CN101119177B - Bit-symbol signal processing method for coherent communication machine - Google Patents

Abstract

The present invention discloses a bit-symbol signal processing method which is applicable for a self-adapting adjudication feedback equalizer and decoder connection and iteration signal process, comprising the following steps: 1 confirm the expression which is the most corresponding to the posterior probability of the signal of the communication receiving; 2 according to SOVA arithmetic, make the highest expression value of the process 1, thereby the posterior probability of the signal of the communication receiving is the highest and the flexible adjudication of the signal is acquired; 3 according to SOVA arithmetic, make the highest expression value of the process 1, thereby the posterior probability of the signal of the communication receiving is the highest and the rigid adjudication is acquired.

Description

The bit-symbol signal processing method that a kind of coherent communication machine is used

Technical field

The present invention relates to signal processing technology, particularly a kind of bit-symbol signal processing method that is applicable to that adaptive decision feedback equalizer and decoder cascade, iteration signal are handled.

Background technology

Underwater acoustic channel is to become and the channel of chromatic dispersion in way more than, time.Because many ways effect, ping of transmitting terminal emission can receive a plurality of pulses at receiving terminal, has produced the time delay diffusion.Again because carrier movement, and the motion of channel interface and medium, multiple motion makes these pulses produce Doppler frequency shift, and has produced Doppler's diffusion.The velocity of sound in the water is being about 1500m/s, and above-mentioned these movement velocitys are compared with the velocity of sound and can not be ignored, and it is quite serious that this spreads Doppler.For above-mentioned reasons, underwater acoustic channel often is called time delay and Doppler's double diffusion channel.And, the characteristic of underwater acoustic channel in time, the variation that do not coexist in place, its effective transmission bandwidth is limited, these all make in underwater acoustic channel transmission information difficulty more.

For in underwater acoustic channel high speed transmission information, generally all adopt the underwater sound coherent communication technology, at transmitting terminal emission multiple phase shift keying (MPSK) or many quadrature amplitude modulation (MQAN) signal, their bandwidth availability ratio height.Adopt the decoder of adaptive equalizer or code signal to overcome the influence of underwater acoustic channel at receiving terminal, still, adopt wherein a kind of technology all to be difficult to reach the good result that overcomes the underwater acoustic channel influence separately.Closely over several years, people begin adaptive equalizer and decoder cascade, carry out iteration signal and handle computing, the output symbol information of adaptive equalizer is sent into decoder, the output bit information feedback input adaptive equalizer of decoder, so iteration, after reaching a certain standard, stop iteration.Obtained United States Patent (USP) " Iterative decision feedback adaptive equalizer " on November 16th, 2004 such as people such as F.A.Blackman, the patent No. is the article " Low complexity iterative decisionfeedback equalizer for 8PSK modulation in time dispersive channel " that people such as US 6819630B1 and M.Marandian deliver calendar year 2001, IEEEInternational Symposium, vol.1.30 sept.-3 oct.2001.pp A102-A106 " in disclosed technology.Disclose decision-feedback adaptive equalizer and decoder cascade in the above-mentioned document, carried out interative computation, how the bit information of decoder output has not been converted to the symbolic information that the decision-feedback adaptive equalizer needs but mention.

In fact, the bit information of decoder output must convert symbolic information to when carrying out above-mentioned interative computation, because adaptive equalizer only carries out computing to symbolic information, in actual applications, just exist bit information to be converted to the signal processing problems of symbolic information.

In the prior art, the key technology of underwater sound coherent communication system is decision-feedback adaptive equalizer and Turbo-TCM (Trellis Coded Modulation) decoder cascade, the row iteration computing of going forward side by side.The soft probabilistic information that is output as systematic bits of Turbo-TCM decoder, generate the estimated value of transmission symbol by it, existing technology is the hard decision recompile to the systematic bits probabilistic information, this coding method is known, such as document " Proakis; J.G.; Digital Communications; Beijing; Publishing House ofElectronics Industry; 2001 " if in disclosed technology. the output of decoder reaches zero bit error rate, and then the method for this recompile is functional, and is easy to realize. still, in the practical application, error bit usually appears in the output of decoder, and when recompile, these error bit information can make that state transition path changes in the grid chart, generate numerous and emission symbol different symbol sebolic addressing, thereby when equalizer and decoder associating iteration, produce a large amount of erroneous transmissions. therefore, though the signal processing method that bit information is converted to symbolic information in the prior art relatively is easy to realize poor-performing.

