CN101355538A - System and method for modulating frequency domain of block transmission system based on filter set - Google Patents

Abstract

The invention provides a system and a method for modulating frequency domain of a block transmission system based on a wave filter set. The method comprises the following: firstly, an inputted serial modulating symbol sequence is divided into D data block sequences, with a length of K for each, the K point orthogonal transformation is performed; secondly, K elements outputted by each orthogonal transformation are respectively mapped to the K subbelts distributed by the system to transmit by the subbelt mapping; thirdly, the D point DFT transformation is performed on the data block with a length of D transmitted on each subbelt, the data block with a length of D*N is expanded in cycle; fourthly, the K data blocks with a length of D*N are respectively multiplied by corresponding spectral window functions on eachbelt to perform the frequency spectrum forming and the data block accumulation so as to form a data block with a length of D*N; and finally, the D*N point inverse Fourier transform to the data block is performed and the cyclic prefix is added to transmit the processed data block, thereby reducing the complexity on the condition of keeping the performance, particularly on the condition that the number of the distributed sub belts is small such as an upstream link circuit.

Description

Frequency domain of block transmission system modulating system and method based on bank of filters

Technical field

The present invention relates to a kind of frequency domain of block transmission system modulating system and method based on bank of filters.

Background technology

In recent years, wireless communication system develops rapidly towards the broadband direction thereupon, and the bandwidth that wireless communication system occupies is more and more higher, and transmission rate is more and more higher, and spectrum efficiency also requires more and more higher.When a plurality of users insert simultaneously in wideband wireless mobile communication system and broadband radio access network, need to adopt multiple access technology.Usually the multiple access technology that adopts mainly contains three kinds: frequency division multiple access, time division multiple access and code division multiple access.The frequency division multiple access technology is that user's information distribution is transmitted to the carrier channel of different frequency; Tdma is that different information distribution is transmitted to different time slots, and a carrier wave can transmit a plurality of users' information by time slot, and the number of users of transmission depends on the number of time slot; CDMA (Code Division Multiple Access) adopts spread spectrum communication mode, a plurality of users' of the pseudo noise code modulation that it can be different with transmission on the same carrier wave at one time signal.According to discovering in recent years,, will become the main multiple access technology of future mobile by the multiple access technology of frequency division multiple access technology (FDMA) and tdma (TDMA) combination for the throughput of effective elevator system.

Existing fdma system mainly contains two kinds of implementations, and a kind of fdma system that is based on the OFDM technology is as OFDM (OFDMA) with based on OFDM (DFT-S-OFDMA) system of DFT spread-spectrum; Another kind is based on the fdma system of bank of filters technology, as filtering multitone (FMT), and generalized multi-carrier (GMC) and based on generalized multi-carrier (DFT-S-GMC) frequency division multiple access scheme of discrete Fourier transform (DFT).Have the multiple access that synchronous error is caused with respect to fdma system and disturb responsive defective based on the OFDM technology; based on the fdma system of bank of filters technology because the bandwidth of its each subband is bigger with respect to carrier wave frequency deviation and Doppler frequency shift; and has certain frequency domain protection between each subband at interval; there is the frequency spectrum of its each subband to have precipitous attenuation outside a channel again, so disturbs between the multi-user that carrier wave frequency deviation and timing error are caused of this system and have stronger robustness.

See also Fig. 1, it is the time domain realization block diagram of existing block transmission system based on bank of filters (DFT-S-GMC).Modulated signal then enters K point orthogonal converter 12 and carries out the orthogonal transform that K is ordered, and then carry out the subband mapping after handling through data block segmenting device 10, serial/parallel conversion equipment 11, supposes that the signal of subband mapping device 13 outputs can be expressed as { e _k, k=0,1 ..., D-1}, here, e _kRepresent that also a number of elements is the column vector of M, wherein M is the total number of sub-bands of many Methods of Subband Filter Banks, and through IFBT converting means 14, after displacement adding up device 15 became block assembly 16 with loop-around data, output signal can be expressed as:

$s\left(n\right)=\underset{k=0}{\overset{D-1}{\mathrm{\&Sigma;}}}{g}_k{\left((n-\mathrm{kN})\right)}_{D\&times;N},$ N=0,1 ..., D * N-1, wherein, $g_k\left(n\right)=\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{e}_k\left(m\right)f_p\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;mn}/M),$ N=0,1 ..., L-1, (()) _QExpression delivery Q computing, f _p(n) be the prototype filter coefficient, g _k(n) be tilde.

