CN100499610C - Low complexity channel estimation method based on orthogonal sequence design - Google Patents

Abstract

The method is used in MIMO-OFDM system having virtual carrier and comprises: the designed orthogonal training sequence is distributed in equal interval in valid sub-carrier on each transmitting antenna; they are pair-wise orthogonal and the rest is filled zero; the transmitting end intercepts a section from the time domain symbol obtained from inverse fast Fourier transform, adds a cycle prefix on it and then transmits it through antenna; the receiving end makes self-copy for the received time domain symbol and then converts it to a frequency domain symbol; the fast Fourier transform, over-sampling and the finite impulse response low-pass interpolating filer are used to make low complexity channel estimation.

Description

A kind of low complexity channel estimation method based on the orthogonal sequence design

Technical field

The invention belongs to MIMO-OFDM (MIMO-OFDM) communication technical field, relate in particular to a kind of low complexity channel estimation method based on the orthogonal sequence design.

Background technology

Reason more and more is subjected to people's attention to OFDM (OFDM, Orthogonal Frequency Division Multiplexing) technology owing to its availability of frequency spectrum height, implementation complexity be low etc.Since the eighties in 20th century, the OFDM technology not only is widely used in broadcast type digital audio and video field, and has become the part of WLAN standard.Along with the enhancing of people to communication dataization, broadband, individualized and mobile demand, the OFDM technology is applied at many high speed information transmission fields.At present, people are considering to use in the radio honeycomb mobile communication system of future generation in future OFDM technology MIMO (Multiple Input Multiple Output) technology to adopt many antennas at transmitting terminal and receiving terminal, can improve power system capacity greatly.For the high speed information transmission system, be frequency-selective channel because the influence of multipath causes mimo channel.The OFDM technology can be converted into the frequency selectivity mimo channel the non-selective mimo channel of many parallel frequencies, has reduced the complexity of receiver.So OFDM technology and MIMO technology combined becomes a focus of studying in the wireless communication field.

Receiving terminal in the MIMO-OFDM system detects and coherent demodulation for signal being carried out MIMO, need estimate that this process is called channel estimating to channel parameter.A kind of typical scheme is the head insertion training sequence (preamble) at a Frame, utilizes these training sequences to carry out channel estimating at receiving terminal.With regard to the channel estimating angle, the optimal training sequence of different transmit antennas design should be satisfied three conditions in the MIMO-OFDM system: 1. constant power; 2. quadrature: 3. shift-orthogonal.As people such as Imad Barhumi at " IEEE Trans.On Signal Processing " vol.51, no.6, pp.1615-1624, June 2003 has delivered " Optimal training design for MIMO OFDM systems in mobile wirelesschannels " (IEEE signal processing journal in June, 2003, the 51st volume, the 6th phase, 1615 to 1624 pages, the optimum of MIMO-OFDM training design in the mobile radio channel).This method for designing has a very big defective, in case be exactly that training sequence shift-orthogonal between the different antennae can not satisfy, will cause the systematic function rapid deterioration, that is to say that this training sequence design does not possess robustness.We know, in actual applications, all there is virtual carrier in ofdm system, the effect of virtual carrier is that it not only makes signal at sideband damping characteristic naturally, form the FFT of " brick wall " shape, and can avoid because frequency shift (FS) is disturbed when above-mentioned training method for designing being applied in the MIMO-OFDM system with virtual carrier the frequency band generation on next door, because the virtual carrier position signalling is zero, this will destroy the shift-orthogonal of training sequence between the different antennae, not only cause the systematic function rapid deterioration, and the computational complexity of channel estimating increases greatly also.Therefore must seek the low complexity channel estimation method of a kind of suitable virtual carrier system.

