CN104135455A - Iterative receiving method for communication system - Google Patents

Abstract

The present invention provides an iterative receiving method for communication system, which combines a Qc-LDPC decoder with a frequency domain determination feedback equalizer, which is based on the minimum mean square error MMSE, with a Turbo iteration principle and builds a joint iterative equalization and decoding algorithm. The equalizer modifies the coefficients of a filter with the soft information provided by the decoder in an iteration manner, eliminates the multipath effect of channel transmission with a low computing complexity, further reduces the bit error rate of data transmission of the communication system and realizes reliable data transmission of the communication system. The present invention is able to have the performance similar with an orthogonal frequency-division multiplexing OFDM system without the problem of high peak-to-average power ratio of OFDM system and suitable for channel condition with serious multipath propagation, and eliminates intersymbol interference introduced by multipath propagation of channel and achieves data transmission with low bit error rate.

Description

A kind of communication system iteration receiving method

Technical field

The invention belongs to wireless communication field, relate to a kind of communication system iteration receiving method based on quasi-circulating low-density parity check Qc-LDPC code and frequency domain decisive feedback equalization device.

Background technology

In wireless communication system, data are when traffic channel, and the multipath effect of dissemination channel can be disturbed ISI between created symbol, thereby reduces the reliability of transfer of data.The multi-carrier communications systems that adopts orthogonal frequency division multiplex OFDM technology is that antagonism channel multi-path is propagated, and realizes the effective technology means of data high-speed transmission.But the multi-carrier communications systems that adopts OFDM technology has the defect of the flat power ratio in peak.

The carrier wave communication system that adopts frequency-domain equalization technology is that a kind of good replacement is selected, and equilibrium is compared with single carrier wave time domain, and frequency domain equalization computation complexity obviously reduces, and has avoided orthogonal frequency division multiplex OFDM system to have the defect of the flat power ratio in peak simultaneously.At present, by numerous communication standards and communication system architecture in the world, adopted, such as: wireless LTE standard, underwater high-speed sound communication system etc.The multichannel DFF of joint space diversity technique, a kind of spatial processing and time domain are processed and combined to eliminate the effective technology means that channel multi-path is propagated the ISI producing, but its time domain realizes, have higher computational complexity, frequency domain is realized and can obviously be reduced the computational complexity that time domain realizes.

Due to the continuous increase of objective demand, more and more higher to the reliability requirement of transfer of data, therefore, use single channel equalization technique to eliminate the ISI that multipath transmisstion produces, cannot meet the performance requirement of the communication system of high speed development.The burst that chnnel coding occurs while being effective correction signal process channel and the effective technology means of random error, reduce the error rate of transfer of data with this, improves the reliability of system data transmission.Channel coding technology and balancing technique are combined, build Joint iteration equilibrium and decoding algorithm, eliminate the symbol error that data produce when multi-path channel transmission, become gradually study hotspot.

Within 2010, Liu waits at < < in SC-FDE system > > mono-literary composition based on Turbo chnnel coding for single carrier data transmission system along blue, proposed to adopt Turbo code decoder and frequency-domain equalizer to carry out the technology of data receiver, its method is: reception data transformation is carried out to equilibrium treatment to frequency domain, convert back again time domain and carry out data demodulates, decoding recovery original transmitted information bit, its weak point is: what (1) frequency-domain equalizer adopted is linear equalization, there is no feedback fraction; (2) decoder and equalizer do not carry out Joint iteration computing.Therefore, the method is under the serious channel condition of multipath transmisstion, and error correcting capability is not enough.

Summary of the invention

In order to overcome the deficiencies in the prior art, the invention provides a kind of communication system iteration receiving method.The inventive method adopts single carrier frequency domain equalization to process, use Turbo iteration theorem, the frequency domain decisive feedback equalization device based on minimizing mean square error MMSE criterion of Qc-LDPC decoder and the present invention's proposition is combined, build Joint iteration equilibrium and decoding algorithm, with lower computation complexity, eliminate channel multi-path and propagate the ISI introducing, further reduce the error rate of communication system data transmission, realize communication system high speed, reliable transfer of data.

