CN104734814A - LDPC coding and decoding method applied in incoherent ultra-wideband communication system - Google Patents

Abstract

The invention discloses an LDPC coding and decoding method applied in an incoherent ultra-wideband communication system, and relates to the field of incoherent ultra-wideband (UWB) communication, in particular to the incoherent UWB communication system based on a transmitter reference pulse cluster (TRPC). According to the method, two kinds of LDPC code construction methods are used in the system, decoding is performed on an LDPC code by using a BP algorithm, and the prior probability needs to be obtained for the channel characteristics by adopting the BP algorithm. The method aims at the channel characteristics of the TRPC system, and the channel noise variance is solved by adopting a sampling value in the actual simulation. The simulation result shows that compared with a RS code, a system long convolution code and a non-system long convolution code which are used by the incoherent ultra-wideband communication system at present, the LDPC coding and decoding method can obtain better performance.

Description

A kind of LDPC coding and interpretation method being applicable to non-coherent ultra-broad band communication system

Technical field

The invention belongs to non-coherent ultra-broad band wireless communication technology field, relate to based on the channel coding method in the non-coherent ultra-broad band communication system of transmission reference pulse bunch, be specifically related to a kind of the LDPC coding and the interpretation method that are applicable to non-coherent ultra-broad band communication system.

Background technology

Along with development, the fusion of radio communication and network technology, ultra broadband (Ultra-wideband, UWB) technology becomes a kind of short-distance wireless communication solution with features such as low cost, low-power consumption and good time domain resolution capabilities.IEEE 802.15.4a standard adopts the UWB scheme based on impulse radio (Impulse radio, IR) standard.IR-UWB system possesses accurate positioning performance, but it still faces the huge challenge of the actual requirement such as low cost, low-power consumption.Therefore, in low bit-rate occasion, be adopt complexity lower, less demanding and do not need the noncoherent receiver of channel estimating to sampling rate more.Wherein, transmission is adopted to receive much concern because its structure is simple, without the need to the accurate synchronous and feature such as channel estimating, Robust Performance with reference to the autocorrelation receiver of (Transmitter Reference, TR) technology.But inter-pulse interference (the Inter-pulse Interference that TR-UWB system causes in order to avoid multi-path delay spread when carrying out auto-correlation and receiving, IPI), need to introduce long delay line to isolate reference pulse and data pulse, this is infeasible with current technical merit.

The proposition of transmission reference pulse bunch (Transmitter Reference pulse cluster, TRPC) system is exactly the long delay line problem in order to solve in TR system.The reference pulse of TRPC signal and data pulse have more unified, compact structure.With less delay T between reference pulse and data pulse _dorder sends, and the minimum delay is exactly the width T of a pulse _p, i.e. T _d=T _p.A reference pulse and a data pulse form a pulse pair, and each pulse is to 2T _drepeating intervally to send of second.

In TRPC system, be difficult to provide desirable bit error rate performance if signal directly sends, therefore carry out chnnel coding before sending to improve systematic function.The coded system specified in IEEE 802.15.4a standard is RS code, but also there are some codes more better than RS code performance in reality, convolution code as long in system, the long convolution code of nonsystematic etc.

In recent years, LDPC code to become the focus of theoretical circles research due to it close to the superperformance of shannon limit.In nineteen sixties, LDPC code does not come into one's own due to the restriction by hardware condition, and nowadays LDPC code has been widely used, is wherein no lack of the application in communication system coding and decoding and high code check UWB.But want LDPC code to be applied in incoherent TRPC-UWB system, also there are the following problems: the parity matrix 1, how constructed, avoid becate and ensure the nonsingularity of submatrix, thus obtaining generator matrix and carry out coding and decoding; When 2, adopting BP algorithm to carry out decoding, need the prior probability of known channel.3, require the prior probability of channel, also need the variance of known multipath channel noise.

Summary of the invention

The object of the invention is to better for performance LDPC code to be used in non-coherent ultra-broad band communication system, solve the problem of LDPC code in this System build code process, propose a kind of the LDPC coding and the interpretation method that are applicable to non-coherent ultra-broad band communication system.

For achieving the above object, present invention employs following technical scheme:

Comprise the following steps:

1), adopt LDPC code to carry out chnnel coding to signal to be sent, obtain sending signal;

2), signal to be sent is sent by UWB channel;

3), receiving terminal receives signal to be sent, and carries out filtering to the signal to be sent received, and obtains Received signal strength;

4), at receiving terminal adopt BP algorithm to carry out decoding to received signal, calculate the parameters needed for decoding, the Received signal strength required for acquisition.