In sum, because the deficiency that prior art exists, the good bit information with decoder output that just needs a kind of performance converts the signal processing method of the symbolic information that the decision-feedback adaptive equalizer needs to.

Summary of the invention

The objective of the invention is to overcome the deficiencies in the prior art, the bit-symbol signal processing method that provides a kind of coherent communication machine to use.

In order to achieve the above object, the present invention takes following technical scheme.

The bit-symbol signal processing method that a kind of coherent communication machine is used comprises the steps:

1) determine that the corresponding expression formula of the posterior probability maximum of the symbol that receives with communication equipment is as follows:

$\underset{t=1}{\overset{T}{\mathrm{\Σ}}}{\log p}_t\left(s_t\right)-\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\{{(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2\}$

P wherein _t(s _t) be prior probability, r _{T, I}And r _{T, Q}Be respectively the t imaginary part and the real part of receiving symbol constantly, s _{T, I} ⁱAnd s _{T, Q} ⁱBe respectively the imaginary part and the real part of respective signal in the planisphere;

2) according to the SOVA algorithm, make the transition formula evaluation maximum of step 1), thereby make the posterior probability maximum of the symbol of communication equipment reception, obtain the soft-decision of symbol;

3) according to the SOVA algorithm, make the transition formula evaluation maximum of step 1), thereby make the posterior probability maximum of the symbol of communication equipment reception, obtain the hard decision of symbol.

Further, described step 2) specifically comprise the steps:

(1), determines to shift in the component convolutional codes grid chart branch metric of road warp according to the SOVA algorithm;

(2) determine path metric;

(3) for the MPSK modulation system, determine the metric of survival route, determine the metric of the contended path corresponding with signal in the planisphere;

(4) determine that systematic bits is respectively 0 and 1 prior probability;

(5) determine t receiving symbol r constantly _tAnd the Euclidean distance in the planisphere between each point;

(6),, get the t probability of check digit constantly according to planisphere by previous step rapid (5);

(7) obtain the probability of each signal in the planisphere, become the soft-decision of symbol, the soft iteration signal that is used for decision-feedback adaptive equalizer and decoder is handled.

Further, step 3) specifically comprises the steps:

(1), determines to shift in the component convolutional codes grid chart branch metric of road warp according to the SOVA algorithm;

(2) determine path metric;

(3) for the MPSK modulation system, determine the metric of survival route, determine the metric of the contended path corresponding with signal in the planisphere;

(4) determine that systematic bits is respectively 0 and 1 prior probability;

(5) determine t receiving symbol r constantly _tAnd the Euclidean distance in the planisphere between each point;

(6),, get the t probability of check digit constantly according to planisphere by step 5);

(7) adjudicate the symbol that t is transmitted constantly, if Λ is (s _t ⁱ) minimum, then the signal of this moment transmission is s ⁱ, the hard decision of formation symbol.

The bit-symbol signal processing method that a kind of coherent communication machine is used comprises the steps:

1) determine that the maximum corresponding expression formula of posterior probability of the symbol that receives with communication equipment is as follows, make the posterior probability maximum of the symbol of communication equipment reception by this expression formula:

$\underset{t=1}{\overset{T}{\mathrm{\Σ}}}{\log p}_t\left(s_t\right)-\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\{{(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2\}---\left(5\right)$

R wherein _{T, I}And r _{T, Q}Be respectively the t imaginary part and the real part of receiving symbol constantly;

2), determine that the branch metric that shifts the road warp in the component convolutional codes grid chart is according to SOVA (Soft Output Viterbi Algorithm) algorithm:

$v_t^{s_t}={(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2-{\log p}_t\left(s_t\right)---\left(6\right)$

3), determine that path metric is for transfer path X:

${\mathrm{\μ}}_t^X=\underset{t^{\′}=1}{\overset{t}{\mathrm{\Σ}}}{v}_{t^{\′}}^{s_{t^{\′}}}={\mathrm{\μ}}_{t-1}^X+v_t^{s_t}---\left(7\right)$

4) for the MPSK modulation system, all corresponding M-1 bar of survival route contended path at any one time, determine that the metric of survival route is:

${\mathrm{\μ}}_{T,\min }=\left(\underset{t^{\′}=1}{\overset{t-1}{\mathrm{\Σ}}}{v}_{t^{\′}}^{s_{t^{\′}}^0}\right)+v_t^{s_t^0}+\left(\underset{t^{\′}=t+1}{\overset{T}{\mathrm{\Σ}}}{v}_t^{s_{t^{\′}}^0}\right)={\mathrm{\Λ}}^{\′}\left(s_t^0\right)---\left(9a\right)$

Determine and signal s ¹, s ²... s ^M-1The metric of corresponding contended path is:

${\mathrm{\μ}}_{t,s^i}=\min \{{\mathrm{\μ}}_{t-1}^f\left(l_{s^i}^{\′}\right)+v_t^{s^i}(l_{s^i}^{\′},l_{s^i})+{\mathrm{\μ}}_t^b\left(l_{s^i}\right)\}={\mathrm{\Λ}}^{\′}\left(s_t^i\right)---\left(9b\right)$

Wherein, t is constantly corresponding with survival route is signal s in the planisphere ⁰, M-1 bar contended path is distinguished respective signal s at this moment ¹, s ²... s ^M-1,

I=1,2 ... M-1 is and signal s ¹, s ²... s ^M-1The metric of corresponding contended path; And

Represent the state of adjacent moment, M _sBe the status number in the grid chart, Be t-1 forward direction survival path metric value constantly,

Be the metric of the constantly reverse survival route of t,

For t generates signal s constantly ⁱThe state transitions branch metric;

5) probability of determining each signal in the planisphere is:

$p({\hat{s}}_t^i=s_t^i)={\mathrm{\Λ}}^{\′}\left(s_t^i\right)/\underset{i=0}{\overset{M-1}{\mathrm{\Σ}}}{\mathrm{\Λ}}^{\′}\left(s_t^i\right)i=\mathrm{0,1,2},......M-1---\left(10\right)$

6) determine systematic bits c _tBeing respectively 0 and 1 prior probability is:

$\left\{\begin{array}{c}p(c_t=1)=\frac{e^{\mathrm{\Λ}\left(c_t\right)}}{1+e^{\mathrm{\Λ}\left(c_t\right)}}\\ p(c_t=0)=\frac{1}{1+e^{\mathrm{\Λ}\left(c_t\right)}}\end{array}\right.---\left(12\right)$

Wherein, prior probability is estimated Λ (c _t) be the soft information of systematic bits probability of decoder output;

7) determine t receiving symbol r constantly _tAnd the Euclidean distance in the planisphere between each point is respectively:

8),, get the t probability of check digit constantly according to planisphere by step 7):

$\left\{\begin{array}{c}p(\mathrm{Check}=0)\\ p(\mathrm{Check}=1)=1-p(\mathrm{Check}=0)\end{array}\right.---\left(14\right)$

9) by the result of step 6), step 7) and step 8),, obtain p according to planisphere _t(s _t);

10) result by step 4), step 5) and step 9) obtains

Become the soft-decision of symbol, the soft iteration signal that is used for decision-feedback adaptive equalizer and decoder is handled.

In technique scheme, further, also comprise step 11) by step 4) and step 9), directly adjudicate the symbol that t is transmitted constantly, if Λ ' is (s _t ⁱ) minimum, then the signal of this moment transmission is s ⁱ, the hard decision of formation symbol, the hard iteration signal that is applied to decision-feedback adaptive equalizer and decoder is handled.

Compared with prior art, beneficial effect of the present invention is:

1) the present invention is based on SOVA (SOft Output Viterbi Algorithm) algorithm and realize to the bit information translation being the signal processing of symbolic information, functional, exist under the situation of error bit in decoder output, the present invention realizes error sign ratio that bit information is converted to symbolic information than two orders of magnitude of prior art performance, and this will obviously improve the performance that decision-feedback adaptive equalizer and decoder cascade, iteration signal are handled.

