CN101557364A - Joint-iterative channel estimation and decoding method of Turbo-OvCDM system - Google Patents

Abstract

The invention relates to a joint-iterative channel estimation and decoding method of a Turbo-OvCDM system. The method comprises the following steps: a small amount of pilot frequency symbols are inserted in the transmitted data by each frame, the channel impact response is estimated by using the known pilot frequency symbols and the unknown coded information and by an iteration method, the pilot frequency symbols are used for proving the original channel estimation information, and the system performs decoding to the received unknown information by using the obtained original channel estimation information; and the estimated information obtained by decoding is utilized as the new pilot frequency symbol required for the channel estimation during the next iteration, and the iteration process is repeated at all time until to meet the requirements. The invention is based on the scheme of combined receiving of the iterative channel estimation feedback by judgment and the decoding, and the simulation result shows that the method has high convergence rate, reduces the complexity of the system under the condition of high-order iteration and has the performance being superior to that of the ordinary channel estimation method.

Description

The associating iterative channel of Turbo-OvCDM system is estimated and interpretation method

Technical field

The present invention relates to the channel estimation methods in a kind of superimposed coding multiplexing (OvCDM) system, belong to the broadband wireless communication technique field.

Background technology

Superimposed coding multiplexing (multiple access) technology (Overlapped Code Division Multiplexing (Access), OvCDM (OvCDMA)) having increased substantially " waveform is cut apart multiple access " is CDMA (Code DivisionMultiple Access, CDMA) transmission reliability of system and spectrum efficiency, be a kind of brand-new spectral efficient coding multiplexing method and technology, made full use of the thought of overlapping multiplexing and the thought of coding, and both unifications have been combined.Utilize code check to be higher than 1 parallel-convolution and encode and increase substantially capacity of communication system and spectrum efficiency, when improving coding gain, also improved the spectrum efficiency of system.

Utilize the OvCDM code check to be higher than 1 character, not only only using binary phase shift keying modulation (Binary Phase ShiftKeying, BPSK) or quaternary phase-shift keying modulation (Quaternary Phase Shift Keying, QPSK) etc. just can obtain very high system spectral efficiency during the low-dimensional modulating data, after constraint length is very big, can also have the strong error correcting capability of approaching Xian Nongjie simultaneously.

The OvCDM coding techniques is the synthesis of a plurality of technology, and it can provide spectrum gain, coding gain and diversity gain.Spectrum efficiency is directly to be reflected on the code check, and coding gain shows as the Euclidean distance that improves the coding output sequence, and diversity gain is that Hamming distance by improving the coding output sequence is to obtain the implied diversity effect in Selective Fading Channel.

Combine OvCDM technology and Turbo iteration thought in the serially concatenated OvCDM system, by the OvCDM coding formation serially concatenated of two-stage.The log likelihood value of receiving terminal soft input and output of feedback between two-stage OvCDM coding carries out the iteration coding, improves performance.Because the decoding of this multi-stage cascade OvCDM system has been adopted and the similar decoding architecture of traditional Turbo code, therefore can be referred to as the Turbo-OvCDM system.The structure of Turbo-OvCDM system as shown in Figure 1.

The channel estimation technique of Turbo-OvCDM system is the technology of a very challenging property.After the OvCDM system proposes, channel estimation technique under this system has had certain research, but (Least Square Error, LSE) channel response that obtains the transmission channel at pilot frequency sequence place utilizes interpolation method to obtain the channel estimating of data division more generally all to be to use traditional channel estimating such as least-squares algorithm.This conventional channel method of estimation causes complexity too high in iterative process when being applied to the Turbo-OvCDM system, has reduced systematic function.

Summary of the invention

Add the not high complexity problem of higher of original channel estimation method performance in the Turbo-OvCDM system but the present invention is directed to, a kind of estimation of associating iterative channel and interpretation method that can improve systematic function and reduce the Turbo-OvCDM system of system complexity is provided.