Described step (1) detailed process is:

1.1), arranging signal to be sent is N _findividual compact arranged pulse is to composition, and each pulse is to comprising a reference pulse and a data pulse, according to delay T _dorder sends.The transmission signal generated at transmitting terminal is:

$\begin{array}{c}{\tilde{s}}_n\left(t\right)=\sqrt{\frac{E_b}{2N_f}}\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}\underset{m=0}{\overset{N_f-1}{\mathrm{\Σ}}}[g(t-i{T}_s-2m{T}_d)+b_n\left(i\right)g(t-i{T}_s-(2m+1))]\\ =\sqrt{\frac{E_b}{2N_f}}\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{s}_i(t-i{T}_s)\end{array}---\left(8\right)$

Wherein, E _brepresent the emitted energy of every bit, N represents the bit number of codeword data bag, N _frepresent pulse in each symbol to the number of times repeated, g (t) represents energy normalized ultra-wideband pulse, b _ni { ± 1} is the binary data symbols of launching to () ∈, T _sfor symbol period, T _pfor pulse duration, m is nonnegative integer, and t is variable continuous time, s _it () is defined as:

$s_i\left(t\right)\stackrel{\mathrm{\Δ}}{=}\underset{m=0}{\overset{N_f-1}{\mathrm{\Σ}}}[g(t-2m{T}_d)+b_n\left(i\right)g(t-(2m+1)T_d)]---\left(9\right)$

1.2), adopt cyclic shift matrices structure QC-LDPC code (N, j, k), N represents code length, and j represents column weight, and k representative row is heavy.First index matrix P is provided:

$P=\left[\begin{array}{cccc}1& a& ...& a^{k-1}\\ b& \mathrm{ab}& ...& a^{k-1}b\\ ...& & & \\ b^{j-1}& a{b}^{j-1}& ...& a^{k-1}{b}^{j-1}\end{array}\right]---\left(3\right)$

Wherein, P is a dimension is the index matrix of j × k, wherein 0≤s≤j-1,0≤t≤k-1 and a, b are prime numbers, and j is less than k; The figure place that the submatrix that the value of each element in index matrix represents check matrix moves right;

Check matrix H is:

$H=\left[\begin{array}{cccc}{I}_1& I_a& ...& I_{a^{k-1}}\\ I_b& I_{\mathrm{ab}}& ...& I_{a^{k-1}b}\\ ...& & & \\ I_{b^{j-1}}& I_{{\mathrm{ab}}^{j-1}}& ...& I_{a^{k-1}{b}^{j-1}}\end{array}\right]---\left(4\right)$

The dimension of H is the dimension that jM × kM, M=N/k represent submatrix;

1.3) H is expressed as form:

H＝[A B] (5)

The wherein square formation of A to be dimension be jM × jM, the dimension of B is jM × (k-j) M; Then generator matrix G:

G＝[(A ^-1·B) ^TI] (6)

Wherein, square formation A is nonsingular matrix.

1.4), according to generator matrix G, signal to be sent is encoded, generate and send signal.

Described square formation A building method is:

2.1), change index matrix P, obtain new index matrix P ₁:

$P_1=\left[\begin{array}{ccccc}0& a& a^2& ...& a^{k-1}\\ z& 0& a^{2}b& ...& a^{k-1}b\\ b^2& z& 0& ...& a^{k-1}{b}^2\\ ...& & & & \\ b^{j-1}& a{b}^{j-1}& a^2{b}^{j-1}& ...& a^{k-1}{b}^{j-1}\end{array}\right]---\left(7\right)$

2.2), in this matrix, what z represented is null matrix, and corresponding check matrix is expressed as:

$H_1=\left[\begin{array}{ccccc}{I}_0& I_a& I_{a^2}& ...& I_{a^{k-1}}\\ 0& I_0& I_{a^{2}b}& ...& I_{a^{k-1}b}\\ I_{b^2}& 0& I_0& ...& I_{a^{k-1}{b}^2}\\ ...& & & & \\ I_{b^{j-1}}& I_{a{b}^{j-1}}& I_{a^2{b}^{j-1}}& ...& I_{a^{k-1}{b}^{j-1}}\end{array}\right]---\left(8\right)$

From H ₁in can obtain A ₁

$A_1=\left[\begin{array}{ccccc}{I}_0& I_a& I_{a^2}& ...& I_{a^{j-1}}\\ 0& I_0& I_{a^{2}b}& ...& I_{a^{j-1}b}\\ I_{b^2}& 0& I_{a^2{b}^2}& ...& I_{a^{j-1}{b}^{j-1}}\\ ...& & & & \\ I_{b^{j-1}}& I_{a{b}^{j-1}}& I_{a^2{b}^{j-1}}& ...& I_0\end{array}\right]---\left(9\right)$

By this change, A ₁nonsingular.