2) signal processing major part of the present invention is to adopt the SOVA algorithm, and the SOVA algorithm is widely used in the decoder, so the present invention is applied widely.

Description of drawings

Fig. 1 is a bit-symbol signal processing method flow chart of the present invention;

Fig. 2 is the schematic diagram that bit-symbol signal processing method of the present invention is used in coherent communication.

Embodiment

Below in conjunction with the drawings and specific embodiments the present invention is described in further detail:

The signal processing method that the bit information that the present invention exports decoder converts the symbolic information of decision-feedback adaptive equalizer needs to can be described as " bit-signal converter " module in application.The present invention adopts the method based on SOVA (SOft Output Viterbi Algorithm) algorithm according to the systematic bits probabilistic information of Turbo-TCM decoder output, can be in the hope of the estimated value of receiver receiving symbol by it.The present invention usually makes mismark seldom, function admirable.Computational complexity involved in the present invention is the SOVA algorithm roughly the same, and is applied widely in actual applications.

Usually adopt the SOVA algorithm in the Turbo decoder, at this moment, also should adopt the SOVA algorithm in bit-symbol transition.And the absolute value of the systematic bits probability Estimation (claiming the systematic bits log-likelihood ratio again) that obtained by the SOVA algorithm of decoder is not too big, spillover can not take place when carrying out exponent arithmetic.

For the convenience of describing, be that example is introduced bit-symbol signal processing method of the present invention in detail below with the qpsk modulation signal.Certainly, the present invention also is applicable to all MPSK modulation signals.

The posterior probability of the symbol that the coherent communication machine receiving terminal receives is

$p_r(S/r_1^T)=\frac{p_r(S,r_1^T)}{p_r\left(r_1^T\right)}---\left(1\right)$

Try to achieve thus

$p_r(S,r_1^T)=p_r\left(S\right)\·p_r(r_1^T/S)---\left(2\right)$

S={s wherein ₁, s ₂..., s _T, the burst that representative is asked, r ₁ ^TRepresent equalizer to be defeated by the symbol sebolic addressing from the moment 1 to moment T of decoder.Make the posterior probability maximum,, only need p for one group of equiprobability burst _r(S, r ₁ ^T) maximum.Logarithm is asked on formula (2) both sides, obtained

${\log p}_r(S,r_1^T)={\log p}_r\left(S\right)+{\log p}_r(r_1^T/S)---\left(3\right)$

The noise of supposing transmission channel is an independent Gaussian noise, and then formula (3) is expressed as

${\log p}_r(S,r_1^T)=\underset{t=1}{\overset{T}{\mathrm{\Σ}}}{\log p}_t\left(s_t\right)-\frac{\mathrm{nT}}{2}\log \left(2\mathrm{\π}\right)-n\log \mathrm{\σ}-\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\frac{{(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2}{2{\mathrm{\σ}}^2}---\left(4\right)$

R wherein _{T, I}And r _{T, Q}Be respectively the t imaginary part and the real part of receiving symbol constantly, s _{T, I} ⁱAnd s _{T, Q} ⁱBe respectively the imaginary part and the real part of respective signal in the planisphere, i=0,1,2,3.p _t(s _t) be prior probability.N is the length of component code.Signal corresponding in the symbol of arbitrary given time and the planisphere is different, then will make the posterior probability maximization, only needs the maximization following formula:

$\underset{t=1}{\overset{T}{\mathrm{\Σ}}}{\log p}_t\left(s_t\right)-\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\{{(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2\}---\left(5\right)$

According to the SOVA algorithm, the branch metric of transfer path is in the definition component convolutional codes grid chart

$v_t^{s_t}={(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2-{\log p}_t\left(s_t\right)---\left(6\right)$

For transfer path X, the definition path metric is

${\mathrm{\μ}}_t^X=\underset{t^{\′}=1}{\overset{t}{\mathrm{\Σ}}}{v}_{t^{\′}}^{s_{t^{\′}}}={\mathrm{\μ}}_{t-1}^X+v_t^{s_t}---\left(7\right)$