Usually, in existing block transmission system (DFT-S-GMC) based on bank of filters, modulate for the K subband, this modulated structure need carry out the orthogonal transform that K is ordered earlier, carry out the IDFT conversion that M is ordered then, then the data block of the IDFT conversion output cycle of carrying out is expanded, and will expand back data block and M Methods of Subband Filter Banks prototype filter coefficient (that is discrete impulse response of prototype filter) dot product the cycle, finish the spectral shaping of each subband in time domain, forming length is the inverse filterbank figure shift of L, that is tilde, subsequently, it is the tilde of DxN (DxN is greater than L) that the zero padding of inverse filterbank figure shift afterbody is formed length, last, the tilde that is DxN with D length adds up after the cyclic shift one by one by shift intervals N, and forming length is many Methods of Subband Filter Banks modulation signal of DxN.Hence one can see that, for time domain filtering group modulated structure, multiplexing (displacement adds up) of moulding of a many subband spectrum and D tilde finishes in time domain, because for realizing the spectral characteristic of each precipitous decay in subband spectrum edge, bank of filters prototype filter coefficient number is more, therefore, time-domain spectral moulding and symbol are multiplexing will cause bigger realization/computation complexity.

In sum, how to solve the existing shortcoming that exists based on the block transmission system of bank of filters, become the technical task that those skilled in the art need to be resolved hurrily in fact.

Summary of the invention

The object of the present invention is to provide a kind of frequency domain of block transmission system modulating system and method, to realize the reduction of communication system complexity based on bank of filters.

In order to achieve the above object, frequency domain of block transmission system modulating system based on bank of filters provided by the invention, comprise the data block segmenting device, the first serial/parallel conversion equipment, K point orthogonal converter, the subband mapping device, and Cyclic Prefix adding set, also comprise and be preset with time-domain multiplexed tilde number, total number of sub-bands, the number of sub-bands that takies, the prototype filter shift-orthogonal at interval, and the bank of filters of prototype filter coefficient, bank of filters comprises again: be connected with described subband mapping device respectively, being used for will being mapped to the first parallel symbol sequence of blocks of data same sequence number element on the respective sub-bands according to default time-domain multiplexed tilde number goes here and there and changes to obtain the second parallel symbol sequence of blocks of data of each subband correspondence, wherein, the number of the second serial/parallel conversion equipment is less than or equal to a plurality of second serial/parallel conversion equipment of the total number of sub-bands of default bank of filters; Be connected with the described a plurality of second serial/parallel conversion equipment respectively, the time-domain multiplexed tilde number that is used to preset carries out a plurality of second parallel symbol data blocks of respective sub-bands correspondence respectively a plurality of discrete Fourier transformers of corresponding second discrete Fourier transform (DFT) of counting; Be respectively applied for a plurality of cycles expansion devices of a plurality of second parallel symbol data blocks of the process discrete Fourier transform (DFT) of each subband correspondence cycle of carrying out respectively being expanded according to described bank of filters prototype filter shift-orthogonal interval; Be respectively applied for a plurality of spectral shaping devices that a plurality of second parallel symbol data blocks through the cycle expansion of each subband correspondence carried out the frequency domain spectral shaping respectively according to bank of filters prototype filter coefficient and each subband center frequency position; Be used for a plurality of second parallel symbol data blocks through spectral shaping of each subband correspondence are superposeed to form many subbands stacking apparatus of corresponding single parallel symbol data block; Be used for the parallel symbol data block that stack forms is carried out exporting to after the corresponding thirdly inverse of a number discrete Fourier transform (DFT) the contrary discrete Fourier transformer of described Cyclic Prefix adding set.

In addition, frequency domain of block transmission system modulator approach based on bank of filters of the present invention comprises step: 1) take number of sub-bands and time-domain multiplexed tilde number is divided into the string character sequence of blocks of data with modulated string character data sequence according to default; 2) described string character sequence of blocks of data is gone here and there and change according to the default number of sub-bands that takies to obtain the first parallel symbol sequence of blocks of data; 3) according to the default number of sub-bands that takies each data block in the described first parallel symbol sequence of blocks of data is carried out corresponding first orthogonal transform of counting; 4) will map to respectively on the corresponding subband through each element of each data block in the first parallel symbol sequence of blocks of data of orthogonal transform; 5) will be mapped in the first parallel symbol sequence of blocks of data on the respective sub-bands in each data block the same sequence number element according to default time-domain multiplexed tilde number goes here and there and changes to obtain a plurality of second parallel symbol data blocks of each subband correspondence; 6) according to default time-domain multiplexed tilde number a plurality of second parallel symbol data blocks of each subband correspondence are carried out corresponding second discrete Fourier transform (DFT) of counting respectively; 7) at interval will carry out the cycle expansion respectively according to bank of filters prototype filter shift-orthogonal through a plurality of second parallel symbol data blocks of each subband correspondence of discrete Fourier transform (DFT); 8) according to bank of filters prototype filter coefficient and each subband center frequency position a plurality of second parallel symbol data blocks of expanding through the cycle of each subband correspondence are carried out the frequency domain spectral shaping respectively; 9) a plurality of second parallel symbol data blocks through spectral shaping with each subband correspondence superpose to form corresponding single parallel symbol data block; 10) the parallel symbol data block that forms that will superpose is carried out corresponding thirdly inverse of a number discrete Fourier transform (DFT) to obtain the cycling wave form sequence; 11) the head or tail portion in the cycling wave form sequence adds the protection of a length-specific at interval to reduce interchannel interference.