Described square formation A building method is:

3.1), first each row of H in exchange equation (5), make the diagonal entry of A be all 1;

3.2), when the row of switching matrix H, matrix column weighs and ring property remains unchanged;

3.3), switching matrix H is converted:

The first step: capable to jth M from the 1st row, replaces with 0 by the element 1 below square formation A diagonal, checks that whether square formation A is nonsingular simultaneously; If not nonsingular, then repeat this step, until A is nonsingular;

Second step: obtain generator matrix from square formation A.

Described step 3) concrete steps be:

In TRPC system, signal is through coding and by after UWB channel and filter, the signal received is:

$r_n\left(t\right)=\tilde{s}\left(t\right)*h\left(t\right)+n\left(t\right)$

$=\sqrt{\frac{E_b}{2N_f}}\underset{k=0}{\overset{K-1}{\mathrm{\Σ}}}{\mathrm{\α}}_k\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{s}_i(t-{\mathrm{\τ}}_k-i{T}_s)+n\left(t\right)---\left(10\right)$

Wherein, * represents linear convolution, and n (t) represents additive white Gaussian noise, and in IEEE802.15.4a standard, multipath channel impulse response is expressed as:

$h\left(t\right)=\underset{k=0}{\overset{K-1}{\mathrm{\Σ}}}{a}_k\mathrm{\δ}(t-{\mathrm{\τ}}_k)---\left(11\right)$

Wherein, α _kand τ _krepresent the multiple fading coefficients of K road multipath component respectively and arrive time delay; In order to ensure there is no intersymbol interference, assuming that T _s>=2N _ft _d+ τ _max, wherein τ _maxrepresent the maximum delay of channel.

Described step 4) concrete steps be:

4.1), receiving terminal to the received signal and self Td postpone carry out auto-correlation, the judgment variables y (i) obtained is expressed as:

$y\left(i\right)={\∫}_{i{T}_S+T_1}^{i{T}_s+T_2}{r}_n\left(t\right){r_n}^{*}(t-T_d)\mathrm{dt}---\left(10\right)$

Wherein, T ₁and T ₂represent starting point and the terminal in autocorrelation integral interval respectively, () ^*represent the conjugate operation of plural number;

4.2), obtain prior probability, and adopt BP algorithm to carry out decoding according to prior probability to judgment variables y (i).

Obtain prior probability specifically to comprise the following steps:

5.1) if x is the stochastic variable that equiprobability produces, so can be obtained by Bayesian formula:

$\frac{p(x=u_0/y)}{p(x=u_1/y)}=\frac{p(y/x=u_0)}{p(y/x=u_1)}---\left(11\right)$

Wherein, u ₀representative be sent as " 0 " time bit, u ₁representative be bit when being sent as " 1 ", y representative be the information bit received.P (y/x=u ₁) what represent is when the information sent is u ₁time receive the probability of y, x is the stochastic variable that equiprobability produces;

5.2), u is established ₀represent when bit " 0 " is by the value after the channel that removes noise, u ₁what represent is that bit " 1 " is by the value after this channel; Y is by value that channel receives in real system; Then formula (13) is:

$\frac{p(y/x=u_0)}{p(y/x=u_1)}=\frac{f(y/x=u_0)}{f(y/x=u_1)}=\frac{\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{-\frac{{(y-u_0)}^2}{2{\mathrm{\δ}}^2}}}{\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{-\frac{{(y-u_1)}^2}{2{\mathrm{\δ}}^2}}}---\left(12\right)$

Definition p (y/x=u ₁)=p ₁, can be obtained by abbreviation (14):

$\frac{1-p_1}{p_1}=e^{\frac{u_1^2-u_0^2+2y{u}_0-2y{u}_1}{2{\mathrm{\δ}}^2}}---\left(13\right)$

$p_1=\frac{1}{1+e^{\frac{u_1^2-u_0^2+2y{u}_0-2y{u}_1}{2{\mathrm{\δ}}^2}}}---\left(14\right)$

5.3), carry out initialization with prior probability after, carry out decoding according to BP algorithm, obtain u ₀, u ₁, variance δ ²depend on the characteristic of channel, can be obtained by sample value.