For the QPSK modulation system, all corresponding 3 contended path of survival route at any one time, suppose the t moment corresponding with survival route be signal s in the planisphere ⁰, then three contended path are distinguished respective signal s at this moment ¹, s ²And s ³Definition μ _{T, min}The metric in path for survival,

I=1,2,3 are and signal s ¹, s ², s ³The metric of corresponding contended path, the path metric that can be obtained correspondence by (7) formula is

$\left\{\begin{array}{c}{\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^0\right)={\mathrm{\&mu;}}_{T,\min }\\ {\mathrm{\&Lambda;}}^{\&prime;}\left(s^1\right)={\mathrm{\&mu;}}_{t,{\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2006100891374E00011.png"\; state="\{\{state\}\}">s</figure-callout>}}^1}\\ {\mathrm{\&Lambda;}}^{\&prime;}\left(s^2\right)={\mathrm{\&mu;}}_{t,s^2}\\ {\mathrm{\&Lambda;}}^{\&prime;}\left(s^3\right)={\mathrm{\&mu;}}_{t,s^3}\end{array}\right.---\left(8\right)$

Each amount is by following formulate in the following formula

${\mathrm{\&mu;}}_{T,\min }=\left(\underset{t^{\&prime;}=1}{\overset{t-1}{\mathrm{\&Sigma;}}}{v}_{t^{\&prime;}}^{s_{t^{\&prime;}}^0}\right)+v_t^{s_t^0}+\left(\underset{t^{\&prime;}=t+1}{\overset{T}{\mathrm{\&Sigma;}}}{v}_t^{s_{t^{\&prime;}}^0}\right)---\left(9a\right)$

${\mathrm{\&mu;}}_{t,s^i}=\min \{{\mathrm{\&mu;}}_{t-1}^f\left(l_{s^i}^{\&prime;}\right)+v_t^{s^i}(l_{s^i}^{\&prime;},l_{s^i})+{\mathrm{\&mu;}}_t^b\left(l_{s^i}\right)\}---\left(9b\right)$

Wherein

Represent the state of adjacent moment, M _sBe the status number in the grid chart.In the grid chart contended path at t-1 constantly from state

I=1,2,3, transfer to state

I=1,2,3, generate signal s ⁱ, i=1,2,3,

Be t-1 forward direction survival path metric value constantly,

Be the metric of the constantly reverse survival route of t,

For t generates signal s constantly ⁱThe state transitions branch metric.The probability of each signal in the planisphere

For

$p({\hat{s}}_t^i=s^i)={\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^i\right)/({\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^0\right)+{\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^1\right)+{\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^2\right)+{\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^3\right)),i=\mathrm{0,1,2,3}---\left(10\right)$

Because used the prior probability of symbol in the formula (6), the probability of the sign check position that should be received by the systematic bits probability and the present receiving machine of decoder output is tried to achieve the prior probability of each point in the planisphere. the soft information of systematic bits probability of decoder output is that prior probability is estimated Λ (c _t), be expressed from the next

$\mathrm{\&Lambda;}\left(c_t\right)=\log \frac{p(c_t=1)}{p(c_t=0)}---\left(11\right)$

Then obtain systematic bits c _tBeing respectively 0 and 1 prior probability is

$\left\{\begin{array}{c}p(c_t=1)=\frac{e^{\mathrm{\&Lambda;}\left(c_t\right)}}{1+e^{\mathrm{\&Lambda;}\left(c_t\right)}}\\ p(c_t=0)=\frac{1}{1+e^{\mathrm{\&Lambda;}\left(c_t\right)}}\end{array}\right.---\left(12\right)$

Λ (the c that the SOVA algorithm obtains _t) absolute value little, its index can not overflow.Λ (the c that the MAP algorithm obtains _t) absolute value bigger, its index can overflow.If the t receiving symbol is r constantly _t, then the Euclidean distance between each signal is respectively in this symbol and the planisphere