Wherein, described step 3) adopts a kind of in discrete Fourier transform (DFT), Walsh-Hadamard transform and the identical transformation to carry out orthogonal transform, described step 4) adopts Continuous Mappings mode or Discrete Mapping mode to shine upon processing, in described step 8), if the parallel symbol sequence of blocks of data of expanding through the cycle is { i _m, m=0,1 ..., M-1}, it is transformed to { l through behind the spectral shaping _m, m=0,1 ..., M-1} satisfies l _m(k)=i _m(k) F _m(k), k=0 ..., D * N-1, m=0,1 ..., M-1, wherein F _m(k) be the frequency response of m subband correspondence, and $F_m\left(k\right)=\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;mn}/M)\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N)),$ { f _p(n), n=0,1,2..., L-1} are bank of filters prototype filter coefficient, L is a filter length, and has $\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)f_p^{*}(n-\mathrm{kN})=\mathrm{\&delta;}\left(k\right),$ N is prototype filter shift-orthogonal interval, and described step 8) also comprises the frequency response F when m subband correspondence _mWhen (k) energy is less than preset threshold value with described F _m(k) be set at 0 step

In sum, of the present invention based on bank of filters the frequency domain of block transmission system modulating system and method by modulate the complexity that can effectively reduce communication system at frequency domain.

Description of drawings

Fig. 1 is that the time domain of the transmitter of existing block transmission system based on bank of filters realizes block diagram.

Fig. 2 is the frequency domain of block transmission system modulating system basic structure schematic diagram based on bank of filters of the present invention.

Fig. 3 is a relatively schematic diagram of the bit error rate performance of the frequency domain of block transmission system modulating system based on bank of filters of the present invention when adopting 1 subband.

Fig. 4 is a relatively schematic diagram of the bit error rate performance of the frequency domain of block transmission system modulating system based on bank of filters of the present invention when adopting 6 subbands.

Embodiment

See also Fig. 2, frequency domain of block transmission system modulating system based on bank of filters of the present invention comprises: data block segmenting device 10, the first serial/parallel conversion equipment 11, the orthogonal converter 12 that K is ordered, subband mapping device 13, be preset with time-domain multiplexed tilde number, total number of sub-bands, the number of sub-bands that takies, the prototype filter shift-orthogonal at interval, and the M Methods of Subband Filter Banks of prototype filter coefficient, with Cyclic Prefix adding set 20, wherein, described M Methods of Subband Filter Banks comprises the second serial/parallel conversion equipment 140 again, 141 and 142, D point discrete Fourier conversion (DFT) device 150,151 and 152, cycle is expanded device 160,161 and 162, spectral shaping device 170,171 and 172, many subbands stacking apparatus 18, and contrary discrete Fourier transform (DFT) (IDFT) device 19 of DxN point.Be simplified illustration, the second serial/parallel conversion equipment, the conversion of D point discrete Fourier and cycle expand device and only illustrate 3 respectively, but are not as limit, and its number number of sub-bands that comprises with described subband mapping device respectively is identical.In addition, described frequency domain of block transmission system modulating system based on bank of filters is to carry out the signal modulation for the shortcoming of avoiding prior art to exist adopts at frequency domain, so its parameter that adopts (for example, total number of sub-bands of bank of filters, prototype filter shift-orthogonal at interval, and the prototype filter coefficient) all determined in advance with reference to prior art.

In addition, need to prove, channel coding device, digital modulation device, RF converter plant and a transmitting antenna as digital communication system transmitter necessary component are not improvements of the present invention, and the structure and the principle of those necessary component are familiar with by those skilled in the art, so no longer be described at this.And, suppose { a earlier for making description more clear _n, n=0,1,2....} is the serial modulated symbols sequence that is input to described data block segmenting device 10.

Described data block segmenting device 10 is used for modulated string character data sequence { a _n, n=0,1,2....} is divided into string character sequence of blocks of data { b _m, m=0,1,2....}, wherein, b _mRepresent that a number of elements equals the row vector of D * K, D is described D point DFT device 150, the DFT conversion counts in 151 and 152, that is time-domain multiplexed tilde number in the signal of default bank of filters modulation output, K is the counting of orthogonal transform in the described orthogonal converter 12, that is the number of sub-bands that takies of the signal of default bank of filters modulation output.Because the digital communication system transmitter is identical and is independently the operation of each serial data block of input, therefore only describes the operation to a serial data block of emitter and receiving system in scheme is thereafter described.