The technical solution adopted for the present invention to solve the technical problems comprises the steps:

Step 1: transmitting terminal is first to binary message sequence { d _ncarry out Qc-LDPC coding, then to the information sequence { c after coding _ncarry out QPSK sign map; Data symbol front end after mapping adds Zadoff-Chu sequence as Frame frame head, forms complete frame data; Between every frame data by the full remainder according to connection, form continuous transmitting data { x _n;

Step 2: at receiving terminal, pass through N _rthe transmitting data of the antenna array receiving step one that individual reception antenna forms, the signal that each reception antenna is received is done as follows: process before each frame data, first remove the full remainder certificate between every frame data, then Frame frame head and data symbol are carried out respectively to discrete Fourier transform DFT, obtain frequency domain frame head with frequency domain data symbol n _r>=2;

Step 3: get any one reception antenna frequency domain frame head that step 2 obtains carry out channel estimating, obtain channel estimation value k _trepresent frame head length, k represents k symbol;

Step 4: step 2 is obtained carry out forward direction filtering operation, obtain wherein, A _kfor forward-direction filter coefficient, filtering result is designated as will obtain with step 6 make it poor;

When carrying out step 4 first,

Step 5: step 4 is obtained carry out inverse discrete Fourier transformer inverse-discrete IDFT, obtain time domain output data will send into Qc-LDPC decoder, the likelihood ratio { L (d of decoder output encoder bit _k); If reach the iterations of regulation, to likelihood ratio { L (d _k) carry out 0,1 bit decision, recover original transmitted information bit otherwise, enter step 6, carry out iterative data processing;

The likelihood ratio of coded-bit { L (d _k) carry out the method for 0,1 bit decision: if L is (d _k) > 1, sentence information bit otherwise, sentence information bit

Step 6: { L (the d that step 5 is obtained _k) carry out the estimation of frequency domain transmitting symbol, obtain frequency domain symbol estimated value will the channel estimating obtaining with step 3 carry out feedback filtering, b _kfor inverse filter coefficient, filtering result is designated as upgrade filter coefficient A simultaneously _kand B _k; By the filter coefficient A after upgrading _kand B _ksend into step 4 and carry out interative computation;

The method of estimation of described frequency domain transmitting symbol is wherein, represent transmitting symbol;

Filter coefficient A _kand B _kupdate method: $A_k={\mathrm{\&sigma;}}_s^2{({\mathrm{\&sigma;}}_v^{2}I+{\hat{H}}_k{\mathrm{\&sigma;}}_s^2{\hat{H}}_k^H)}^{-1}{\hat{H}}_k^H;$ $B_k=A_k{\hat{H}}_k^H{\mathrm{\&gamma;}}_{\widehat{S,S}}-{\mathrm{\&gamma;}}_{\hat{S},S}.$ Filter coefficient A _kinitial value be made as complete 1, wherein, noise variance, the oblique variance of frequency domain transmitted data symbols, it is the correlation function of time domain transmitting symbol and estimate symbol.

The invention has the beneficial effects as follows:

(1) a kind of communication system iteration receiving method that the present invention proposes is applied to carrier wave communication system, adopts frequency domain data processing mode, can obtain the performance close with ofdm system, and there is no the problem that ofdm system peak-to-average ratio is high.

(2) a kind of communication system iteration receiving method that the present invention proposes, the frequency domain decisive feedback equalization device of Qc-LDPC decoder and proposition is combined, build Joint iteration data processing, the mode that equalizer utilizes soft information exchange that decoder provides to cross iteration is upgraded the coefficient of filter, revise filtering result, elimination, because channel multi-path is propagated the intersymbol interference of introducing, obtains the transfer of data of low error rate.

(3) a kind of communication system iteration receiving method that the present invention proposes, is applicable to the serious channel condition of multipath transmisstion.

Accompanying drawing explanation

Fig. 1 the present invention relates to communication system transmitting data framing structure schematic diagram;

Fig. 2 a is the emitting structural that the present invention relates to communication system;

Fig. 2 b is the iterative receiver structure based on Qc-LDPC code and frequency domain decisive feedback equalization device;

Fig. 3 is the bit error rate curve comparison figure that adopts Joint iteration method of reseptance of the present invention and there is no iteration;

Fig. 4 is the inventive method equalizer and decoder Joint iteration while getting different iterations, ber curve comparison diagram;

Fig. 5 is the inventive method equalizer and decoder Joint iteration 2 times, while adopting 2 antennas and 3 antenna reception, and ber curve comparison diagram.

