CN101299600B - Design method for digital filter based on OFDM system - Google Patents

Abstract

The invention provides a design method of a digital filter based on the OFDM system, including determining the resolvable interpolation factor through the frequency spectrum continuation of the pass band; determining the combining filter cascaded with the first order filter and a known second order filter, selecting the amplitude frequency characteristic of the interpolation factor combining filter of the first order filter, computing the amplitude frequency characteristic of the first order filter according to the amplitude frequency characteristic of the second order filter and the combining filter; setting the weighted error of the first order filter through extending the frequency band of the first order filter; computing the filter coefficient according to the computing the amplitude frequency and the weighted error of the first order filter, and regarding the combining filter as the digital filter of the OFDM system. The combining filter based on the OFDM system can be designed by the design method of the filter in the invention, furthermore improving the jitter of the pass band ripple, optimizing the EVM of the system and providing the performance of the system.

Description

A kind of design digital filter based on ofdm system

Technical field

The present invention relates to the Design of Filter technical field, relate in particular to a kind of design digital filter based on ofdm system.

Background technology

In the back Development Technology Scheme Choice of 3G (3G (Third Generation) Moblie technology), transmission rate, efficiency of transmission that the support of requirement system is higher, possess better locomotivity and covering power, in wide-band mobile communication, channel can suffer serious multipath fading, and existing mobile communication system is difficult to satisfy the demand of future mobile communications.And OFDM (Orthogonal Frequency Division Multiplexing, OFDM) technology can overcome the problem that the intersymbol interference of single-carrier system when supporting high speed data transfers increases, and the spectrum efficiency height is arranged, hardware is implemented advantages such as simple, therefore in the scheme of E-UTRA, determined (Frequency Division Multiple Access with OFDM/FDMA, frequency division inserts) be the basic transmission technology and the multiple access technology of core, promptly adopt the OFDM technology at down direction, up direction adopts SC-FDMA (single carrier frequency is tapped into) technology.

Wireless access technology based on OFDM comprises FDD (Frequency Division Duplex, Frequency Division Duplexing (FDD)) pattern and TDD (Time Division Duplex, time division duplex) pattern.The shared same frequency range of the up-downgoing of TDD so be that technology combines that more reliable technique support is provided for smart antenna and MIMO (Multiple Input Multiple Output, many antennas import many antennas and export).Under tdd mode, adopt common pilot symbols or special-purpose frequency pilot sign to realize the channel inequality of link, the pilot tone of the quadrature that different antennae adopts can be used for supporting multi-antenna technology, as MIMO.So TD-SCDMA (Time Division Synchronous CDMA, Time Division-Synchronous Code Division Multiple Access) follow-up developments technology adopts the OFDM-TDD pattern, for guaranteeing compatibility and evolution to existing TD-SCDMA system, its frame structure and bandwidth demand and present TD-SCDMA LCR (Low Chip Rate, low spreading rate) are consistent.

TD-SCDMA system physical layer speed is very low now is 1.28MHz, and intermediate frequency speed is 76.8MHz, the speed multiple can reach 60 times, 60 times of interpolations have also been finished simultaneously, so high interpolation factor makes Design of Digital Filter that very big flexibility be arranged, therefore can adopt various junction filter designs, for example can use CIC (Cascaded integrator-comb, comb filter), HB (Half-Band, half-band filter), PFIR (Program Finite Impulse Response, programmable finite impulse response filter) Combination Design is finished the design of intermediate-frequency filter.But OFDM_TDD system physical layer speed has reached 30.72MHz, and simultaneously intermediate frequency speed is subjected to the influence of existing device, generally can not surpass 150MHz, so the speed multiple equals 150/30.72 less than 5 times, promptly maximum can reach 4 times of speed; And the transmitted bandwidth BW=20MHz of the physical layer signal of agreement regulation, 1200 subcarriers are so effective bandwidth is 18MHz, then the intermediate frequency transmission rate is f to the maximum _s=30.72 * 4=122.88MHz, according to formula:

$\frac{\mathrm{BW}}{f_s}=\frac{1+\mathrm{\α}}{L}---(1-1)$

Obtain the interpolation factor L=7 of filter, rolloff-factor α=0.14.The passband of the filter of designing even so but because interpolation factor L=7 then can't resolve into multiple filter, can only be realized with single-section filter just in time in the agreement predetermined band.