Described first string and conversion equipment 11 are used for string character sequence of blocks of data { b _m, m=0,1, each data block is gone here and there and conversion operations among the 2....}, to form D the first parallel symbol data block { c _k, k=0,1 ..., D-1}, wherein, c _kBe a number of elements with orthogonal converter 12 in the conversion identical column vector of K of counting;

Described orthogonal converter 12 is used for D the first parallel symbol data block { c to input _k, k=0,1 ..., D-1} carries out the orthogonal transform of K point.Wherein, can adopt discrete Fourier transform (DFT) (DFT), (WH) conversion of Walsh-Hadamard or identical transformation (being that conversion output signal vector and input signal vector are identical) to carry out orthogonal transform.Optimal way is for adopting K point DFT conversion.In orthogonal converter 12, the first parallel data block sequence { c of input _k, k=0,1 ..., after the D-1} process K point DFT conversion, be transformed into corresponding data block sequence { d _k, k=0,1 ..., D-1}, relation is each other obeyed d _k=FFT (c _k), wherein, d _kAlso represent a number of elements and the equirotal column vector of orthogonal transform.Orthogonal transform size K equals the number of sub-bands that transmission signals takies, and can carry out the self adaptation adjustment according to the required transmission rate of communication system.

Described subband mapping device 13 is used for the data block d through DFT conversion output _kIn each element be mapped to respectively on the corresponding subband and transmit, for the subband transmission 0 that does not have data map.The mode of mapping can be the Continuous Mappings mode, and soon each element map in the data block is to the continuously arranged a plurality of subbands of frequency spectrum upper frequency; Also can be the Discrete Mapping mode, soon each element map in the data block be to the spaced a plurality of subbands of frequency spectrum upper frequency.Through the subband mapping device, the sequence of blocks of data { d that input is parallel _k, k=0,1 ..., D-1} is transformed into corresponding data block sequence { e _k, k=0,1 ..., D-1}, e _kRepresent that also a number of elements is the column vector of M, wherein M is the total number of sub-bands of many Methods of Subband Filter Banks, and described number of sub-bands is definite in advance.

The described second serial/parallel conversion equipment 140,141 and 142 is respectively applied for according to default time-domain multiplexed tilde number being mapped to the second parallel symbol sequence of blocks of data { e that transmits on each subband _k, k=0,1 ..., each data block same sequence number element is gone here and there and conversion operations among the D-1}, to form M the parallel data block { g that size is D _m, m=0,1 ..., M-1}, wherein, g _mRepresent in a number of elements and the DFT converting means 150,151 and 152 the conversion the same column vector of D of counting, the number of the second serial/parallel conversion equipment is less than or equal to the total number of sub-bands of default bank of filters.

Described discrete Fourier transform (DFT) (DFT) device 150,151 and 152 is respectively applied for according to the second parallel symbol sequence of blocks of data { g of described default time-domain multiplexed tilde number to each subband correspondence of input _m, m=0,1 ..., M-1} carries out corresponding second, and to count be the DFT conversion that D is ordered.Through the DFT converting means, the parallel sequence of blocks of data of input is transformed into corresponding data block sequence { h _m, m=0,1 ..., M-1}, relation is each other obeyed $h_m\left(k\right)=\underset{n=0}{\overset{D-1}{\mathrm{\&Sigma;}}}{g}_m\left(n\right)\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/D),$ K=0 ..., D-1, m=0,1 ..., M-1, wherein, h _mRepresent the column vector that number of elements is D.

The described cycle is expanded device 160,161 and 162, is respectively applied for according to the second parallel symbol data block preface { h of described bank of filters prototype filter shift-orthogonal interval to each subband correspondence of input _m, m=0,1 ..., the M-1} cycle of carrying out expands.Through cycle expansion device, the second parallel data block sequence transformation of input becomes corresponding parallel data block sequence { i _m, m=0,1 ..., M-1}, relation is each other obeyed i _m(k)=h _m((k)) _D, k=0 ..., D * N-1, m=0,1 ..., M-1, (()) _DExpression delivery D computing, N is bank of filters prototype filter shift-orthogonal interval, that is the up-sampling rate, its value is definite in advance, wherein, i _mRepresent the column vector that number of elements is D * N.

Described spectral shaping device 170,171 and 172 is respectively applied for according to bank of filters prototype filter coefficient and the frequency domain transmission signals of each subband center frequency position to each subband correspondence of input, the i.e. second parallel symbol data block preface { i _m, m=0,1 ..., M-1} carries out the frequency domain spectral shaping.Through the spectral shaping device, the parallel sequence of blocks of data of input is transformed into corresponding parallel data block sequence { l _m, m=0,1 ..., M-1}, relation is each other obeyed l _m(k)=i _m(k) F _m(k), k=0 ..., D * N-1, m=0,1 ..., M-1, wherein F _m(k) be the frequency response of m subband correspondence, and $F_m\left(k\right)=\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;mn}/M)\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N)),$ { f wherein _p(n), n=0,1,2..., L-1} are bank of filters prototype filter coefficient (that is impulse response), and L is a filter length, and its frequency response is the frequency response of list low-pass filter, and this filter satisfies the shift-orthogonal condition: $\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)f_p^{*}(n-\mathrm{kN})=\mathrm{\&delta;}\left(k\right),$ N is filter shift-orthogonal interval, that is the up-sampling rate, here, and l _mRepresent the column vector that number of elements is D * N; Need to prove, because the energy of each sub-bands of frequencies response | F _m(k) | ²Mainly concentrate on the limited minority Frequency point, therefore can only keep the N of each sub-bands of frequencies response energy maximum _mValue on the individual frequency, and directly its residual value is made as zero.That is, for m subband, its approximate frequency response can equivalence be ${\tilde{F}}_m\left(k\right)=\left\{\begin{array}{cc}{F}_m\left(k\right),& {\left|F_m\left(k\right)\right|}^2>\mathrm{\&xi;}\\ 0,& \mathrm{otherwise}\end{array}\right.,$ ξ is given threshold value.Under optimal situation, by frequency response zero setting, can make just in time phase non-overlapping copies of frequency number that adjacent each subband takies, promptly the frequency that takies of all subbands is counted sum and is just in time equaled D * N.Like this, under the very little condition of performance loss, the implementation complexity of the spectral shaping computing of each subband can significantly reduce.