In IEEE Std 802.16-2004, worked out wireless access system air-interface standard based on the MIMO-OFDM technology, structure when in 802.16 standards, having adopted Alamouti empty, for avoiding the aliasing of receiving terminal, an antenna is when pilot sub-carrier is uploaded transporting frequency information, it is idle that other transmitting antenna must keep, thereby utilize the channel estimation methods of single antenna ofdm system to estimate the characteristic of channel between each transmitting antenna and each reception antenna.This method realizes simple, but exist 2 inherent shortcomings: this method of the first need be transmitted continuous a plurality of OFDM symbol, in mobile time varying channel, part is transmitted channel information that the period reflects can not reflect frequency domain response statistical information on the corresponding pilot sub-carrier of continuous a plurality of OFDM symbol; It two is because each antenna is selected different subcarrier emission pilot tones in same period, though avoided the phase mutual interference, has also caused the loss of signal to noise ratio simultaneously, as adopting N _iDuring two transmit antennas, can cause snr loss 10log ₁₀N _iDB.

G.L.St " people such as uber is at Proceedings of the IEEE; vol.92; No.2; pp.271-294.Feb.2004 has delivered " Broadband MIMO-OFDM wireless communications in addition "; utilize the training sequence of Space Time Coding structural design MIMO-OFDM system; its shortcoming has been to adopt too much OFDM symbol as training sequence, performance loss is bigger in time varying channel.People such as Z.Wu have delivered at IEEE WCNC 2005 " Design of Optimal Pilot-tones for Channel Estimation in MIMO-OFDMSystems ", utilize the training sequence design of space-frequency coding structural design MIMO-OFDM system, this method has very strong robustness to time varying channel, but performance loss is more serious when channel exponent number is big.

Summary of the invention

The objective of the invention is to propose a kind of low complexity channel estimation method that has stronger robustness, low error rate performance and can reduce hardware realization power consumption based on the orthogonal sequence design.

The present invention adopts following technical scheme:

A kind of low complexity channel estimation method based on the quadrature training sequence design at the multi-input multi-output-orthogonal frequency-division multiplexing system that has virtual carrier, be spacedly distributed in effective subcarrier of designed quadrature training sequence on each transmitting antenna, and pairwise orthogonal, all the other zero paddings; Add Cyclic Prefix after one section of the time-domain symbol intercepting that transmitting terminal obtains anti-fast fourier transform and launch by antenna; Receiving terminal is forwarding frequency domain symbol to receiving to after time-domain symbol is carried out self-replacation again; Utilize fast fourier transform, over-sampling and finite impulse response low-pass interpolation filters to carry out the low complex degree channel estimating.

Compared with prior art, the present invention has following advantage:

Be the requirement of taking all factors into consideration channel estimating performance and complexity according to an aspect of the present invention, a kind of MIMO-OFDM short training sequence scheme of orthogonal design is proposed, every transmit antennas adopts the OFDM Short Training symbol identical with number of transmit antennas, on frequency domain, adopt and be spacedly distributed, space-number is consistent with number of transmit antennas, and the training sequence between per two antennas all is a mutually orthogonal.Amplitude equates between a plurality of OFDM Short Training symbols of every transmit antennas, and phase place is by the equal difference value.The design of this OFDM Short Training symbol does not need to satisfy the shift-orthogonal characteristic, and has considered the virtual carrier influence, has very strong robustness; Simultaneously, utilize this orthogonal sequence, can adopt the channel estimation methods of low complex degree at receiving terminal.

According to another aspect of the present invention, propose a kind of low complexity channel estimation method and device of the MIMO-OFDM short training sequence based on above orthogonal design, comprise initial channel estimation module and frequency domain channel interpose module.It has utilized FFT computing, over-sampling and digital lowpass interpolation filtering technology.

Beneficial effect of the present invention:

(1) training sequence of the present invention design is simple with channel estimation methods, and highly versatile, can be used for any MIMO-OFDM system;

(2) training sequence design of the present invention has stronger robustness with channel estimation methods, can not cause systematic function rapid deterioration, stable performance along with the slight variation of channel condition

(3) training sequence of the present invention design reduces greatly with the conventional method computation complexity that channel estimation methods compares, and bit error rate performance also increases, and is practical, greatly reduces hard-wired power consumption.

Description of drawings

Fig. 1 is the frequency domain training sequence schematic diagram.

Fig. 2 is the training sequence emission reception programme figure in the MIMO-OFDM system.