An important indicator of OFDM-TDD systematic function quality is EVM (error vectormagnitude, the error vector amplitude), if the intermediate frequency digital filter design is bad, will make signal pass through after the intermediate frequency interpolation filtering, the EVM variation of signal, and then cause the EVM index of whole system to worsen.For intermediate-frequency filter,, can influence the subcarrier on other physical carrier if Design of Filter is bad; And if the corrugated design of passband is bad, can influence all subcarriers of this frequency range.Because in time domain, an OFDM symbol is a rectangle, so present the shape of SINC function at frequency domain, ofdm system satisfies Nyquist does not have the intersymbol interference criterion, but the symbol moulding of this moment is not to carry out pulse-shaping in time domain as common system, but realize at frequency domain.Therefore, because the duality relation of time-frequency, the intersymbol interference (ISI) in the system has become and has disturbed (ICI) between the subcarrier in the ofdm system usually, promptly when the signal of OFDM receiver demodulation subcarrier 1, can introduce the interference of subcarrier 2 to it, and vice versa.This mainly is owing to no longer be integral multiple in the cycle of FFT two subcarriers in the time of integration, thereby can not guarantee due to the orthogonality.Simultaneously, because the shake of passband edge also can make the orthogonal property of subcarrier of band edge be damaged, the consequence that the frequency error of a few subcarrier causes is that these subcarrier signals energy will be distributed in adjacent a plurality of carrier waves, to cause the more orthogonality forfeiture of multi-subcarrier, thereby also can introduce ICI, cause the decline of systematic function.

In order to reduce ICI, the OFDM symbol can send the cyclic extensions signal in guard time, be called Cyclic Prefix (CP).Cyclic Prefix is the signal of OFDM symbol afterbody to be moved head constitute, and can guarantee that like this ofdm signal of time delay always has the integral multiple cycle in the FFT integration period.Therefore as long as multipath delay less than guard time, just can not cause inter-carrier interference.Though can overcome the multipath interference yet increase CP, but can't eliminate the passband fluctuation that digital filter brings, be subjected to existing device ADC simultaneously, the influence of DAC, intermediate frequency speed can not be too high and physical layer rate is very high, when making designing filter interpolation factor limited, can't decompose, thereby cause in ofdm system, can't designing multiple filter.

Summary of the invention

In view of this, the present invention proposes a kind of design digital filter based on ofdm system, has solved the problem that can't design multiple filter in the prior art based on ofdm system.

For this reason, technical scheme provided by the invention is as follows:

A kind of design digital filter based on ofdm system is characterized in that, this method comprises:

Obtain decomposable interpolation factor by frequency spectrum continuation with passband;

Determine junction filter by decomposable interpolation factor, with the junction filter of first order filter and second level filter cascade digital filter as ofdm system by first order filter and second level filter cascade;

With the interpolation factor of described first order filter interpolation factor the amplitude-frequency characteristic of described junction filter is calculated as described junction filter;

Calculate the amplitude-frequency characteristic of first order filter according to the amplitude-frequency characteristic of described second level filter and junction filter;

The transition band of continuation first order filter, and the weighted error of first order filter is set according to the continuation amount of transition band;

Amplitude-frequency characteristic and weighted error according to described first order filter are calculated its filter coefficient.

Preferably, it is poor that transmitted bandwidth and effective bandwidth are got, and this difference is the maximum continuation amount of described passband;

With the passband after the continuation is new passband, calculates interpolation factor.

Preferably, the continuation process of transition band is: the passband of first order filter is put forward predetermined band and stopband delayed one section allocated frequency band the last period.

Preferably, with transition band by (1-α)/L _PFIR＜f≤(1+ α)/L _PFIRContinuation is (1-α-aheadpass)/L _PFIR＜f≤(1+ α+delaystop)/L _PFIR, wherein, aheadpass is the frequency band that shifts to an earlier date in the passband, and delaystop is the frequency band that stopband is delayed, and α is a rolloff-factor, L _PFIRInterpolation factor for first order filter.