Described many subbands stacking apparatus 18 is used for each the subband transmission signals behind the spectral shaping, i.e. parallel symbol data block preface { l _m, m=0,1 ..., M-1} superposes to form corresponding single parallel symbol data block.Through too much subband stacking apparatus, a plurality of parallel data block sequence transformations of input become corresponding parallel data block o (k), k=0,1 ..., D * N-1}, relation is each other obeyed $o\left(k\right)=\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{l}_m\left(k\right),$ K=0 ..., D * N-1, here, { o (k) } represents the column vector that number of elements is D * N.

Described IDFT (contrary discrete Fourier transform (DFT)) device 14, be used for to input the parallel symbol sequence of blocks of data o (k), k=0,1 ..., D * N-1} carries out thirdly that number is the inverse Fourier transform that D * N is ordered.Through the IDFT converting means, the parallel sequence of blocks of data of input be transformed into corresponding parallel data block s (n), n=0,1 ..., D * N-1}, relation is each other obeyed $s\left(n\right)=\frac{1}{D\&times;N}\underset{k=0}{\overset{D\&times;N-1}{\mathrm{\&Sigma;}}}o\left(k\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N)),$ n＝0，1，...，D×N-1。Here { s (n) } is expressed as the parallel data block that block length is D * N.In fact,

$s\left(n\right)=\frac{1}{D\&times;N}\underset{k=0}{\overset{D\&times;N-1}{\mathrm{\&Sigma;}}}\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{l}_m\left(k\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N))$

$=\frac{1}{D\&times;N}\underset{k=0}{\overset{D\&times;N-1}{\mathrm{\&Sigma;}}}\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{i}_m\left(k\right)F_m\left(k\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N))$

$=\frac{1}{D\&times;N}\underset{k=0}{\overset{D\&times;N-1}{\mathrm{\&Sigma;}}}\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}\underset{d=0}{\overset{D-1}{\mathrm{\&Sigma;}}}{g}_m\left(d\right)\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kd}/D)F_m\left(k\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N))$

$=\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}\underset{d=0}{\overset{D-1}{\mathrm{\&Sigma;}}}{g}_m\left(d\right)f_m{\left((n-\mathrm{dN})\right)}_{D\&times;N}$

$f_m\left(n\right)=\frac{1}{D\&times;N}\underset{k=0}{\overset{D\&times;N-1}{\mathrm{\&Sigma;}}}{F}_m\left(k\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N))$

$=f_p\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;mn}/M)$

In the following formula,

Order ${\tilde{g}}_d\left(n\right)=\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{g}_m\left(d\right)f_p\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;mn}/M),$ Then the IDFT output signal can be expressed as

$s\left(n\right)=\underset{d=0}{\overset{D-1}{\mathrm{\&Sigma;}}}{\tilde{g}}_d{\left((n-\mathrm{dN})\right)}_{D\&times;N}$

Because g _m(d)=e _k(m), d=k, so this signal expression becomes the signal expression behind the block assembly identical through loop-around data during with available technology adopting time domain modulator approach.

Described Cyclic Prefix adding set 20, the protection that is used for adding a length-specific in the head or tail portion of cycling wave form sequence are used to reduce interchannel interference (preferably, this protection length at interval should greater than channel maximum delay extension length) at interval.Preferably, protection adding set at interval can be adopted Cyclic Prefix (CP) adding set, and a part that also is about to described data block afterbody copies to its front end, forms the data block symbols of final band CP.Through the Cyclic Prefix adding set, input data sequence s (n), n=0,1 ..., D * N-1} be transformed into complete data block symbols sequence t (n), n=0,1 ..., P-1}, wherein, P=D * N+C, C are circulating prefix-length.

Frequency domain of block transmission system modulator approach based on bank of filters of the present invention mainly may further comprise the steps:

The first step: take number of sub-bands and time-domain multiplexed tilde number is divided into the string character sequence of blocks of data with modulated string character data sequence according to default, be about to { a _n, n=0,1,2....} is divided into string character sequence of blocks of data { b _m, m=0,1,2....}, wherein, b _mRepresent that a number of elements equals the row vector of D * K, D is the counting of DFT conversion in described D point DFT device 150,151 and 152, and K is the counting of orthogonal transform in the described orthogonal converter 12.

