the I circuit-switched data that peak power point corresponds to greater than zero, then corresponding with peak clipping sequence of I road signal data subtracts each other, otherwise addition.

if?I _{max_m}＞0

I＝ I- h

else

I＝ I+ h

end (4-4)

For the Q road also is same judgement

if?Q _{max_m}＞0

Q＝ Q- h

else

Q＝ Q+ h

end (4-5)

Because the length of peak clipping sequence and participation data length equate, so the direct plus and minus calculation in corresponding position just can between the vector, redundant data can not appear later in peak clipping yet, the method of this and time domain peak clipping frequency filtering is completely different, the sort of method is owing to the reason of volume machine has redundant data, when eliminating the redundant data of volume machine, can introduce nonlinear operation, so nonlinear distortion is necessarily arranged.But there is not nonlinear distortion in this kind method, has improved the efficient that reduces peak-to-average force ratio, and the operation of peak clipping simultaneously is simple plus and minus calculation, so simple, processing speed is fast.

But the peak power point data that surpasses detection threshold after the one-level peak clipping can not once just be eliminated under threshold value or the data of other small leaks are not detected, so this patent adopts multistage PDE (peak value detects and eliminates process) to eliminate remaining peak power, progression has determined the effect of peak clipping to a certain extent.Multistage PDE handles as schematically shown in Figure 6, and the power threshold that cascade PDE is provided with is to successively decrease step by step.Because the data of peak value are detected and eliminate at first, the setting of successively decreasing step by step is for guaranteeing that big peak-data is at first detected and eliminating, big more peak-data is not eliminated, to cause big more nonlinear distortion at PA (Power Amplifir), so the power threshold of the first order PDE that is provided with is bigger, exemplary value is (9dB, 10dB), the power threshold of afterbody PDE is less, exemplary value is (6dB, 7dB), the peak-data of the data of small leak and regeneration will be eliminated by the PDE of back one-level like this.By the time after data dealt from afterbody PDE, the data of big peak value will be eliminated fully, and the small leak data will occur with very low probability, will can not impact system.

The invention also discloses a kind of communication ends that realizes cutting peak power, comprising:

Some phase place deflectors: insert training sequence code on each time slot of each road frequency point data, the training sequence code of each frequency carries out different phase place deflection

Some interpolation filters: be used for a circuit-switched data is carried out L times of interpolation filtering;

Some frequency converters: each converter unit connects an interpolation unit, is used for the data after the interpolation are converted to a corresponding frequency point data;

One superimposer: connect all converter units, be used for all frequency point data after the frequency conversion in the enterprising line linearity stack of time domain;

One interpolation despiker: described interpolation despiker comprises interpolation filtering unit, peak clipping unit and filtering extraction unit, wherein,

Interpolation filtering unit: be used for superimposed data is carried out M times of interpolation filtering;

Peak clipping unit: be used for that filtered data are carried out the peak value detection and eliminate with peak value;

Filtering extraction unit: be used for data after the peak clipping cell processing are carried out 1/M times of filtering extraction.

Described peak clipping unit comprises peak detector at least, adder and memory cell, wherein,

Memory cell is used for each grade of buffer memory peak value and detects and eliminate the power threshold of process and the data symbol after the buffer memory sampling;

Peak detector: connected storage is used to seek peak power;

Adder: connect peak detector and memory cell, be used for the plus and minus calculation of peak signal and peak clipping sequence.

Interpolation filter, frequency converter and superimposer are realized by digital up converter DUC (Digital Up-Converter).The interpolation filtering unit is that digital up converter DUC realizes that the filtering extraction unit is to be realized by digital down converter DDC (Digital Down-Converter).

Next, use graphic-arts technique to illustrate the effect of digital interpolation method.Physical layer data is after carrying out L*M (60*4=240) times interpolation and filtering, carry out the peak value detection and eliminate (PDE) processing, obtain the peak point of data, as shown in Figure 7, the data of each (L*M)/2 before and after the while buffer memory, the length of the interpolation filter coefficient that sets in advance also is (L*M) (240), after detecting peak-data, this two columns is according to carrying out simple additive operation in time domain, just obtain the figure of anticipating shown in Fig. 8, the data of peak power point are eliminated, and the trend of data power variation simultaneously still maintains the original state.

The digital interpolation method is subtracted each other at detected peak-data symbol of time domain and peak clipping pulse, because being the frequency domain with data symbol, the frequency domain of peak clipping pulse of definition is complementary, so when time domain is carried out plus and minus calculation, can not cause the spectrum leakage of signal at frequency domain, owing to also have filtering extraction after the peak clipping, so sometimes also can improve the performance of ACLR, as Fig. 9 signal is the power spectral density contrast figure of peak clipping pulse, input signal.

TD-SCDMA has training sequence in the data of each time slot of receiving terminal on the other hand, and estimate channel by training sequence (Midamble sign indicating number), because before digital despicking, the phase place of MIDAMBLE sign indicating number is rotated, reduced after frequency stack MIDAMBLE partly peak power and make this part data randomization of Midamble sign indicating number, more approaching with the fluctuating of business datum peak value.Carry out peak clipping behind Midamble sign indicating number and time slot of business datum composition handles together.Because peak clipping is handled training sequence has been carried out same operation, so just comprised the influence of peak clipping in the channel impulse response that estimates by training sequence.

Channel impulse response and spreading code are formed compound spreading code, use compound spreading code to form sytem matrix A, just can better detect initial data when the using system matrix A is carried out joint-detection.This method with the frame structure of TD-SCDMA and channel estimating is relevant.Figure 10 is to use the power contrast figure before and after phase place rotation-multistage digital interpolation method peak clipping.