Preferably, the weighted error after the transition band continuation is: the passband weighted error is δ _pThe transition band weighted error is δ _αThe stopband weighted error is δ _s

Preferably, this method also comprises: the frequency band shake that adds the known second level filter when weighted error of described first order filter passband is set.

Preferably, the passband weighted error after the frequency band shake of adding second level filter is H _CIC(f) δ _p, H wherein _CIC(f) be the ripple shake of CIC.

Preferably, this method also comprises: the redundant digit that convolution causes is carried out asymmetric cut position, and the few intercepting of a part before the time slot, the time slot rear section intercepts more.

Preferably, the signal length of many interceptings is a 1-50 sampled point.

Preferably, described first order filter is the PFIR filter.

Preferably, described second level filter is a cic filter.

As can be seen, adopt the design digital filter the present invention is based on ofdm system, thereby can design multiple filter by passband being carried out the interpolation factor that continuation obtains decomposing; And when the first order filter of design in the multiple filter, considered the frequency band shake of second level filter, thereby the band attenuation that brings can compensate the second level filter filtering time; Simultaneously the transition band of the first order filter in the multiple filter is carried out certain continuation, effectively improved the influence of the ripple shake of transition band the passband ripple shake; And each step is adjusted the shake that has also well improved passband ripple designing on the basis of multiple filter, improved the EVM of system, well improved the performance of system.

Description of drawings

Fig. 1 is the FB(flow block) that the present invention is based on the design digital filter of ofdm system.

Embodiment

Basic thought of the present invention is to provide a kind of design digital filter based on ofdm system.This method proposes the passband continuation, it is the frequency spectrum that passband captures a part of transition band place, make that the interpolation factor of filter can be adjusted (L can decompose) as required, thereby can design junction filter when the filter of design OFDM_TDD system.The junction filter that designs like this can be so that the ripple of passband be more smooth, and then improves the EVM of system, and the performance of system is improved.Simultaneously, the transition band that is cascaded into the single-section filter of this multiple filter is carried out certain continuation, some arrives stopband the frequency spectrum of transition band frequency spectrum, other frequency bands have been captured with regard to some transition band like this, such way seems some waste of frequency spectrum resource, can the quality of other band signal not impacted but do so in fact, this be because ofdm signal at physical layer spectral shaping just; Because the frequency spectrum moulding of base band, so if the expansion of intermediate-frequency filter frequency spectrum is limited at this moment, just can not cause deterioration to the frequency spectrum of other channel, again because transition band does not cause interference to the signal of this frequency range, so EVM can keep within the system requirements, do not appear at as much as possible in the passband can effectively raising EVM quality for transition band for this reason, and then improves systematic function.

As shown in Figure 1, technical solution of the present invention is to realize ofdm system Filter Design method by the following method:

Step 101: obtain decomposable interpolation factor by frequency spectrum continuation with passband;

Step 102: determine junction filter by decomposable interpolation factor, with the junction filter of first order filter and second level filter cascade digital filter as ofdm system by first order filter and second level filter cascade;

Step 103: the amplitude-frequency characteristic of described junction filter is calculated with the interpolation factor of described first order filter interpolation factor as described junction filter;

Step 104: the amplitude-frequency characteristic of calculating first order filter according to the amplitude-frequency characteristic of described second level filter and junction filter;

Step 105: the transition band of continuation first order filter, and the weighted error of first order filter is set according to the continuation amount of transition band;

Step 106: amplitude-frequency characteristic and weighted error according to described first order filter are calculated its filter coefficient.