Second step: described string character sequence of blocks of data is gone here and there and change according to the default number of sub-bands that takies, be about to { b to obtain the first parallel symbol sequence of blocks of data _m, m=0,1,2....} is transformed to D the first parallel symbol data block { c _k, k=0,1 ..., D-1}, obviously, c _kBe a number of elements with orthogonal converter 12 in the conversion identical column vector of K of counting.

The 3rd step: the described first parallel symbol sequence of blocks of data is carried out corresponding first orthogonal transform of counting according to the default number of sub-bands that takies, can adopt the orthogonal transform of carrying out in discrete Fourier transform (DFT), Walsh-Hadamard transform or the identical transformation, in the present embodiment, adopt K point DFT conversion with { c _k, k=0,1 ..., D-1} is transformed to { d _k, k=0,1 ..., D-1}.

The 4th step: will map to respectively on the corresponding subband through each element in the first parallel symbol sequence of blocks of data of orthogonal transform, can adopt Continuous Mappings mode or Discrete Mapping mode to shine upon processing, in the present embodiment, handle { d through mapping _k, k=0,1 ..., D-1} is transformed into corresponding data block sequence { e _k, k=0,1 ..., D-1}, e _kRepresent that also a number of elements is the column vector of M, wherein M is the total number of sub-bands of many Methods of Subband Filter Banks.

The 5th step: will be mapped in the first parallel symbol sequence of blocks of data on the respective sub-bands in each data block the same sequence number element according to default time-domain multiplexed tilde number and go here and there and change, and be about to { e to obtain the second parallel symbol sequence of blocks of data of each subband correspondence _k, k=0,1 ..., each data block same sequence number element is gone here and there and conversion operations among the D-1}, to form M the parallel data block { g that size is D _m, m=0,1 ..., M-1}.

The 6th step: according to default time-domain multiplexed tilde number the second parallel symbol sequence of blocks of data of respective sub-bands correspondence is carried out corresponding second discrete Fourier transform (DFT) of counting, be about to the second parallel symbol sequence of blocks of data { g _m, m=0,1 ..., M-1} carries out the DFT that D orders and is transformed into corresponding data block sequence { h _m, m=0,1 ..., M-1}, wherein, $h_m\left(k\right)=\underset{n=0}{\overset{D-1}{\mathrm{\&Sigma;}}}{g}_m\left(n\right)\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/D),$ K=0 ..., D-1, m=0,1 ..., M-1, h _mRepresent the column vector that number of elements is D.

The 7th step: will expand through the second parallel symbol data block of the discrete Fourier transform (DFT) cycle of carrying out respectively at interval according to bank of filters prototype filter shift-orthogonal, the number of times that the cycle expands is determined at interval by bank of filters prototype filter shift-orthogonal, is about to { h _m, m=0,1 ..., M-1} carries out N-1 cycle expansion and is transformed into corresponding parallel data block sequence { i _m, m=0,1 ..., M-1}, wherein, i _m(k)=h _m((k)) _D, k=0 ..., D * N-1, m=0,1 ..., M-1, (()) _DExpression delivery D computing, N is bank of filters prototype filter shift-orthogonal interval, that is the up-sampling rate, wherein, i _mRepresent the column vector that number of elements is D * N.

The 8th step: will carry out the frequency domain spectral shaping through the second parallel symbol sequence of blocks of data that the cycle expands according to bank of filters prototype filter coefficient and each subband center frequency position, if the parallel symbol sequence of blocks of data of expanding through the cycle is { i _m, m=0,1 ..., M-1}, it is transformed to { l through behind the spectral shaping _m, m=0,1 ..., M-1} satisfies l _m(k)=i _m(k) F _m(k), k=0 ..., D * N-1, m=0,1 ..., M-1, wherein F _m(k) be the frequency response of m subband correspondence, and $F_m\left(k\right)=\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;mn}/M)\exp (-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N)),$ { f _p(n), n=0,1,2..., L-1} are bank of filters prototype filter coefficient, L is a filter length, and has $\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)f_p^{*}(n-\mathrm{kN})=\mathrm{\&delta;}\left(k\right),$ N is filter shift-orthogonal interval, because the energy of each sub-bands of frequencies response | and F _m(k) | ²Mainly concentrate on the limited minority Frequency point, therefore in this step, when | F _m(k) | ²During less than preset threshold value ξ, can be set is 0, only keeps the N of each sub-bands of frequencies response energy maximum _mValue on the individual frequency.That is, for m subband, its approximate frequency response can equivalence be ${\tilde{F}}_m\left(k\right)=\left\{\begin{array}{cc}{F}_m\left(k\right),& {\left|F_m\left(k\right)\right|}^2>\mathrm{\&xi;}\\ 0,& \mathrm{otherwise}\end{array}\right.,$ ξ is a pre-set threshold.Under optimal situation, by frequency response zero setting, can make just in time phase non-overlapping copies of frequency number that adjacent each subband takies, promptly the frequency that takies of all subbands is counted sum and is just in time equaled D * N.Like this, under the very little condition of performance loss, the implementation complexity of the spectral shaping computing of each subband can significantly reduce.