Fig. 3 is the initial channel estimation schematic diagram.

Fig. 4 is a frequency domain channel interpolation schematic diagram.

Fig. 5 is the bit error rate comparison diagram of the following two kinds of channel estimating of low speed TU channel circumstance.

Embodiment

A kind of low complexity channel estimation method based on the quadrature training sequence design at the multi-input multi-output-orthogonal frequency-division multiplexing system that has virtual carrier, be spacedly distributed in effective subcarrier of designed quadrature training sequence on each transmitting antenna, and pairwise orthogonal, all the other zero paddings; Add Cyclic Prefix after one section of the time-domain symbol intercepting that transmitting terminal obtains anti-fast fourier transform and launch by antenna; Receiving terminal is forwarding frequency domain symbol to receiving to after time-domain symbol is carried out self-replacation again; Utilize fast fourier transform, over-sampling and finite impulse response low-pass interpolation filters to carry out the low complex degree channel estimating.

Utilize above-mentioned orthogonal sequence, can adopt the channel estimation methods of low complex degree at receiving terminal, this sequence designs as follows:

The first step: according to the position of effective subcarrier, virtual carrier and the number N of transmitting antenna _i, the position of selection training sequence place subcarrier makes the sub-carrier positions of training sequence effectively be spacedly distributed in the subcarrier, and is spaced apart N _i, the training symbol number that is located in the OFDM symbol is

, corresponding subcarrier sequence number is followed successively by $[u_0,u_1\·\·\·u_{\tilde{N}-1}];$

Second step: design on the 1st antenna in the 1st OFDM symbol

The value of individual training symbol makes this

Individual training symbol power equates that its phase place becomes equal difference to distribute, and is made as $\mathrm{\λ}={[{\mathrm{\λ}}_0,{\mathrm{\λ}}_1,\·\·\·{,\mathrm{\λ}}_{\tilde{N}-1}]}^T$ , except this

Individual subcarrier sequence number is

Training symbol, the equal zero filling of all the other sub-carrier positions;

The 3rd step: in the 1st OFDM symbol on the 1st antenna The value λ of individual training symbol is a benchmark, designs on other antennas and the training symbol design of other positions, makes the 2nd on the 1st antenna to N _iIn the individual OFDM symbol

The value of individual training symbol is λ; N on the 2nd antenna _iThe 1st N that follows on the antenna that the value of group training symbol compares _iThe rotation of group training symbol phase place is followed successively by

The 3rd with N on the antenna _iThe 2nd N that follows on the antenna that the value of group training symbol compares _iThe rotation of group training symbol phase place is followed successively by $\frac{2\mathrm{\π}\·0}{N_i},\frac{2\mathrm{\π}\·0}{N_i},\·\·\·\frac{2\mathrm{\π}\·(N_i-1)}{N_i}$ , the rest may be inferred;

The 4th step: at transmitting terminal, with s ⁽ⁿ⁾(k) k group frequency domain training sequence on the expression n transmit antennas, n=1 wherein, 2 ..., N _iAnd k=1,2 ..., N _i, with each frequency domain training sequence s ⁽ⁿ⁾(k) comprise in

Individual symbol according to sub-carrier positions is Obtain the OFDM frequency domain symbol after in the sequence that to be filled into a length be N, then, the OFDM frequency domain symbol is carried out anti-fast fourier transform, and the preceding N/N after getting it and handling _iIndividual time-domain symbol, last, before time-domain symbol, add behind the Cyclic Prefix separately respectively by each transmission antennas transmit;

The 5th step: at receiving terminal, to obtaining time-domain symbol with time-domain symbol self-replacation N behind each reception antenna received signal removal Cyclic Prefix _iCarry out fast fourier transform again after part, carry out subcarrier then and select, therefrom select the frequency domain training sequence sub-carrier positions Corresponding frequency domain symbol obtains N _rLength is on the root antenna

Frequency domain receiving symbol r (m) (k), m=1 wherein, 2 ..., N _rAnd k=1,2 ..., N _i, the method for utilizing these frequency domain receiving symbols to carry out channel estimating is as follows:

The first step: utilize the characteristics of designed orthogonal sequence, the frequency domain receiving symbol on each group same sub-carrier location is carried out N _iThe Fourier transform of point has obtained the subcarrier sequence number and is

Channel frequency domain response estimation;

Second step: on the every reception antenna

Zero insertion between the individual subcarrier channel estimation expands to whole effective subcarrier;

The 3rd step: to every reception antenna, the channel frequency domain that can estimate whole effective subcarrier by the frequency domain channel interpolation is estimated, specifically comprises following 3 steps:

(1). design a normalization cut-off frequency and be the N of α=(L+1) _iThe FIR low-pass interpolation filters of/N

(2). the FIR interpolation filter of signal that previous step is obtained by designing;

(3). the output displacement of filter is obtained whole effective subcarrier channel estimation.

Below in conjunction with attached circle the present invention is described further:

1. training sequence design

Fig. 1 is the frequency domain training sequence schematic diagram of the short symbol of an OFDM, and dark grid is represented among the figure

Individual training sequence, light grid represent zero.Consider a MIMO-OFDM system, number of transmit antennas N _i, reception antenna is counted N _r, adopt N point FFT, remove the virtual subnet carrier wave and middle direct current subcarrier on low frequency and high frequency both sides after, effective subcarrier number is N ₀Suppose number of transmit antennas N _iCan divide exactly N.Training sequence structure adopts N on time domain _iIndividual length is N/N _i+ L _CPThe short symbol of OFDM to replace original length be N+L _CPThe OFDM symbol; On frequency domain, effectively adopt the training sequence that is spacedly distributed in the subcarrier, be spaced apart N _i, all the other zero paddings.Suppose to be total in the short symbol of an OFDM

Individual training sequence, corresponding subcarrier sequence number is

That is to say except at subcarrier

The place sends outside the training sequence, other subcarrier zero paddings.If channel exponent number is L, the applicable elements of this training sequence design is: antenna number N _i* channel exponent number L≤FFT points N.

Fig. 2 is the training sequence emission reception programme figure in the MIMO-OFDM system.

At transmitting terminal, s ⁽ⁿ⁾(k) k is individual on the expression n transmit antennas

Frequency domain training sequence, n=1 wherein, 2 ..., N _iAnd k=1,2 ..., N _is ⁽ⁿ⁾(k) include in

Individual symbol is with this

Individual symbol according to the subcarrier sequence number is

Being filled into a length is in the sequence of N, i.e. s ⁽ⁿ⁾(k) the 0th symbol filled out at position u ₀, the 1st symbol filled out at position μ ₁, the rest may be inferred, and last symbol is filled out in the position

Wherein,

In effective subcarrier, be spacedly distributed, be spaced apart number of transmit antennas N _iThe IFFT computing that the frequency domain symbol of the N that obtains like this * 1 is ordered through N obtains the time-domain symbol of N * 1, and the time-domain symbol of this N * 1 is divided into N _iEqual portions can prove, this N _iThe symbol of part is on all four, and we only get the 1st part of N/N _iThe symbol of * I, remaining is all given up, and obtains (N/N after adding CP _i+ L _CPThe symbol of) * 1 carries out D/A conversion and up-conversion then, launches by antenna again.

At receiving terminal, through down-conversion, A/D conversion and synchronous, the every (N/N that reception antenna receives _i+ L _CPThe symbol of) * 1 removes CP and obtains N/N afterwards _i* 1 symbol, self-replacation N _iPart obtains the time-domain symbol of N * 1, and the FFT computing that process N is ordered obtains the frequency domain symbol of N * 1, according to the subcarrier sequence number

Extraction obtains

Symbol r ^(m)(k), m=1 wherein, 2 ..., N _rAnd k=1,2 ..., N _i, this part is used for carrying out channel estimating, extracts in effective subcarrier all the other

Individual subcarrier obtains

Symbol z ^(m)(k), m=1 wherein, 2 ..., N _rAnd k=1,2 ..., N _i, this part is used for Noise Variance Estimation.