Describe technology of the present invention in detail below in conjunction with specific embodiment and realize details:

According to LTE agreement 3GPP TR25.814V1.2.1 (2006-2), the transmitted bandwidth of physical layer signal is BW=20MHz, and 1200 subcarrier passband effective bandwidths are EBW=18MHz, and the intermediate frequency transmission rate is f _s=30.72 * 4=122.88MHz is so the transmitted bandwidth of signal is the 20/122.88=0.162 of whole transmission rate, promptly $\frac{\mathrm{BW}}{f_s}=\frac{1+\mathrm{\α}}{L}=0.162,$ And because rolloff-factor α is generally about 0.1, so interpolation factor should get 1/6 or 1/7; But learn by analysis, ofdm signal is at physical layer spectral shaping just, if filter passband can have certain continuation, it is the frequency spectrum that passband captures a part of transition band place, transition band is captured the frequency band of a part of stopband correspondence, if the frequency spectrum continuation within the specific limits, the expansion of intermediate-frequency filter frequency spectrum can not cause deterioration to the frequency spectrum of other channel so, obtain the interpolation factor L=6 of filter this moment according to formula, rolloff-factor α=0.07, then when interpolation factor L=6, just can design junction filter easily, first order filter interpolation factor L1=3, the interpolation factor L2=2 of second level filter, though this moment, the filter passband that designs was less times greater than agreement regulation system bandwidth (because (1+0.07)/6〉(1+0.14)/7), but the passband ripple of the filter that designs like this is far smaller than the filter ripple that designs in the prior art, and this moment, the transition band of filter did not enter in the effective bandwidth of adjacent frequency band, and the roll-off factor α that aforementioned calculation is come out=0.07 interpolation factor L=6 calculates according to this formula: $\frac{\mathrm{BW}+(\mathrm{BW}-\mathrm{EBW})}{f_s}=\frac{1+\mathrm{\α}}{L},$ Because (1+0.07)/6＜(22+2)/122.88, so the signal of this frequency range does not cause negative effect to the effective subcarrier in the adjacent frequency band, total folder says that the method for calculating interpolation factor and rolloff-factor is as follows: $\frac{\mathrm{BW}+(\mathrm{BW}-\mathrm{EBW})}{f_s}>\frac{1+\mathrm{\α}}{L}>\frac{\mathrm{BW}}{f_s},$ The continuation amount maximum of passband is no more than (BW-EBW).

If stipulate designing filter according to agreement fully in the prior art, only system bandwidth ability protocol compliant is stipulated when interpolation factor L=7, but can't decompose because of interpolation factor this moment, can only design single-section filter, and the passband ripple of well-known single-section filter is shaken obviously greater than multiple filter, thereby makes systematic function can not get guaranteeing.

The baseband signal of physical layer and filter carry out convolution in time domain, if the result of convolution is more little from frequency domain response rolloff-factor α, attenuation outside a channel is also more little, and ripple is also big more simultaneously.This explanation transition band is precipitous more, and ripple and attenuation outside a channel performance all can be poor more in the band.And the embodiment of the invention two method by adopting the multiple-stage filtering cascade to handle makes that ripple satisfies system requirements in the attenuation outside a channel that can also guarantee filter under the certain situation of α and the band.Junction filter in the present embodiment is interpolation or the extraction of making the big factor with cic filter, use PFIR filter compensation CIC to in the decay that causes.The CIC amplitude-frequency characteristic is a pectination:

$\left|H\left(e^{\mathrm{j\ω}}\right)\right|=\left|H_1\left(e^{\mathrm{j\ω}}\right)\right|\·\left|H_2\left(e^{\mathrm{j\ω}}\right)\right|=\frac{\sin \left(\frac{\mathrm{\ω}\·L}{2}\right)}{\sin \left(\frac{\mathrm{\ω}}{2}\right)}=\frac{L\·\sin \left(\frac{\mathrm{\ω}\·L}{2}\right)/\frac{\mathrm{\ω}\·L}{2}}{\sin \left(\frac{\mathrm{\ω}}{2}\right)/\frac{\mathrm{\ω}}{2}}=L\·\sin c\left(\frac{\mathrm{\ω}\·L}{2}\right)\·\sin c{\left(\frac{\mathrm{\ω}}{2}\right)}^{-1}---(1-2)$

It is very poor that but the secondary lobe of single-stage cic filter suppresses, in order to improve inhibition ability to secondary lobe, improve stopband characteristic, can be with the cascade of single-stage cic filter, form cascade integral comb filter (CascadedIntegrator Comb), its frequency response is shown below:

$\left|H\left(e^{\mathrm{jw}}\right)\right|={\left[\frac{\sin c(\mathrm{\ωL}/2)}{\sin c(\mathrm{\ω}/2)}\right]}^Q---(1-3)$

Here, Q is the progression of cascade.The cic filter coefficient obtains according to the mode of launching below:

CIC _Q(z)＝(1+z ^-1+z ^-2+…z ^-(L-1)) ^Q(1-4)

The filter coefficient that obtains can be kept among the RAM, finishes filter function as common FIR filter with convolution algorithm.Though cic filter has low-pass characteristic, also uneven in the frequency band, but certain roll-offing arranged, thus make frequency spectrum produce distortion.The wide more distortion that then produces of the bandwidth of signal is serious more, therefore need compensate with the PFIR filter.

The present embodiment junction filter is set to the RRC filter, and its amplitude-frequency characteristic is as follows:

$H_{\mathrm{RRC}}\left(f\right)=\left\{\begin{array}{cc}1& 0\&le;f\&le;(1-\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\\ \mathrm{sqrt}\left\{\frac{1}{2}[1+\cos \left(\frac{\mathrm{\&pi;}\&CenterDot;(f\&CenterDot;L_{\mathrm{PFIR}}-(1-\mathrm{\&alpha;}))}{2\mathrm{\&alpha;}}\right)]\right\}& (1-\mathrm{\&alpha;})/L_{\mathrm{PFIR}}<f\&le;(1+\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\\ \mathrm{eps}& f>(1+\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\end{array}\right.---(1-5)$

The later Filter Structures interpolation factor of known combination is L _RRC(L _RRC=L _CIC* L _PFIR), but because CIC has designed and has finished, and the amplitude-frequency characteristic in strict accordance with junction filter during design PFIR filter coefficient designs, and the PFIR filter of design is in order to compensate the decay of CIC in passband, so when the later frequency response of composite design, only need consider the characteristic of PFIR, so present embodiment interpolation factor when calculating the frequency response of RRC junction filter is still selected L _PFIRIf this moment, the frequency response of cic filter was H _CIC(z), the frequency response H of then programmable compensating filter in passband _PFIR(z) be:

$H_{\mathrm{PFIR}}\left(z\right)=\frac{H_{\mathrm{RRC}}\left(z\right)}{H_{\mathrm{CIC}}\left(z\right)}---(1-6)$

H wherein _RRC(z) be the frequency response of the junction filter of system requirements, can design the amplitude-frequency characteristic of first order filter PFIR filter thus.

When design PFIR filter, always wish to obtain optimum efficiency, the frequency response that said here the best is meant filter in the interested frequency range and the maximum approximate error minimum between the frequency response values of ideal filter, implication is as follows:

δ _minimax＝minimize{maximum(e(ω)|ω∈[0，ω _s/2)}(1-7)

The error signal e of frequency domain measurement (ω) is defined as follows:

e(ω)＝W(ω)|H _d(e ^jω)-H(e ^jω)|(1-8)

Wherein W (ω) 〉=0 is weighted error (error weight), and this weighted error is a key of design PFIR filter coefficient.Carry out the design of different errors during the mistake in computation weighted factor according to different frequency ranges.PFIR will make up with CIC in embodiments of the present invention, so when the passband ripple of design PFIR, considered the frequency response of CIC in passband simultaneously, the overall passband ripple of filter will improve so, this is the reason that has overcome the ripple shake of CIC owing to design PFIR filter, therefore present embodiment is when being provided with the passband weighted error of PFIR filter, the frequency band shake that has added CIC, can not consider the influence that CIC causes PFIR for stopband and transition band weighted error, decay increases because the target of stopband is requirement, do not consider for the shake that decay is inner, so weighted error is set to:

$\mathrm{weight}\left(f\right)=\left\{\begin{array}{cc}{H}_{\mathrm{CIC}}\left(f\right)\&CenterDot;{\mathrm{\&delta;}}_p& 0\&le;f\&le;(1-\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_a& (1-\mathrm{\&alpha;})/L_{\mathrm{PFIR}}<f\&le;(1+\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_s& f>(1+\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\end{array}\right.---(1-9)$

The shake of the filter ripple so designed obtains improvement to a certain degree, and the shake strictness of ripple the vibration of ripple such as presents, and this means that for OFDM is this EVM can have greatly improved concerning the system of frequency requirement strict synchronism.