Embodiment

Below in conjunction with drawings and Examples, the present invention is further described, the present invention includes but be not limited only to following embodiment.

The present invention includes following steps:

Step 1: transmitting terminal is first to binary message sequence { d _ncarry out Qc-LDPC coding, then to the information sequence { c after coding _ncarry out QPSK sign map.Symbol front end after mapping adds Zadoff-Chu sequence as Frame frame head, forms complete frame data; Between every frame data by the full remainder according to connection, form continuous transmitting data { x _n;

In step 1, as shown in Figure 1, the emitting structural of the related communication system of our bright method as shown in Figure 2 a for transmitting data framing mode;

Step 2: at receiving terminal, pass through N _r(N _r>=2) transmitting of the antenna array receiving step one that individual reception antenna forms, the signal that each reception antenna is received is done as follows: process before each frame data, first remove the zero-filled data of Frame tail end, then Frame frame head and data symbol are carried out respectively to discrete Fourier transform DFT, obtain frequency domain frame head with frequency domain data symbol

Step 3: get the arbitrary reception antenna frequency domain frame head data that step 2 obtains carry out channel estimating, obtain channel estimation value standby;

Step 4: step 2 is obtained carry out forward direction filtering operation, wherein, A _kfor forward-direction filter coefficient, filtering result is designated as will obtain with step 6 make it poor;

When carrying out step 4 first, step 6 is not inputted data and can be utilized, and therefore, step 6 is output as zero,

Step 5: step 4 is obtained carry out inverse discrete Fourier transformer inverse-discrete IDFT, obtain time domain output data will send into Qc-LDPC decoder, carry out soft decoding processing, the likelihood ratio { L (d of decoder output encoder bit _k); If reach the iterations of regulation, to likelihood ratio { L (d _k) carry out 0,1 bit decision, recover original transmitted information bit otherwise, continue step 6, carry out iterative data processing;

Ldpc decoder adopts and amasss BP soft decoding algorithm, BP algorithm referring to [Xiao Yang .Turbo and LDPC encoding and decoding and application [M] thereof. People's Telecon Publishing House, 2010:78-80].

The likelihood ratio of coded-bit { L (d _k) carry out the method for 0,1 bit decision: if L is (d _k) > 1, sentence information bit otherwise, sentence information bit

Step 6: { L (the d that step 5 is obtained _k) carry out the soft estimation of frequency domain transmitting symbol, obtain frequency domain symbol estimated value will the channel estimating obtaining with step 3 carry out feedback filtering, b _kfor inverse filter coefficient, filtering result is designated as upgrade filter coefficient A simultaneously _kand B _k; By the filter coefficient A after upgrading _kand B _ksend into step 4 and carry out interative computation;

Frequency domain transmitting symbol estimation method: ${\hat{S}}_k=\frac{1}{K}\underset{i=0}{\overset{K-1}{\mathrm{\&Sigma;}}}E\left({\hat{x}}_k\right)\exp (-j\frac{2\mathrm{\&pi;ki}}{K});$

Filter coefficient A _kand B _kupdate method: $A_k={\mathrm{\&sigma;}}_s^2{({\mathrm{\&sigma;}}_v^{2}I+{\hat{H}}_k{\mathrm{\&sigma;}}_s^2{\hat{H}}_k^H)}^{-1}{\hat{H}}_k^H;$ $B_k=A_k{\hat{H}}_k^H{\mathrm{\&gamma;}}_{\widehat{S,S}}-{\mathrm{\&gamma;}}_{\hat{S},S}.$ Filter coefficient A _kinitial value be made as complete 1.

The reception structure of the related communication system of this method as shown in Figure 2 b.