The associating iterative channel of Turbo-OvCDM of the present invention system is estimated and interpretation method, be to send at each frame to insert a small amount of frequency pilot sign in the data, mode by iteration, utilize known pilot symbols and unknown coded message to come estimated channel impulse response, frequency pilot sign provides original channel estimating information, and the original channel estimated information that system's utilization obtains is deciphered the unknown signaling that receives; The required new frequency pilot sign of channel estimating when the estimated information that obtains through decoding is utilized as next iteration is till this iterative process is repeated until and satisfies condition; Specifically may further comprise the steps:

(1) iteration for the first time, sending in data according to data and frequency pilot sign at each frame is 20: 1 ratio insertion frequency pilot sign, utilize known pilot symbols and unknown coded message to adopt least mean-square error (Minimum Mean Square Error, MMSE) algorithm for estimating carries out initial channel estimation to channel gain, suppose that the frequency pilot sign vector for P, then gains corresponding to initial channel

Can try to achieve by following formula:

${\hat{H}}^{\left(0\right)}={(P^{*}P+{\mathrm{\σ}}_n^{2}I)}^{-1}{P}^{*}r$

Wherein, ^*The expression conjugation, σ _n ²The power of expression additive white Gaussian noise, r is a received signal information, I representation unit matrix; Utilize this initial estimate again

Carry out Turbo-OvCDM decoding; Supposing to send vector satisfies

Wherein, E[] represent the content in the bracket is averaged, ^*The expression conjugation, X[l] represent to send to quantity symbol, The transmission that expression is estimated to obtain is to quantity symbol, and the soft-decision of each symbol is exported in decoding, and adding frequency pilot sign becomes the known training symbol of hypothesis, the soft-decision of the symbol of input channel estimator, and then output information once more and the soft-decision of redundant symbol;

The inferior iteration of (2) n (n＞1), the required new frequency pilot sign of channel estimating when the estimated information that obtains through decoding is utilized as next iteration, the soft-decision of Turbo-OvCDM decoder output when channel estimator adopts last iteration, do following two steps and handle:

A. Nonlinear Processing: hard decision is done in the output to relevant all symbols of Turbo-OvCDM decoder, and imitation sends and through the process of channel, the receiving symbol sequence that obtains being similar to then Detailed process is as follows:

The purpose of Nonlinear Processing is in order to recover receiving sequence from the soft-decision output of decoder, to be used for carrying out channel estimating next time; Conventional letter is l at interval, and the soft-decision of decoder output is output as:

$p\left(\hat{X}\right[l]=x_m|r)$

Wherein, x _m∈ 1,2} be belong to binary phase-shift modulation (Binary Phase Shift Keying, the signal of set of signals BPSK) has obtained now at mark space l, each sends the probability of symbol, again because have

r[l]＝H[l]X[l]+n[l]

Wherein, r[l] be through received signal information behind the channel, X[l] for expression sends to quantity symbol, H[l] be channel impulse response, n[l] be additive white Gaussian noise, so approximate receiving sequence is

Probability p be:

$p\left(\hat{r}\right|r)=p\left(\hat{X}\right[l]=x_m|r)$

For the soft-decision from decoding is exported the receiving sequence that obtains being similar to

So approximate receiving symbol of l mark space For:

$\hat{r}\left[l\right]=H\hat{X}\left[l\right]$

In the following formula

Hard decision is done in the output of decoder to be obtained

$\hat{X}\left[l\right]=\arg \underset{m}{\max }p\left(\hat{r}\right|r)=\arg \underset{m}{\max }p\left(\hat{X}\right[l]=x_m|r)$

By top derivation as can be known, the process of Nonlinear Processing is exactly: hard decision is done in the output to relevant all symbols of Turbo-OvCDM decoder, and imitation sends and pass through the process of channel, the receiving symbol sequence that obtains being similar to then

B. insert frequency pilot sign;

After inserting frequency pilot sign, channel estimator carries out the feedback of decoder the channel estimating of next round as training sequence.

The present invention both can regard combined channel as and estimate and decoding scheme, can be regarded as a kind of suboptimum decoding algorithm of incoherent coding again, this is that the process of channel estimating also can be used as decode procedure and understands because channel condition information can be understood as a kind of state variable.

The present invention is based on the iterative channel estimation of decision-feedback and the associating reception programme of decoding, and by simulation result as can be seen, this method fast convergence rate under the situation of high order iteration, has reduced system complexity, and performance is better than the conventional channel method of estimation more.

Description of drawings

Fig. 1 is the structured flowchart of Turbo-OvCDM system.

Fig. 2 adds the structured flowchart of channel estimating for the Turbo-OvCDM system receiving terminal.