From figure, use the power of MIDAMBLE sign indicating number part significantly not excessive as can be seen, so use multistage digital interpolation method just can reach good like this effect this moment on this basis with respect to time slot power.Peak power is all eliminated under the power threshold of user's appointment.

Figure 11 as can be seen from Figure 11, does not cause the leakage of ACLR for frequency spectrum contrast figure before and after the peak clipping on frequency domain before and after the peak clipping.(Complementary cumulativedistribution function: contrast figure contrary integral power probability-distribution function), the peak power on the figure roughly is reduced to 7db by 9db to Figure 12, has eliminated the peak value of 2db for CCDF before and after the peak clipping.

More than disclosed only be several specific embodiment of the present invention, but the present invention is not limited thereto, any those skilled in the art can think variation, all should fall within the scope of the invention.

Claims (14)

1, a kind of method of cutting peak power is characterized in that, comprising:

(1) each circuit-switched data behind the modulation spread spectrum is carried out L times of interpolation filtering, and frequency conversion becomes the different frequent points data;

(2) all frequency point data are in the enterprising line linearity stack of time domain;

(3) superimposed data is carried out M times of interpolation filtering earlier, carries out peak clipping again, after finish 1/M times of filtering extraction, wherein, L, M are the interpolation multiple.

2, the method for cutting peak power as claimed in claim 1 is characterized in that, step (1) also comprised before L times of interpolation filtering:

Insert training sequence code on each time slot of each circuit-switched data behind the modulation spread spectrum, the training sequence code of different frequent points carries out different phase place deflection.

3, the method for cutting peak power as claimed in claim 2 is characterized in that, the training sequence code of different frequent points carries out different phase place deflection and is specially: the Midamble code data is multiplied by exp (j* (n-1) * π), and j is an imaginary symbols, and n is a carrier index.

4, the method for cutting peak power as claimed in claim 1 or 2 is characterized in that, step (3) peak clipping comprises that the one-level peak value detects and eliminates process at least, and wherein each grade peak value detects and eliminates process and may further comprise the steps:

A1: detection power surpasses the data symbol of the power threshold at the corresponding levels of preserving in advance, and obtains peak value;

A2: interpolation method eliminates peak value.

5, the method for cutting peak power as claimed in claim 4 is characterized in that, when peak clipping comprised multistage peak value detection and eliminates process, the power threshold that step a1 is provided with was successively decreased step by step.

6, as the method for claim 4 or 5 described cutting peak powers, it is characterized in that step a1 is specially:

B1: be docile and obedient the data cached symbol of preface, and find power to surpass the data symbol of predefined power threshold at the corresponding levels;

B2: buffer memory data symbol subsequently, and whether the power of judging the data in buffer symbol less than the power of last data in buffer symbol, if then the power of last data cached symbol is peak value, otherwise, proceed b2;

B3: obtain K data symbol before this data symbol of the data symbol that comprises the peak power correspondence, and further K data symbol after the data symbol of buffer memory peak power correspondence.

7, the method for cutting peak power as claimed in claim 6 is characterized in that, step a2 is specially:

If the I circuit-switched data that peak power corresponds to is greater than zero, then subtract each other the corresponding position with default peak clipping sequence of the I circuit-switched data of the 2*K of Bao Cuning data symbol, otherwise the corresponding position addition with default peak clipping sequence of the I circuit-switched data of 2*K the data symbol of preserving;

If the Q circuit-switched data that peak power corresponds to is greater than zero, then subtract each other the corresponding position with default peak clipping sequence of the Q circuit-switched data of the 2*K of Bao Cuning data symbol, otherwise the corresponding position addition with default peak clipping sequence of the Q circuit-switched data of 2*K the data symbol of preserving.

8, the method for cutting peak power as claimed in claim 6 is characterized in that, K is (L*M)/2.

9, the method for cutting peak power as claimed in claim 7 is characterized in that, K is (L*M)/2.

10, a kind of communication ends that realizes cutting peak power is characterized in that, comprising:

Some interpolation filters: be used for the circuit-switched data behind the ovennodulation spread spectrum is carried out L times of interpolation filtering;

Some frequency converters: be used for the data after the interpolation are converted to a corresponding frequency point data;

One superimposer: connect all converter units, be used for all frequency point data after the frequency conversion in the enterprising line linearity stack of time domain;

One interpolation despiker: described interpolation despiker comprises interpolation filtering unit, peak clipping unit and filtering extraction unit, wherein,

Interpolation filtering unit: be used for superimposed data is carried out M times of interpolation filtering;

Peak clipping unit: be used for that filtered data are carried out the peak value detection and eliminate with peak value;

Filtering extraction unit: be used for data after the peak clipping cell processing are carried out 1/M times of filtering extraction.

11, the communication terminal of realization cutting peak power as claimed in claim 10 is characterized in that, also comprises:

Some phase place deflectors: after inserting training sequence code on each time slot of each circuit-switched data, the phase place deflection that the training sequence code of different frequent points is correlated with.

As the communication terminal of claim 10 or 11 described realization cutting peak powers, it is characterized in that 12, interpolation filter, frequency converter and superimposer are realized by digital up converter DUC.

As the communication terminal of claim 10 or 11 described realization cutting peak powers, it is characterized in that 13, the interpolation filtering unit is that digital up converter DUC realizes that the filtering extraction unit is to be realized by digital down converter DDC.

As the communication terminal of claim 10 or 11 described realization cutting peak powers, it is characterized in that 14, described peak clipping unit comprises peak detector, memory cell and adder at least, wherein,

Memory cell: be used for the power threshold of each grade of buffer memory peak value detection and the data symbol after the buffer memory sampling;

Peak detector; Connected storage is used to seek peak power;

Adder: connect peak detector and memory cell, be used for the plus and minus calculation of peak signal and peak clipping sequence.

Images