Embodiment as shown in Figure 2 a and 2 b, comprises following step:

Step 1: transmitting terminal is first to binary message sequence { d _ncarry out Qc-LDPC coding, then to the information sequence { c after coding _ncarry out QPSK sign map.Symbol front end after mapping adds Zadoff-Chu sequence as Frame frame head, forms complete frame data; Between every frame data by the full remainder according to connection, form continuous transmitting data { x _n;

In step 1, as shown in Figure 1, the emitting structural of the related communication system of our bright method as shown in Figure 2 a for transmitting data framing mode;

Step 2: at receiving terminal, pass through N _r(N _r>=2) transmitting of the antenna array receiving step one that individual reception antenna forms, the signal that each reception antenna is received is done as follows: process before each frame data, first remove the zero-filled data of Frame tail end, then Frame frame head and data symbol are carried out respectively to discrete Fourier transform DFT, obtain frequency domain frame head with frequency domain data symbol

Step 3: get any one reception antenna frequency domain frame head data that step 2 obtains carry out channel estimating, obtain channel estimation value standby;

The time-domain response of tentation data transmission channel is:

h＝[h ₀,h ₁,...,h _L-1]

Wherein, L is the number that channel multi-path is propagated, i the multipath composition that hi is channel;

The channel frequency domain response estimation of corresponding k data symbol is:

$H\left(k\right)=\underset{l=0}{\overset{L-1}{\mathrm{\&Sigma;}}}{\hat{h}}_l\exp (-j\frac{2\mathrm{\&pi;}}{K}\mathrm{kl}),k=\mathrm{0,1},...,K-1$

Wherein, K is the length of data symbol.

Step 4: step 2 is obtained carry out forward direction filtering operation, filtering result is designated as will obtain with step 6 make it poor;

When carrying out step 4 first, step 6 is not inputted data and can be utilized, and therefore, step 6 is output as zero,

Step 5: step 4 is obtained carry out inverse discrete Fourier transformer inverse-discrete IDFT, obtain time domain output data will send into Qc-LDPC decoder, carry out soft decoding processing, the likelihood ratio { L (d of decoder output encoder bit _k); If reach the iterations of regulation, to likelihood ratio { L (d _k) carry out 0,1 bit decision, recover original transmitted information bit otherwise, continue step 6, carry out iterative data processing;

Ldpc decoder adopts and amasss BP soft decoding algorithm, BP algorithm referring to [Xiao Yang .Turbo and LDPC encoding and decoding and application [M] thereof. People's Telecon Publishing House, 2010:78-80].

The likelihood ratio of coded-bit { L (d _k) carry out the method for 0,1 bit decision: if L is (d _k) > 1, sentence information bit otherwise, sentence information bit

Step 6: { L (the d that step 5 is obtained _k) carry out the soft estimation of frequency domain transmitting symbol, obtain frequency domain symbol estimated value will the channel estimating obtaining with step 3 carry out feedback filtering, filtering result is designated as upgrade filter coefficient A simultaneously _kand B _k; By the filter coefficient A after upgrading _kand B _ksend into step 4 and carry out interative computation;

In step 6, symbol estimator completes the soft estimation to frequency domain transmitting symbol, obtains estimated value method of estimation is:

${\hat{S}}_k=\frac{1}{K}\underset{i=0}{\overset{K-1}{\mathrm{\&Sigma;}}}E\left({\hat{x}}_k\right)\exp (-j\frac{2\mathrm{\&pi;ki}}{K})$

Wherein, be the average (estimated value) of time domain transmitting symbol, its computational methods are:

Wherein, the set of QPSK sign map, represent a value in assemble of symbol, expression time domain transmitting sign estimation value is got the probability of some symbols in assemble of symbol.

Probability computational methods as follows:

$P({\hat{x}}_k={\mathrm{\&alpha;}}_i)=\underset{l}{\mathrm{\&Pi;}}P\left(d_{k,l}\right),l=\mathrm{0,1}$

Wherein, P (d _k,l) representing that launching l coded-bit of symbol for k gets 0,1 probability, its value is used the likelihood ratio { L (d of the coded-bit of decoder output in step 5 _k) try to achieve, that is:

$P\left(d_{k,l}\right)=\frac{\exp \left(d_{k,l}L\left(d_k\right)\right)}{1+\exp \left(L\left(d_k\right)\right)}$

Forward-direction filter and the feedback filter coefficient update method of the frequency domain decisive feedback equalization device relating in step 6 are as follows:

The data symbol that single antenna is received is set up following equation at frequency domain:

Wherein, S _d={ S ₁... S _kthe DFT conversion of transmitted data symbols, V={v ₀..., v _k-1the DFT conversion that channel adds noise, K is the length of data symbol, the diagonalization of Diag () representing matrix.