Fig. 3 uses the present invention to unite iterative channel estimation and decoding algorithm and conventional channel algorithm for estimating performance comparison figure down for Rayleigh slow fading (speed of a motor vehicle 30km/h) channel in the Turbo-OvCDM system.(solid line estimates and decoding algorithm that for the associating iterative channel dotted line is a linear interpolation algorithm among the figure)

Fig. 4 uses the present invention to unite iterative channel estimation and decoding algorithm and conventional channel algorithm for estimating performance comparison figure down for Rayleigh rapid fading (speed of a motor vehicle 250km/h) channel in the Turbo-OvCDM system.(solid line estimates and decoding algorithm that for the associating iterative channel dotted line is a secondary Gauss interpolation algorithm among the figure)

Embodiment

Iteration after the receiving terminal access channel is estimated receives structure as shown in Figure 2.The process of whole Turbo-OvCDM system channel estimation and decoding is as follows:

(1) iteration for the first time, the initial frequency pilot sign of channel estimator utilization, and the employing least mean-square error (Minimum MeanSquare Error, MMSE) algorithm for estimating carries out preliminary channel estimating to channel gain, suppose that the frequency pilot sign vector for P, then gains corresponding to initial channel

Can try to achieve by following formula:

${\hat{H}}^{\left(0\right)}={(P^{*}P+{\mathrm{\σ}}_n^{2}I)}^{-1}{P}^{*}r$

Wherein, ^*The expression conjugation, σ _n ²The power of expression additive white Gaussian noise, r is a received signal information, I representation unit matrix.Utilize this initial estimate again

Carry out Turbo-OvCDM decoding, the soft-decision of each symbol of decoding output, adding frequency pilot sign becomes the known training symbol of hypothesis, input channel estimator once more.It should be noted that in independent decoding, do not consider only to be concerned about information symbol under the condition of channel estimating that decoding only needs the soft decision information of output information symbol to get final product.And in combined channel estimation of the present invention and decoding scheme, because during next iteration, channel estimator needs complete symbol sebolic addressing as training sequence, so will make an amendment to original Turbo-OvCDM decoder, not only require the soft-decision output of information symbol, also will obtain the soft-decision output of redundant symbol.Ask the method and soft-decision and the indistinction of asking information symbol of the soft-decision output of redundant symbol.

When the inferior iteration of (2) n (n＞1), channel estimator adopt last iteration, the output of the soft-decision of Turbo-OvCDM decoder, do following two steps and handle:

The a Nonlinear Processing:

Conventional letter is l at interval, and the soft-decision of decoder output is output as:

$p\left(\hat{X}\right[l]=x_m|r)$

Wherein, x _m∈ 1,2} be belong to binary phase-shift modulation (Binary Phase Shift Keying, the BPSK) signal of set of signals has obtained now at mark space l, each sends the probability of symbol.Again because have

r[l]＝H[l]X[l]+n[l]

Wherein, r[l] be through received signal information behind the channel, X[l] be through input signal information before the channel, H[l] be channel impulse response, n[l] be additive white Gaussian noise, so approximate receiving sequence is Probability p be:

$p\left(\hat{r}\right|r)=p\left(\hat{X}\right[l]=x_m|r)$

For the soft-decision from decoding is exported the receiving sequence that obtains being similar to

So approximate receiving symbol of l mark space

For:

$\hat{r}\left[l\right]=H\hat{X}\left[l\right]$

In the following formula By being done hard decision, the output of decoder obtains

$\hat{X}\left[l\right]=\arg \underset{m}{\max }p\left(\hat{r}\right|r)=\arg \underset{m}{\max }p\left(\hat{X}\right[l]=x_m|r)$

B inserts frequency pilot sign:

After inserting frequency pilot sign, estimator carries out the next round channel estimating to the feedback of decoder as training sequence.

For Turbo-OvCDM code multiplexing system, the speed of a motor vehicle that the process channel estimating obtains is that the associating iterative channel under 30km/h and the 250km/h is estimated and decoding algorithm.Because the result in existing research shows that linear interpolation method was better when the speed of a motor vehicle was 30km/h, the quadratic interpolation method was better when the speed of a motor vehicle was 250km/h.When therefore comparing, Fig. 3 is when the speed of a motor vehicle is 30km/h, the comparison of associating iterative channel algorithm for estimating and linear interpolation algorithm, and Fig. 4 is when the speed of a motor vehicle is 250km/h, the comparison of associating iterative channel algorithm for estimating and quadratic interpolation algorithm.By simulation result as can be seen, this method fast convergence rate, under the situation of high order iteration, performance is better than the conventional channel method of estimation more.