And simultaneously when designing filter, also want the minimum value of estimation filter exponent number N: normalization passband cut-off angular frequency ω _p, normalization stopband cut-off angular frequency ω _s, passband weighted error δ _p, stopband weighted error δ _sIf the transition band of filter is precipitous more, be exactly that the frequency band that occupies of transition band is narrow more, the stopband attenuation of the filter that designs so is more little, and the ripple shake is big more, as formula:

$N\&ap;\frac{-20{\log }_{10}\left(\sqrt{{\mathrm{\&delta;}}_p{\mathrm{\&delta;}}_s}\right)-13}{14.6({\mathrm{\&omega;}}_s-{\mathrm{\&omega;}}_p)/2\mathrm{\&pi;}}---(1-10)$

PFIR filter order N and transition band are inversely proportional to as can be seen from the above equation, that is to say that filter coefficient must be grown if transition band is wanted to do precipitously.If so the stopband decay is big more, passband ripple is more little, needed filter order is also long more.Therefore, the embodiment of the invention is when the weighted error of design PFIR filter, frequency band to transition band also limits: promptly passband shifts to an earlier date a bit of frequency band and stopband postpones a bit of frequency band in design PFIR filter weight error, and then the frequency band of transition band is: (1-α-aheadpass)/L _PFIR＜f≤(1+ α+delaystop)/L _PFIR, wherein aheadpass is the frequency band that shifts to an earlier date in the passband, delaystop is the frequency band that stopband postpones.So the frequency band of design transition band makes transition band widen to some extent, has so just saved the filter order of PFIR, and additional filter coefficient just can effectively increase the stopband of filter and reduce the interior ripple of passband.

Simultaneously, in conjunction with top design, weighted error can be provided with as follows at each frequency band:

$\mathrm{weight}\left(f\right)=\left\{\begin{array}{cc}{H}_{\mathrm{CIC}}\left(f\right)\&CenterDot;{\mathrm{\&delta;}}_p& 0\&le;f\&le;(1-\mathrm{\&alpha;}-\mathrm{aheadpass})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_a& (1-\mathrm{\&alpha;}-\mathrm{aheadpass})/L_{\mathrm{PFIR}}<f\&le;(1+\mathrm{\&alpha;}+\mathrm{delaystop})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_s& f>(1+\mathrm{\&alpha;}+\mathrm{delaystop})/L_{\mathrm{PFIR}}\end{array}\right.---(1-11)$

The setting of weighted error obtains according to following formula in the prior art:

$\mathrm{weight}\left(f\right)=\left\{\begin{array}{cc}{\mathrm{\&delta;}}_p& 0\&le;f\&le;(1-\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_a& (1-\mathrm{\&alpha;})/L_{\mathrm{PFIR}}<f\&le;(1+\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_s& f>(1+\mathrm{\&alpha;})/L_{\mathrm{PFIR}}\end{array}\right.---(1-12)$

Though yet do not leak into other frequency bands with the filter transition band that this designs, but the shake meeting of transition band exerts a certain influence to the orthogonality of filter, the passband ripple of filter can not reach desirable effect, thereby makes the EVM of system relatively worsen.

After the amplitude-frequency characteristic of above-mentioned PFIR filter and weighted error design are finished, call function firpm has just finished the design of PFIR filter coefficient, and filter coefficient can download in FPGA (the field programmable gate array Field Programmable Gate Array) chip and use like this.The multiple filter of PFIR and CIC cascade just can be finished the filter function to many speed, and the multiple filter that designs like this can well improve the shake of passband ripple, reduces the infringement to signal, and systematic function is provided.

After the design of having finished junction filter, present embodiment has proposed the method for a kind of further minimizing digital filter to the infringement of signal.