So, to k, receive data symbol S _kthere is following relation:

Z _d,k＝H(k)S _k+v _k

So, for N _r(N _r>=2) individual reception antenna, N _rk the frequency domain transmitted data symbols S that individual reception antenna receives _kthere is following relation:

$Z_{d,k}=H\left(k\right)S_k+V_k\&DoubleRightArrow;\left[\begin{array}{c}{Z}_{d,k}^1\\ .\\ .\\ .\\ Z_{d,k}^{N_r}\end{array}\right]=\left[\begin{array}{c}{H}_k^1\\ .\\ .\\ .\\ H_k^{N_r}\end{array}\right]S_k+\left[\begin{array}{c}{v}_k^1\\ .\\ .\\ .\\ v_k^{N_r}\end{array}\right]$

Wherein, Z _d,kby N _rthe data vector that the DFT conversion of k the data symbol that individual reception antenna receives forms, H (k) is by N _rthe channel response vector that the channel frequency domain response of k the data symbol that individual reception antenna receives forms, V _kit is the DFT conversion that channel adds noise.

Frequency domain decisive feedback equalization device frequency domain evaluated error as follows:

$E=S_k-{\tilde{S}}_k=S_k-(A_k^H{Z}_{d,k}-B_k^H{\hat{S}}_k)=S_k-A_k^H{Z}_{d,k}+B_k^H{\hat{S}}_k$

Wherein, () ^hrepresenting matrix conjugate transpose.

So according to least mean-square error (MMSE) criterion, the forward-direction filter of frequency domain decisive feedback equalization device and feedback filter coefficient more new formula are:

$A_k=\mathrm{cov}(S_k,S_k){({\mathrm{\&sigma;}}_v^{2}I+{\hat{H}}_k\mathrm{cov}(S_k,S_k){\hat{H}}_k^H)}^{-1}{\hat{H}}_k^H$

$B_k=A_k{\hat{H}}_k^{H}E({\hat{S}}_k,S_k^H)-E({\hat{S}}_k,S_k)$

Wherein, noise variance, cov (S _k, S _k) be frequency domain transmitted data symbols auto-correlation, it is the cross-correlation of frequency domain transmitting symbol and its frequency domain estimate symbol.

Cov (S _k, S _k) computational methods as follows:

${\mathrm{\&sigma;}}_s^2=\mathrm{cov}(S_k,S_k)=E\left({{|S}_k|}^2\right)-{\left|E\left(S_k\right)\right|}^2$

Wherein, E (S _k) be the average of frequency domain transmitting symbol, E (| S _k| ²) be the average of frequency domain transmitting symbol square; Its computing formula is as follows:

$E\left(S_k\right)=\frac{1}{K}\underset{i=0}{\overset{K-1}{\mathrm{\&Sigma;}}}E\left({\hat{x}}_k\right)\exp (-j\frac{2\mathrm{\&pi;ki}}{K})$

$E\left({\left|S_k\right|}^2\right)=\frac{1}{K}\underset{i=0}{\overset{K-1}{\mathrm{\&Sigma;}}}E\left({\left|{\hat{x}}_k\right|}^2\right)\exp (-j\frac{2\mathrm{\&pi;ki}}{K})$

Wherein, the average of time domain transmitting sign estimation, it is the average of time domain transmitting sign estimation square; Its computing formula is as follows:

computational methods as follows:

So, the forward-direction filter A of frequency domain decisive feedback equalization device _kwith feedback filter coefficient B _kmore new formula can be reduced to:

$A_k={\mathrm{\&sigma;}}_s^2{({\mathrm{\&sigma;}}_v^{2}I+{\hat{H}}_k{\mathrm{\&sigma;}}_s^2{\hat{H}}_k^H)}^{-1}{\hat{H}}_k^H;$

$B_k=A_k{\hat{H}}_k^H{\mathrm{\&gamma;}}_{\widehat{S,S}}-{\mathrm{\&gamma;}}_{\hat{S},S}$

Embodiment: shallow underwater sound communication system, 50 meters of the depth of waters, Qc-LDPC code code rate 1/2, Frame frame head length is 128 bits, and information data length is 600 bits, and underwater acoustic channel model is for mixing sound velocity gradient channel.