Transmitting terminal N _iN on the transmit antennas _iThe orthogonal design of individual continuous training sequence can be described as, and the 1st with being N on the transmitting antenna _iThe short symbol of individual identical training; The 2nd with N on the antenna _iThe short symbol of individual training compare the 1st with N on the antenna _iThe short symbol phase rotation of individual training is followed successively by

The 3rd with N on the antenna _iThe short symbol of individual training compare the 2nd with N on the antenna _iThe short symbol phase rotation of individual training is followed successively by

The rest may be inferred.Matrix notation is as follows:

Wherein, $\mathrm{\λ}={[{\mathrm{\λ}}_0,{\mathrm{\λ}}_1,\·\·\·{,\mathrm{\λ}}_{\tilde{N}-1}]}^T$ It is one

Vector.If get N _iBe 2,3 and 4, corresponding orthogonal design is respectively

$\left[\begin{array}{cc}\mathrm{\&lambda;}& \mathrm{\&lambda;}\\ \mathrm{\&lambda;}& -\mathrm{\&lambda;}\end{array}\right],\left[\begin{array}{ccc}\mathrm{\&lambda;}& \mathrm{\&lambda;}& \mathrm{\&lambda;}\\ \mathrm{\&lambda;}& {\mathrm{\&lambda;e}}^{\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C200610039551E00142.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2}{3}\mathrm{\&pi;}}& {\mathrm{\&lambda;e}}^{\mathrm{<figure-callout\; id="4"\; label="j"\; filenames="C200610039551E00143.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{4}{3}\mathrm{\&pi;}}\\ \mathrm{\&lambda;}& {\mathrm{\&lambda;e}}^{\mathrm{<figure-callout\; id="4"\; label="j"\; filenames="C200610039551E00143.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{4}{3}\mathrm{\&pi;}}& {\mathrm{\&lambda;e}}^{\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C200610039551E00142.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2}{3}\mathrm{\&pi;}}\end{array}\right],\left[\begin{array}{cccc}\mathrm{\&lambda;}& \mathrm{\&lambda;}& \mathrm{\&lambda;}& \mathrm{\&lambda;}\\ \mathrm{\&lambda;}& {\mathrm{\&lambda;e}}^{\mathrm{<figure-callout\; id="1"\; label="j"\; filenames="C200610039551E00143.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{1}{2}\mathrm{\&pi;}}& -\mathrm{\&lambda;}& {\mathrm{\&lambda;e}}^{j\frac{3}{2}\mathrm{\&pi;}}\\ \mathrm{\&lambda;}& -\mathrm{\&lambda;}& \mathrm{\&lambda;}& -\mathrm{\&lambda;}\\ \mathrm{\&lambda;}& {\mathrm{\&lambda;e}}^{j\frac{3}{2}}& -\mathrm{\&lambda;}& {\mathrm{\&lambda;e}}^{\mathrm{<figure-callout\; id="1"\; label="j"\; filenames="C200610039551E00143.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{1}{2}\mathrm{\&pi;}}\end{array}\right]$

2. channel estimation methods and device

According to above training sequence design, we can write out received signal r on the m root reception antenna ^(m)Model

r ^(m)＝S·H ^(m)+η ^(m)

Wherein,

$r^{\left(m\right)}=\left[\begin{array}{c}{r}^{\left(m\right)}\left(0\right)\\ r^{\left(m\right)}\left(1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r^{\left(m\right)}(N_i-1)\end{array}\right],$

$H^{\left(m\right)}=\left[\begin{array}{c}{H}^{(m,1)}\\ H^{(m,2)}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ H^{(m,N_i)}\end{array}\right],$

${\mathrm{\&eta;}}^{\left(m\right)}=\left[\begin{array}{c}{\mathrm{\&eta;}}^{\left(m\right)}\left(0\right)\\ {\mathrm{\&eta;}}^{\left(m\right)}\left(1\right)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\mathrm{\&eta;}}^{\left(m\right)}(N_i-1)\end{array}\right]$

H ^{(m, n)}Represent that the subcarrier sequence number is between n transmit antennas and the m root reception antenna

Channel frequency domain response, η ^(m)(k) k short symbol of OFDM on the expression m root reception antenna