The end of the data of a time slot of ofdm system physical layer is made up of TI (Timeslot Interval), and what TI partly transmitted all is invalid numerical value, and cyclic prefix CP is to be used for overcoming multipath to disturb, and needs to keep.Present embodiment proposes the redundant digit that convolution causes is carried out asymmetric cut position for this reason, the few intercepting of a part before the time slot, and the time slot rear section intercepts more, and the signal length of preferred many interceptings is [1～50] individual sampled points.This cut position method can make the demodulation quality of signal be improved, and then further improves systematic function.

Above-described embodiment only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (9)

1. the design digital filter based on ofdm system is characterized in that, this method comprises:

Obtain decomposable interpolation factor by frequency spectrum continuation with passband;

Determine junction filter by decomposable interpolation factor, with the junction filter of first order filter and second level filter cascade digital filter as ofdm system by first order filter and second level filter cascade;

With the interpolation factor of described first order filter interpolation factor the amplitude-frequency characteristic of described junction filter is calculated as described junction filter;

Calculate the amplitude-frequency characteristic of first order filter according to the amplitude-frequency characteristic of described second level filter and junction filter;

The transition band of continuation first order filter, and the weighted error of first order filter is set according to the continuation amount of transition band;

Amplitude-frequency characteristic and weighted error according to described first order filter are calculated its filter coefficient.

2. method according to claim 1 is characterized in that:

It is poor that transmitted bandwidth and effective bandwidth are got, and this difference is the maximum continuation amount of described passband;

With the passband after the continuation is new passband, obtains decomposable interpolation factor.

3. method according to claim 1 is characterized in that, the transition band of described continuation first order filter is specially:

The passband of first order filter is put forward predetermined band and stopband delayed one section allocated frequency band the last period.

4. method according to claim 3 is characterized in that: with transition band by (1-α)/L _PFIR＜f≤(1+ α)/L _PFIRContinuation is (1-α-aheadpass)/L _PFIR＜f≤(1+ α+delaystop)/L _PFIR, wherein, aheadpass is the frequency band that shifts to an earlier date in the passband, and delaystop is the frequency band that stopband is delayed, and α is a rolloff-factor, L _PFIRBe the interpolation factor of first order filter, f is a frequency band.

5. method according to claim 4 is characterized in that, the weighted error after the transition band continuation is:

$\mathrm{weight}\left(f\right)=\left\{\begin{array}{cc}{H}_{\mathrm{CIC}}\left(f\right)\&CenterDot;{\mathrm{\&delta;}}_p& 0\&le;f\&le;(1-\mathrm{\&alpha;}-\mathrm{aheadpass})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_a& (1-\mathrm{\&alpha;}-\mathrm{aheadpass})/L_{\mathrm{PFIR}}<f\&le;(1+\mathrm{\&alpha;}+\mathrm{delaystop})/L_{\mathrm{PFIR}}\\ {\mathrm{\&delta;}}_s& f>(1+\mathrm{\&alpha;}+\mathrm{delaystop})/L_{\mathrm{PFIR}}\end{array}\right.$

Wherein, H _CIC(f) δ _pBe the passband weighted error after the frequency band shake that adds second level filter, H _CIC(f) be the ripple shake of cascade pectination integral comb filter, aheadpass is the frequency band that shifts to an earlier date in the passband, and delaystop is the frequency band that stopband is delayed, and α is a rolloff-factor, L _PFIRBe the interpolation factor of first order filter, f is a frequency band, δ _pBe passband weighted error, δ _aBe transition band weighted error, δ _sBe the stopband weighted error.

6. method according to claim 1 is characterized in that, this method also comprises behind described calculating filter coefficient:

The redundant digit that convolution causes is carried out asymmetric cut position, the few intercepting of a part before the time slot, the time slot rear section intercepts more.

7. method according to claim 6 is characterized in that:

The signal length of many interceptings is a 1-50 sampled point.

8. according to each described method of claim 1-5, it is characterized in that: described first order filter is programmable finite impulse response filter.

9. according to each described method of claim 1-5, it is characterized in that: described second level filter is a cascade pectination integral comb filter.

Images