While adopting the inventive method, bit error rate curve comparison as shown in Figure 3-Figure 5.As seen from Figure 3, do not compare with there is no the situation of Joint iteration, use Joint iteration processing method of the present invention, 2 times iteration can obtain the snr gain that is greater than 5dB; As seen from Figure 4, Joint iteration number of times balanced and decoding is more, and systematic function is better; Fig. 5 further illustrates, and uses more reception antenna, can obtain more space diversity gain, thereby obtains more snr gain, promotes the receptivity of communication system, obtains the more transfer of data of low error rate.

Claims (1)

1. a communication system iteration receiving method, is characterized in that comprising the steps:

Step 1: transmitting terminal is first to binary message sequence { d _ncarry out Qc-LDPC coding, then to the information sequence { c after coding _ncarry out QPSK sign map; Data symbol front end after mapping adds Zadoff-Chu sequence as Frame frame head, forms complete frame data; Between every frame data by the full remainder according to connection, form continuous transmitting data { x _n;

Step 2: at receiving terminal, pass through N _rthe transmitting data of the antenna array receiving step one that individual reception antenna forms, the signal that each reception antenna is received is done as follows: process before each frame data, first remove the full remainder certificate between every frame data, then Frame frame head and data symbol are carried out respectively to discrete Fourier transform DFT, obtain frequency domain frame head with frequency domain data symbol n _r>=2;

Step 3: get any one reception antenna frequency domain frame head that step 2 obtains carry out channel estimating, obtain channel estimation value k _trepresent frame head length, k represents k symbol;

Step 4: step 2 is obtained carry out forward direction filtering operation, obtain wherein, A _kfor forward-direction filter coefficient, filtering result is designated as will obtain with step 6 make it poor;

When carrying out step 4 first,

Step 5: step 4 is obtained carry out inverse discrete Fourier transformer inverse-discrete IDFT, obtain time domain output data will send into Qc-LDPC decoder, the likelihood ratio { L (d of decoder output encoder bit _k); If reach the iterations of regulation, to likelihood ratio { L (d _k) carry out 0,1 bit decision, recover original transmitted information bit otherwise, enter step 6, carry out iterative data processing;

The likelihood ratio of coded-bit { L (d _k) carry out the method for 0,1 bit decision: if L is (d _k) > 1, sentence information bit otherwise, sentence information bit

Step 6: { L (the d that step 5 is obtained _k) carry out the estimation of frequency domain transmitting symbol, obtain frequency domain symbol estimated value will the channel estimating obtaining with step 3 carry out feedback filtering, b _kfor inverse filter coefficient, filtering result is designated as upgrade filter coefficient A simultaneously _kand B _k; By the filter coefficient A after upgrading _kand B _ksend into step 4 and carry out interative computation;

The method of estimation of described frequency domain transmitting symbol is wherein, represent transmitting symbol;

Filter coefficient A _kand B _kupdate method: $A_k={\mathrm{\&sigma;}}_s^2{({\mathrm{\&sigma;}}_v^{2}I+{\hat{H}}_k{\mathrm{\&sigma;}}_s^2{\hat{H}}_k^H)}^{-1}{\hat{H}}_k^H;$ $B_k=A_k{\hat{H}}_k^H{\mathrm{\&gamma;}}_{\widehat{S,S}}-{\mathrm{\&gamma;}}_{\hat{S},S}.$ Filter coefficient A _kinitial value be made as complete 1, wherein, noise variance, the oblique variance of frequency domain transmitted data symbols, it is the correlation function of time domain transmitting symbol and estimate symbol.