The 9th step: will superpose through the second parallel symbol sequence of blocks of data of spectral shaping to form corresponding parallel symbol data block, and be about to { l _m, m=0,1 ..., M-1} be superimposed as corresponding parallel data block o (k), k=0,1 ..., D * N-1}, wherein, $o\left(k\right)=\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{l}_m\left(k\right),$ k＝0，…，D×N-1

The tenth step: the parallel symbol data block that forms that will superpose is carried out corresponding thirdly inverse of a number discrete Fourier transform (DFT) to obtain the cycling wave form sequence, be about to { o (k), k=0,1 ..., D * N-1} carries out the inverse Fourier transform that D * N orders and becomes corresponding parallel data block { s (n), n=0,1 ..., D * N-1}, relation is each other obeyed $s\left(n\right)=\frac{1}{D\&times;N}\underset{k=0}{\overset{D\&times;N-1}{\mathrm{\&Sigma;}}}o\left(k\right)\exp (\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A20071004413200181.png,A20071004413200182.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;kn}/(D\&times;N)),$ n＝0，1，...，D×N-1。

The 11 step: the head or tail portion in the cycling wave form sequence adds the protection of a length-specific at interval with minimizing interchannel interference, soon { s (n), n=0; 1; ..., D * N-1} is transformed into complete data block symbols sequence { t (n), n=0; 1; ..., P-1}, wherein; P=D * N+C, C are circulating prefix-length.

Below will further specify its advantage by the emulation to the frequency domain of block transmission system modulating system based on bank of filters of the present invention, the simulation parameter that is adopted is as follows:

Systematic sampling frequency: 7.68MHz

Sub-band sum order (M): 28

Take number of sub-bands (K): 1 subband/6 subband

Prototype filter: root raised cosine

Prototype filter length (L): 392

Filter up-sampling rate (N): 32

Subband mapping mode: concentrate mapping

Coded system/code check: Turbo (1/2)

Modulation system: QPSK

Antenna configurations: receive for 11

Channel model: PB (3km/h)

Equalization algorithm: MMSE frequency domain equalization

The frequency domain equalization Q:512 that counts

The multiplexing tilde of each data block is counted D:16

Every subband frequency number: Round (512/28), wherein Round (.) represents to round up computing

Contrast simulation system: frequency-domain modulation approach and time domain modulator approach

See also Fig. 3 and Fig. 4, it is respectively relatively schematic diagram of the bit error rate performance of the frequency domain of block transmission system modulating system based on bank of filters of the present invention when adopting 1 subband and 6 subbands, when 1 of obvious employing and 6 subbands, DFT-S-GMC system based on frequency domain modulation compares with existing DFT-S-GMC system based on the time domain modulation, bit error rate promptly adopts frequency domain modulation almost not cause systemic loss of energy much at one.

See also following table 1 again, it is based on time domain modulation with based on the comparison sheet of the complexity of frequency domain modulation:

Modulation of table 1 time domain and frequency domain modulation complexity are relatively

Wherein, 28 FFT is equivalent to FFT and 7 FFT of 4 of 74 times, and 512 and 4 FFT adopt 2-base algorithms, and 7 FFT adopts WFTA algorithm (taking advantage of again for totally 9 times).Because generally speaking, as 10 bit quantizations, the computation complexity of addition is than the low order of magnitude of multiplication, so, the amount of calculation of more not considering addition of complexity.

By table 1 as seen, adopt the transmitter complexity of frequency domain modulation less than the transmitter complexity that adopts the time domain modulation, especially in the less situation of receiving terminal (base station) CU number of sub-bands.As taking 28 subbands when (expiring subband), the real multiplier of frequency domain modulation is about 4/5ths of time domain modulation.And when taking 1 subband, the real multiplier of frequency domain modulation has only 1/2nd of time domain modulation.And if 512 IFFT and 16 FFT of frequency domain modulation adopt basic 4 algorithms, then the complexity of frequency domain modulation will further reduce.

In sum, frequency domain of block transmission system modulating system and method based on bank of filters of the present invention compared with existing delivery plan based on the modulation of time domain filtering group, under the condition that keeps identical performance, it has lower implementation complexity, especially when the less situation of number of sub-bands of distributing, as up link.