Claims (2)

1. the associating iterative channel of a Turbo-OvCDM system is estimated and interpretation method, it is characterized in that:

Send a small amount of frequency pilot sign of insertion in the data at each frame, mode by iteration, utilize known pilot symbols and unknown coded message to come estimated channel impulse response, frequency pilot sign provides original channel estimating information, and the original channel estimated information that system's utilization obtains is deciphered the unknown signaling that receives; The required new frequency pilot sign of channel estimating when the estimated information that obtains through decoding is utilized as next iteration is till this iterative process is repeated until and satisfies condition; Specifically may further comprise the steps:

(1) iteration for the first time, sending in data according to data and frequency pilot sign at each frame is 20: 1 ratio insertion frequency pilot sign, utilize known pilot symbols and unknown coded message to adopt the least mean-square error algorithm for estimating that channel gain is carried out initial channel estimation, suppose that the frequency pilot sign vector for P, then gains corresponding to initial channel

Can try to achieve by following formula:

${\hat{H}}^{\left(0\right)}={(P^{*}P+{\mathrm{\σ}}_n^{2}I)}^{-1}{P}^{*}r$

Wherein, ^*The expression conjugation, σ _n ²The power of expression additive white Gaussian noise, r is a received signal information, I representation unit matrix; Utilize this channel estimation value again

Carry out Turbo-OvCDM decoding, suppose to send vector and satisfy $E\left[{\hat{X}}^{*}\right[l\left]X\right[l\left]\right]=1,$ Wherein, E[] represent the content in the bracket is averaged, ^*The expression conjugation, X[l] represent to send to quantity symbol,

The transmission that expression is estimated to obtain is to quantity symbol, and the soft-decision of each symbol is exported in decoding, and adding frequency pilot sign becomes the known training symbol of hypothesis, the soft-decision of the symbol of input channel estimator, and then output information once more and the soft-decision of redundant symbol;

The soft-decision output of Turbo-OvCDM decoder when (2) the n time iteration, the required new frequency pilot sign of channel estimating when the estimated information that obtains through decoding is utilized as next iteration, channel estimator adopt last iteration, do following two steps and handle:

A. Nonlinear Processing: hard decision is done in the output to relevant all symbols of Turbo-OvCDM decoder, and imitation sends and through the process of channel, the receiving symbol sequence that obtains being similar to then

B. insert frequency pilot sign;

After inserting frequency pilot sign, channel estimator carries out the feedback of decoder the channel estimating of next round as training sequence.

2. the associating iterative channel of Turbo-OvCDM according to claim 1 system estimates and interpretation method that it is characterized in that: the detailed process of described Nonlinear Processing is as follows:

The purpose of Nonlinear Processing is in order to recover receiving sequence from the soft-decision output of decoder, to be used for carrying out channel estimating next time; Conventional letter is l at interval, and the soft-decision of decoder output is output as:

$p\left(\hat{X}\right[l]=x_m|r)$

Wherein, x _m{ 1,2} is the signal that belongs to the set of signals of binary phase-shift modulation to ∈, has obtained now at mark space l, and each sends the probability of symbol, again because have

r[l]＝H[l]X[l]+n[l]

Wherein, r[l] for through received signal information behind the channel, X[l] for sending to quantity symbol H[l] be channel impulse response, n[l] be additive white Gaussian noise, therefore be similar to receiving sequence and be

Probability p be:

$p\left(\hat{r}\right|r)=p\left(\hat{X}\right[l]=x_m|r)$

For the soft-decision from decoding is exported the receiving sequence that obtains being similar to So approximate receiving symbol of l mark space

For:

$\hat{r}\left[l\right]=H\hat{X}\left[l\right]$

In the following formula Hard decision is done in the output of decoder to be obtained

$\hat{X}\left[l\right]=\arg \underset{m}{\max }p\left(\hat{r}\right|r)=\arg \underset{m}{\max }p\left(\hat{X}\right[l]=x_m|r).$

Images