The frequency domain noise,

Expression diagonalization computing.According to top signal model, the orthogonal design of combined training sequence obtains u between n transmit antennas and the m root reception antenna _pThe estimated value of the channel frequency domain response of individual subcarrier

${\hat{H}}_{u_p}^{(m,n)}=\frac{1}{N_i{\mathrm{\&lambda;}}_p}\underset{k=0}{\overset{N_i-1}{\mathrm{\&Sigma;}}}{r}_p^{\left(m\right)}\left(k\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="C200610039551E00142.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2\mathrm{\&pi;}}{N_i}(n-1)k},p=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;,\tilde{N}-1$

This process can realize with FFT, as shown in Figure 3.

Fig. 3 is that m root reception antenna is corresponding to the subcarrier sequence number

The channel frequency domain response estimation scheme.M root reception antenna u _pThe N of individual subcarrier _iIndividual continuous symbol

Through the FFT computing, multiply by the factor 1/ (N again _iλ _p) promptly obtain channel estimating ${\hat{H}}_{u_p}^{(m,1)},{\hat{H}}_{u_p}^{(m,2)},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{H}}_{u_p}^{(m,N_i)}.$

Having obtained the subcarrier sequence number is Channel frequency domain response estimation after can estimate whole effective subcarrier by the frequency domain channel interpolation the channel frequency domain estimate that we can design a normalization cut-off frequency is the N of α=(L+1) _tThe FIR low pass filter of/N is realized interpolation.The realization of interpolation filter can be divided into a step: (1) will Zero insertion between the individual subcarrier channel estimation expands to whole effective subcarrier; (2) the FIR interpolation filter of the signal that previous step is obtained by designing; (3) the output displacement with filter obtains whole effective subcarrier channel estimation.Theoretically, the exponent number of interpolation filter is high more, and channel estimated accuracy is high more. and the frequency domain channel interpolation is as shown in Figure 4.

The amount of calculation of traditional channel estimating need be IFFT computing and the N that 1 N is ordered to each reception antenna _iIndividual N/N _iThe FFT computing of point, N _rIt is N that the channel estimating of root reception antenna needs the complex multiplication number of times _r(NlogN+Nlog (N/N _i)).In new channel estimation scheme, the amount of calculation of initial channel estimation needs N to each subcarrier of each reception antenna _iLogN _iInferior complex multiplication needs so altogether

Inferior complex multiplication; Suppose that in addition interpolation filter is the M rank, interpolation needs N altogether so _rMN _oInferior complex multiplication.The complex multiplication number of times that whole channel estimating needs is , when interpolation filter exponent number M gets hour, the computation complexity of new channel estimation scheme is being much smaller than conventional channel estimation scheme obviously.

Embodiment:

We are example with the MIMO-OFDM system of 4 transmit antennas and 4 reception antennas, and how transmitting training sequence and the corresponding channel estimation methods of receiving terminal are described.The total number of sub-carriers N=1024 of OFDM, the CP length L _CP=216, effectively sub-carrier number is 884,70 virtual carriers of low frequency, and 70 virtual carriers of high frequency, training sequence place subcarrier sequence number is 70,74 ..., 954, and constant power distributes.According to the method for designing of front, at transmitting terminal, the training sequence on 4 transmit antennas is followed successively by

$\left[\begin{array}{cccccccccccccccc}1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}\\ 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& j& e^{\frac{129\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{349\&CenterDot;\mathrm{j\&pi;}}{256}}& -1& e^{\frac{257\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{477\&CenterDot;\mathrm{j\&pi;}}{256}}& -j& e^{\frac{385\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{605\&CenterDot;\mathrm{j\&pi;}}{256}}\\ 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& -1& e^{\frac{257\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{477\&CenterDot;\mathrm{j\&pi;}}{256}}& 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& -1& e^{\frac{257\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{477\&CenterDot;\mathrm{j\&pi;}}{256}}\\ 1& e^{\frac{1\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{221\&CenterDot;\mathrm{j\&pi;}}{256}}& -j& e^{\frac{385\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{605\&CenterDot;\mathrm{j\&pi;}}{256}}& -1& e^{\frac{257\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{477\&CenterDot;\mathrm{j\&pi;}}{256}}& j& e^{\frac{129\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{349\&CenterDot;\mathrm{j\&pi;}}{256}}\end{array}\right]$