Claims (6)

1. frequency domain of block transmission system modulating system based on bank of filters, it comprises data block segmenting device, the first serial/parallel conversion equipment, K point orthogonal converter, subband mapping device, reaches the Cyclic Prefix adding set, it is characterized in that also comprising: be preset with time-domain multiplexed tilde number, total number of sub-bands, the number of sub-bands that takies, prototype filter shift-orthogonal at interval, and the bank of filters of prototype filter coefficient, described bank of filters comprises:

The a plurality of second serial/parallel conversion equipment, be connected with described subband mapping device respectively, being used for will being mapped to the first parallel symbol sequence of blocks of data same sequence number element on the respective sub-bands according to default time-domain multiplexed tilde number goes here and there and changes to obtain a plurality of second parallel symbol data blocks of each subband correspondence, wherein, the number of the second serial/parallel conversion equipment is less than or equal to the total number of sub-bands of default bank of filters;

A plurality of discrete Fourier transformers, be connected with the described a plurality of second serial/parallel conversion equipment respectively, be used for a plurality of second parallel symbol data blocks of respective sub-bands correspondence being carried out corresponding second discrete Fourier transform (DFT) of counting respectively according to described default time-domain multiplexed tilde number;

A plurality of cycles are expanded device, are respectively applied for according to described bank of filters prototype filter shift-orthogonal interval a plurality of second parallel symbol data blocks of the process discrete Fourier transform (DFT) of each subband correspondence cycle of carrying out is respectively expanded;

A plurality of spectral shaping devices are respectively applied for according to bank of filters prototype filter coefficient and each subband center frequency position a plurality of second parallel symbol data blocks through the cycle expansion of each subband correspondence are carried out the frequency domain spectral shaping respectively;

Many subbands stacking apparatus is used for a plurality of second parallel symbol data blocks through spectral shaping of each subband correspondence are superposeed to form corresponding single parallel symbol data block;

Contrary discrete Fourier transformer is used for the parallel symbol data block that stack forms is carried out exporting described Cyclic Prefix adding set to after the corresponding thirdly inverse of a number discrete Fourier transform (DFT).

2. frequency domain of block transmission system modulator approach based on bank of filters is characterized in that comprising step:

1) takies number of sub-bands and time-domain multiplexed tilde number is divided into the string character sequence of blocks of data with modulated string character data sequence according to default;

2) described string character sequence of blocks of data is gone here and there and change according to the default number of sub-bands that takies to obtain the first parallel symbol sequence of blocks of data;

3) according to the default number of sub-bands that takies each data block in the described first parallel symbol sequence of blocks of data is carried out corresponding first orthogonal transform of counting;

4) will map to respectively on the corresponding subband through each element of each data block in the first parallel symbol sequence of blocks of data of orthogonal transform;

5) will be mapped in the first parallel symbol sequence of blocks of data on the respective sub-bands in each data block the same sequence number element according to default time-domain multiplexed tilde number goes here and there and changes to obtain a plurality of second parallel symbol data blocks of each subband correspondence;

6) according to default time-domain multiplexed tilde number a plurality of second parallel symbol data blocks of each subband correspondence are carried out corresponding second discrete Fourier transform (DFT) of counting respectively;

7) at interval will carry out the cycle expansion respectively according to bank of filters prototype filter shift-orthogonal through a plurality of second parallel symbol data blocks of each subband correspondence of discrete Fourier transform (DFT);

8) according to bank of filters prototype filter coefficient and each subband center frequency position a plurality of second parallel symbol data blocks of expanding through the cycle of each subband correspondence are carried out the frequency domain spectral shaping respectively;

9) a plurality of second parallel symbol data blocks through spectral shaping with each subband correspondence superpose to form corresponding single parallel symbol data block;

10) the parallel symbol data block that forms that will superpose is carried out corresponding thirdly inverse of a number discrete Fourier transform (DFT) to obtain the cycling wave form sequence;

11) the head or tail portion in the cycling wave form sequence adds the protection of a length-specific at interval to reduce interchannel interference.

3. the frequency domain of block transmission system modulator approach based on bank of filters as claimed in claim 1 is characterized in that: described step 3) adopts a kind of in discrete Fourier transform (DFT), Walsh-Hadamard transform and the identical transformation to carry out orthogonal transform.

4. the frequency domain of block transmission system modulator approach based on bank of filters as claimed in claim 1 is characterized in that: described step 4) adopts Continuous Mappings mode or Discrete Mapping mode to shine upon processing.

5. the frequency domain of block transmission system modulator approach based on bank of filters as claimed in claim 1 is characterized in that: in described step 8), if the parallel symbol sequence of blocks of data of expanding through the cycle is { i _m, m=0,1 ..., M-1}, it is transformed to { l through behind the spectral shaping _m, m=0,1 ..., M-1} satisfies l _m(k)=i _m(k) F _m(k), k=0 ..., D * N-1, m=0,1 ..., M-1, wherein F _m(k) be the frequency response of m subband correspondence, and $F_m\left(k\right)=\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)\exp (j2\mathrm{\&pi;mn}/M)\exp (-j2\mathrm{\&pi;kn}/(D\&times;N)),$ { f _p(n), n=0,1,2..., L-1} are bank of filters prototype filter coefficient, L is a filter length, and has $\underset{n=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{f}_p\left(n\right)f_p^{*}(n-\mathrm{kN})=\mathrm{\&delta;}\left(k\right),$ N is filter shift-orthogonal interval.

6. the frequency domain of block transmission system modulator approach based on bank of filters as claimed in claim 5 is characterized in that also comprising the frequency response F when m subband correspondence _mWhen (k) energy is less than preset threshold value with described F _m(k) be set at 0 step.

Images