$\left\{\begin{array}{c}L\left(s_t^0\right)={(r_{t,I}-s_{t,I}^0)}^2+{(r_{t,Q}-s_{t,Q}^0)}^2\\ L\left(s_t^1\right)={(r_{t,I}-s_{t,I}^1)}^2+{(r_{t,Q}-s_{t,Q}^1)}^2\\ L\left(s_t^2\right)={(r_{t,I}-s_{t,I}^2)}^2+{(r_{t,Q}-s_{t,Q}^2)}^2\\ L\left(s_t^3\right)={(r_{t,I}-s_{t,I}^3)}^2+{(r_{t,q}-s_{t,Q}^3)}^2\end{array}\right.---\left(13\right)$

Then the probability of t moment check digit is respectively

$\left\{\begin{array}{c}p(\mathrm{Check}=0)=[e^{-L\left(s_t^0\right)}+e^{-L\left(s_t^1\right)}]/[e^{-L\left(s_t^0\right)}+e^{-L\left(s_t^1\right)}+e^{-L\left(s_t^2\right)}+e^{-L\left(s_t^3\right)}]\\ p(\mathrm{Check}=1)=1-p(\mathrm{Check}=0)\end{array}\right.---\left(14\right)$

Can think that each bit is separate in each code word, so current transmission symbol is that the prior probability of each point in the planisphere is respectively

$\left\{\begin{array}{c}p(s_t^0=s^0)=p(c_t=0)\&CenterDot;p(\mathrm{Check}=0)\\ p(s_{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2006100891374E00011.png"\; state="\{\{state\}\}">t</figure-callout>}}^1=s^1)=p(c_t=0)\&CenterDot;p(\mathrm{Check}=1)\\ p(s_t^2=s^2)=p(c_t=1)\&CenterDot;p(\mathrm{Check}=0)\\ p(s_{\mathrm{<figure-callout\; id="3"\; label="t"\; filenames="H2006100891374E00011.png"\; state="\{\{state\}\}">t</figure-callout>}}^3=s^3)=p(c_t=1)\&CenterDot;p(\mathrm{Check}=1)\end{array}\right.---\left(15\right)$

So just obtained symbol prior probability p required in the formula (6) _t(s _t).

Formula (6) to formula (15) has provided the algorithm of bit-symbol transition signal processing.

Try to achieve easily by formula (8), (9), (10) and (15)

Become the soft-decision of symbol, also can directly adjudicate the symbol that t is transmitted constantly, if Λ ' is (s according to formula (8), (9) and (15) _t ⁱ) minimum, then the signal of this moment transmission is s ⁱ, the hard decision of formation symbol.Above-mentioned symbol soft-decision and hard decision are applied to the soft hard iterative algorithm of decision-feedback adaptive equalizer and decoder respectively.

The present invention is suitable for the MPSK modulation signal, describes from signal processing flow of the present invention below, and be that example illustrates with QPSK as shown in Figure 1.The soft information of decoder output is that prior probability is estimated Λ (c _t), be input in the operation blocks 101 of bit-signal converter 100, get p (c with formula (12) _t).The data flow r that receiver receives ₁ ^TBe input in the operation blocks 104, get L (s by formula (13) _t ⁱ).L (s _t ⁱ) being input to must p (Check) in the operation blocks 103.r ₁ ^T, p (c _t) and p (Check) be input in the operation blocks 102, get p (s with (15) formula _t).{ r _tAnd p (s _t) import in the operation blocks 105, get by formula (6)

In the input operation blocks 106, get μ with formula (7) _t ^xμ _t ^xSend into operation blocks 107, with formula (9a) and (9b) must μ _{T, min}With

Obtain Λ ' (s thus _t ⁱ), be used for the hard iterative algorithm of decision-feedback adaptive equalizer and decoder.μ _{T, min}With

Input operation blocks 108 is got by (10) formula

The soft iterative algorithm that is used for decision-feedback adaptive equalizer and decoder. operation blocks 105,106,107 and 108 has been formed operation blocks 109, it is the SOVA algorithm. need to prove, above-mentioned operation blocks both can realize with software, also can realize with circuit devcie. the present invention can be applied to the MQAM modulation signal.