4 line displays, 4 transmit antennas wherein, continuous 4 groups of frequency domain short training sequences are shown in 888 tabulations, the length of every group of training is 222, every group of first symbol of training sequence be ± 1 or ± j.Every group of training sequence on every antenna carries out zero padding then, is example with the 3rd group of training sequence on the 2nd antenna, before the zero padding is $\begin{array}{ccccc}-1& e^{\frac{257\&CenterDot;\mathrm{j\&pi;}}{256}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{\frac{476\&CenterDot;\mathrm{j\&pi;}}{256}}& e^{\frac{477\&CenterDot;\mathrm{j\&pi;}}{256}}\end{array}$ , after the zero padding be

Wherein, the position at top digitized representation training sequence place, the concrete numerical value of following numeral training sequence, the signal at this place of light grey box indicating are zero.The training sequence of other antennas and other groups is similar with it.Carry out the IFFT computing according to flow process shown in Figure 2 after the zero padding, and intercept preceding 256 symbols, abandon back 768 symbols.Add after the CP, carry out D/A conversion and up-conversion, transmit by antenna at last.Training sequence and data symbol be emission alternately, inserts 19 OFDM data-signals between the adjacent training sequence, and the modulation system that transmits is 4QAM.Channel circumstance is low speed typical urban (TU) channel.

At receiving terminal, through down-conversion, A/D conversion and synchronous, every every group code that reception antenna receives, obtain 768 symbols after removing CP, self-replacation obtains 1024 time-domain symbol for 4 parts, obtains 1024 frequency domain symbols through 1024 FFT computings, extracting wherein, the position is 70,74 ..., 954 symbol obtains 222 symbols, for 4 reception antennas, 4 group codes, such one total 222*4*4 symbol, these symbols are estimated as subsequent communication channel.

For 222*4*4 symbol, represent 222 subcarriers, 4 reception antennas, 4 group codes.For each subcarrier, every reception antenna carries out FFT computing shown in Figure 3 at 4, obtains initial channel estimation; Obtain whole effective subcarrier channel estimation through 4 times of over-samplings shown in Figure 4 and interpolation filter then.The frequency domain interpolation of channel estimating adopts 8 rank and 16 rank FIR low-pass interpolation filters, and the time domain interpolation adopts linear interpolation.MIMO detects and adopts ZF to detect and ordering SIC detection.

To the bit error rate simulation result of system as shown in Figure 5, we detect in (ZF) in ZF as can be seen from figure, and the performance of BER of new method exceeds about 2dB than traditional method when low signal-to-noise ratio, when signal to noise ratio improved, the performance of BER of two kinds of methods reached unanimity; In ordering serial interference elimination (SIC) detects, the performance of BER of new method when low signal-to-noise ratio (＜12dB) exceed about ldB than traditional method, two kinds of method performances are the same when signal to noise ratio arrives 15dB, the slightly inferior properties of new method when signal to noise ratio is higher than 23dB.On the other hand, when interpolation filter adopted 8 rank and 16 rank, the performance of BER of system was almost approaching, but it is more than what to adopt 8 rank computation complexities big to adopt 16 rank obviously, recommends the interpolation filter on selection 8 rank during practical application.

Claims (2)

1, a kind of low complexity channel estimation method based on the quadrature training sequence design at the multi-input multi-output-orthogonal frequency-division multiplexing system that has virtual carrier, it is characterized in that: be spacedly distributed in effective subcarrier of designed quadrature training sequence on each transmitting antenna, and pairwise orthogonal, all the other zero paddings; The preceding N/N of time-domain symbol intercepting that transmitting terminal obtains anti-fast fourier transform _tAdd Cyclic Prefix after the individual time-domain symbol and by antenna emission, N is that fast fourier transform is counted N _tIt is number of transmit antennas; Receiving terminal is forwarding frequency domain symbol to receiving to after time-domain symbol is carried out self-replacation again; Utilize fast fourier transform, over-sampling and finite impulse response low-pass interpolation filters to carry out the low complex degree channel estimating; Described quadrature training sequence designs as follows:

The first step: according to the position of effective subcarrier, virtual carrier and the number N of transmitting antenna _t, the position of selection training sequence place subcarrier makes the sub-carrier positions of training sequence effectively be spacedly distributed in the subcarrier, and is spaced apart N _t, the training symbol number that is located in the OFDM symbol is

Corresponding subcarrier sequence number is followed successively by $[u_0,u_1\&CenterDot;\&CenterDot;\&CenterDot;u_{\tilde{N}-1}];$

Second step: design on the 1st antenna in the 1st OFDM symbol

The value of individual training symbol makes this

Individual training symbol power equates that its phase place becomes equal difference to distribute, and is made as $\mathrm{\&lambda;}={[{\mathrm{\&lambda;}}_0,{\mathrm{\&lambda;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;{\mathrm{\&lambda;}}_{\tilde{N}-1}]}^T,$ Except this

Individual subcarrier sequence number is

Training symbol, the equal zero filling of all the other sub-carrier positions;

The 3rd step: in the 1st OFDM symbol on the 1st antenna

The value λ of individual training symbol is a benchmark, designs on other antennas and the training symbol design of other positions, makes the 2nd on the 1st antenna to N _tIn the individual OFDM symbol

The value of individual training symbol is λ; N on the 2nd antenna _tN on the 1st antenna that the value of group training symbol compares _tThe rotation of group training symbol phase place is followed successively by

N on the 3rd antenna _tN on the 2nd antenna that the value of group training symbol compares _tThe rotation of group training symbol phase place is followed successively by The rest may be inferred;

The 4th step: at transmitting terminal, with s ⁽ⁿ⁾(k) k group frequency domain training sequence on the expression n transmit antennas, n=1 wherein, 2 ..., N _tAnd k=1,2 ..., N _t, with each frequency domain training sequence s ⁽ⁿ⁾(k) comprise in

Individual symbol according to sub-carrier positions is

Obtain the OFDM frequency domain symbol after in the sequence that to be filled into a length be N, then, the OFDM frequency domain symbol is carried out anti-fast fourier transform, and the preceding N/N after getting it and handling _tIndividual time-domain symbol, N is that fast fourier transform is counted N _tBe number of transmit antennas, last, before time-domain symbol, add behind the Cyclic Prefix separately respectively by each transmission antennas transmit;

The 5th step: at receiving terminal, to obtaining time-domain symbol behind each reception antenna received signal removal Cyclic Prefix, with time-domain symbol self-replacation N _tCarry out fast fourier transform again after part, carry out subcarrier then and select, therefrom select the frequency domain training sequence sub-carrier positions

Corresponding frequency domain symbol obtains N _rLength is on the root antenna Frequency domain receiving symbol r ^(m)(k), m=1 wherein, 2 ..., N _rAnd k=1,2 ..., N _t, these frequency domain receiving symbols are as channel estimating.

2, the low complexity channel estimation method based on the quadrature training sequence design according to claim 1 is characterized in that:

The first step: utilize the characteristics of designed orthogonal sequence, the frequency domain receiving symbol on each group same sub-carrier location is carried out N _tThe Fourier transform of point has obtained the subcarrier sequence number and is

Channel frequency domain response estimation;

Second step: on the every reception antenna

Zero insertion between the individual subcarrier channel estimation expands to whole effective subcarrier;

The 3rd step: to every reception antenna, the channel frequency domain that can estimate whole effective subcarrier by the frequency domain channel interpolation is estimated, specifically comprises following 3 steps:

(1). design a normalization cut-off frequency and be the N of α=(L+1) _tThe FIR low-pass interpolation filters of/N, L is a channel exponent number;

(2). the FIR interpolation filter of signal that previous step is obtained by designing;

(3). the output displacement of filter is obtained whole effective subcarrier channel estimation.

Images