Relatively obtained the performance of two kinds of methods of transmitted symbol streams below by the decoder output bit flow, a kind of is bit-signal converter output, and another kind of is that recompile is exported.With the component code generator polynomial is that { 23,35, the 8PSK Turbo-TCM code signal of 33} is tested, and frame length is 1936, and output adds σ to coding ²=0.07 white Gaussian noise, decoder iterations are 0 (not carrying out interative computation, directly output).Table 1 has provided the output result of 10 frame data of selecting arbitrarily.Can see that when the bit error rate of decoder output was 0, the performance of recompile output was better than bit-signal converter slightly, the bit of this moment-signal converter output may have a little several mismark, and the error sign ratio of recompile output is 0.In case there is error bit in decoder output, it is big that the error sign ratio of recompile output sharply becomes, and bit-signal converter output is far better, and the latter is than two orders of magnitude of the former performance.This is because when there is error bit in decoder output, if with these bit recompiles, the state transition path of encoder is made a mistake, thereby cause numerous mismarks.The result was similar when output added different white noises to coding.

Table 1 is obtained two kinds of methods of transmitted symbol streams by the decoder output bit flow performance compares

Frame number	Decoding output bit error rate	Bit-signal converter output error sign ratio	Recompile output error sign ratio
				1	0.00077	0.01085	0.26033
2	0.00387	0.00981	0.19421
				3	0.00646	0.02273	0.25362
4	0.00000	0.00413	0.00000
				5	0.00000	0.00568	0.00000
6	0.00310	0.01188	0.25981
				7	0.00207	0.01446	0.22521
8	0.00077	0.00878	0.18957
				9	0.00129	0.00723	0.17252
10	0.00336	0.00981	0.24948

Below in conjunction with Fig. 2 the application of the present invention in adaptive decision feedback equalizer and decoder cascade described.See Fig. 2.The data flow r of equalizer output ₁ ^TBe divided into two-way, the one tunnel directly imports first bit-signal converter 201, and another road is through interleaver 203 input second bit-signal converters 204.The output prior probability of decoder 200 is estimated Λ (c _t) be divided into two-way, one the tunnel through deinterleaver 202 input first bit-signal converters 201, another road is imported the output and the second bit-signal converter 204 of second bit-signal converter 204. first bits-signal converter 201 and is deleted computing through the output process of deinterleaver 205, gets Λ ' (s _t ⁱ) or

Behind really up to the mark or soft-decision 206, obtain desired sign estimation

It should be noted last that above embodiment is only unrestricted in order to technical scheme of the present invention to be described.Although the present invention is had been described in detail with reference to embodiment, those of ordinary skill in the art is to be understood that, technical scheme of the present invention is made amendment or is equal to replacement, the spirit and scope that do not break away from technical solution of the present invention, it all should be encompassed in the middle of the claim scope of the present invention.

Claims (2)

1. the bit-symbol signal processing method that coherent communication machine is used comprises the steps:

1) determine that the corresponding expression formula of the posterior probability maximum of the symbol that receives with communication equipment is as follows:

$\underset{t=1}{\overset{T}{\mathrm{\&Sigma;}}}{\log p}_t\left(s_t\right)-\underset{t=1}{\overset{T}{\mathrm{\&Sigma;}}}\{{(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2\}$

P wherein _t(s _t) be prior probability, r _{T, I}And r _{T, Q}Be respectively the t imaginary part and the real part of receiving symbol constantly, s _{T, I} ⁱAnd s _{T, Q} ⁱBe respectively the imaginary part and the real part of respective signal in the planisphere;

2), determine that the branch metric that shifts the road warp in the component convolutional codes grid chart is according to the SOVA algorithm:

$v_t^{s_t}={(r_{t,I}-s_{t,I}^i)}^2+{(r_{t,Q}-s_{t,Q}^i)}^2-{\log p}_t\left(s_t\right)$

3), determine that path metric is for transfer path X:

${\mathrm{\&mu;}}_t^X=\underset{t^{\&prime;}=1}{\overset{t}{\mathrm{\&Sigma;}}}{v}_{t^{\&prime;}}^{s_{t^{\&prime;}}}={\mathrm{\&mu;}}_{t-1}^X+v_t^{s_t}$

4) for the MPSK modulation system, all corresponding M-1 bar of survival route contended path at any one time, determine that the metric of survival route is:

${\mathrm{\&mu;}}_{T,\min }=\left(\underset{t^{\&prime;}=1}{\overset{t-1}{\mathrm{\&Sigma;}}}{v}_{t^{\&prime;}}^{s_{t^{\&prime;}}^0}\right)+v_t^{s_t^0}+\left(\underset{t^{\&prime;=t+1}}{\overset{T}{\mathrm{\&Sigma;}}}{v}_t^{s_{t^{\&prime;}}^0}\right)={\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^0\right)$

Determine and signal s ¹, s ²... s ^M-1The metric of corresponding contended path is:

${\mathrm{\&mu;}}_{{t,s}^i}=\min \{{\mathrm{\&mu;}}_{t-1}^f\left(l_{s^i}^{\&prime;}\right)+v_t^{s^i}(l_{s^i}^{\&prime;},l_{s^i})+{\mathrm{\&mu;}}_t^b\left(l_{s^i}\right)\}={\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^i\right)$

Wherein, t is constantly corresponding with survival route is signal s in the planisphere ⁰, M-1 bar contended path is distinguished respective signal s at this moment ¹, s ²... s ^M-1, For with signal s ¹, s ²... s ^M-1The metric of corresponding contended path; And

Represent the state of adjacent moment, M _SBe the status number in the grid chart,

Be t-1 forward direction survival path metric value constantly,

Be the metric of the constantly reverse survival route of t,

For t generates signal s constantly ⁱThe state transitions branch metric;

5) probability of determining each signal in the planisphere is:

$p({\hat{s}}_t^i=s_t^i)={\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^i\right)/\underset{i=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{\mathrm{\&Lambda;}}^{\&prime;}\left(s_t^i\right)i=\mathrm{0,1,2},......M-1$

6) determine systematic bits c _tBeing respectively 0 and 1 prior probability is:

$\left\{\begin{array}{c}p(c_t=1)=\frac{e^{\mathrm{\&Lambda;}\left(c_t\right)}}{1+e^{\mathrm{\&Lambda;}\left(C_t\right)}}\\ p(c_t=0)=\frac{1}{1+e^{\mathrm{\&Lambda;}\left(c_t\right)}}\end{array}\right.$

Wherein, prior probability is estimated Λ (c _t) be the soft information of systematic bits probability of decoder output;

7) determine t receiving symbol r constantly _tBe respectively with the Euclidean distance of each signal in the planisphere:

$\left\{\begin{array}{c}L\left(s_t^0\right)={(r_{t,I}-s_{t,I}^0)}^2+{(r_{t,Q}-s_{t,Q}^0)}^2\\ L\left(s_t^I\right)={(r_{t,I}-s_{t,I}^1)}^2+{(r_{t,Q}-s_{t,Q}^1)}^2\\ L\left(s_t^2\right)={(r_{t,I}-s_{t,I}^2)}^2+{(r_{t,Q}-s_{t,Q}^2)}^2\\ \begin{array}{ccc}.& .& .\end{array}\\ \begin{array}{ccc}.& .& .\end{array}\\ \begin{array}{ccc}.& .& .\end{array}\\ L\left(s_t^{M-1}\right)={(r_{t,I}-s_{t,I}^{M-1})}^2+{(r_{t,Q}-s_{t,Q}^{M-1})}^2\end{array}\right.$

8),, get the t probability of check digit constantly according to planisphere by step 7):

$\left\{\begin{array}{c}p(\mathrm{Check}=0)\\ p(\mathrm{Check}=1)=1-p(\mathrm{Check}=0)\end{array}\right.$

9) by the result of step 6), step 7) and step 8),, obtain p according to planisphere _t(s _t);

10) result by step 4), step 5) and step 9) obtains

Become the soft-decision of symbol, the soft iteration signal that is used for decision-feedback adaptive equalizer and decoder is handled.

2. the bit-symbol signal processing method of using according to the described coherent communication machine of claim 1 is characterized in that, also comprises:

Step 11): by described step 4) and described step 9), the signal that judgement t is transmitted constantly is if Λ ' is (s _t ⁱ) minimum, then the signal of this moment transmission is s ⁱ, form hard decision to symbol.

Images