Compared with the prior art, the present invention has following beneficial effect:

The present invention is directed to the chnnel coding problem of TRPC-UWB system, propose a kind of the LDPC coding and the interpretation method that are applicable to non-coherent ultra-broad band communication system.First the building method of two kinds of LDPC code is used for, in TRPC system, then obtaining prior probability for this characteristic of channel by the inventive method, also tries to achieve the variance of interchannel noise.Required by more than utilizing, BP algorithm is adopted to carry out decoding.The TRPC system of also being encoded by LDPC contrasts with adopting the system of RS code, the long convolution code of system and the long convolution code of nonsystematic, show that the system using LDPC code to encode has better error performance than using the system of other coded systems existing.Visible, the present invention can well improve the performance of TRPC system.

Accompanying drawing explanation

The system model of Fig. 1 for applying in the present invention;

Fig. 2 is the TRPC system and the performance comparison figure of uncoded TRPC system under channel 1 environment that use two kinds of LDPC code in the present invention;

Fig. 3 is the TRPC system and the performance comparison figure of uncoded TRPC system under channel 8 environment that use two kinds of LDPC code in the present invention;

Fig. 4 uses the TRPC system of LDPC code and the performance comparison figure of TRPC system under channel 1 environment using RS code, the long convolution code of system and the long convolution code of nonsystematic in the present invention;

Fig. 5 uses the TRPC system of LDPC code and the performance comparison figure of TRPC system under channel 8 environment using RS code, the long convolution code of system and the long convolution code of nonsystematic in the present invention;

Fig. 6 is FB(flow block) of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further detail:

As a shown in Figure 6, in exemplary Fig. 1, give application scenarios, be applicable to LDPC coding and the interpretation method of non-coherent ultra-broad band communication system, specifically comprise the following steps:

1), adopt LDPC code to carry out chnnel coding to signal to be sent, obtain sending signal;

1.1), arranging signal to be sent is N _findividual compact arranged pulse is to composition, and each pulse is to comprising a reference pulse and a data pulse, according to delay T _dorder sends.The transmission signal generated at transmitting terminal is:

$\begin{array}{c}{\tilde{s}}_n\left(t\right)=\sqrt{\frac{E_b}{2N_f}}\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}\underset{m=0}{\overset{N_f-1}{\mathrm{\Σ}}}[g(t-i{T}_s-2m{T}_d)+b_n\left(i\right)g(t-i{T}_s-(2m+1))]\\ =\sqrt{\frac{E_b}{2N_f}}\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{s}_i(t-i{T}_s)\end{array}---\left(15\right)$

Wherein, E _brepresent the emitted energy of every bit, N represents the bit number of codeword data bag, N _frepresent pulse in each symbol to the number of times repeated, g (t) represents energy normalized ultra-wideband pulse, b _ni { ± 1} is the binary data symbols of launching to () ∈, T _sfor symbol period, T _pfor pulse duration, m is nonnegative integer, and t is variable continuous time, s _it () is defined as:

$s_i\left(t\right)\stackrel{\mathrm{\Δ}}{=}\underset{m=0}{\overset{N_f-1}{\mathrm{\Σ}}}[g(t-2m{T}_d)+b_n\left(i\right)g(t-(2m+1)T_d)]---\left(16\right)$

Uncoded system is difficult to provide desirable bit error rate performance, adopts chnnel coding to improve systematic function.Although IEEE 802.15.4a standard defines the coded systems such as RS code, cascaded code, still there is the better code of some performances, convolution code as long in system, the long convolution code of nonsystematic etc.Consider that LDPC code is the ideal good code of a kind of performance close to shannon limit, be all applied in a lot of fields of ultra-wideband communications in recent years.Therefore LDPC code is adopted as the channel coding method of TRPC system to obtain better performance.

1.2), for the performance improving LDPC code reduces the becate in check matrix factor figure, adopt cyclic shift matrices structure QC-LDPC code (N, j, k), N represents code length, and j represents column weight, and k representative row is heavy, first provides index matrix P:

$P=\left[\begin{array}{cccc}1& a& ...& a^{k-1}\\ b& \mathrm{ab}& ...& a^{k-1}b\\ ...& & & \\ b^{j-1}& a{b}^{j-1}& ...& a^{k-1}{b}^{j-1}\end{array}\right]---\left(3\right)$

Wherein, P is a dimension is the index matrix of j × k, wherein 0≤s≤j-1,0≤t≤k-1 and a, b are prime numbers, and j is less than k; The figure place that the submatrix that the value of each element in index matrix represents check matrix moves right;

Check matrix H is:

$H=\left[\begin{array}{cccc}{I}_1& I_a& ...& I_{a^{k-1}}\\ I_b& I_{\mathrm{ab}}& ...& I_{a^{k-1}b}\\ ...& & & \\ I_{b^{j-1}}& I_{{\mathrm{ab}}^{j-1}}& ...& I_{a^{k-1}{b}^{j-1}}\end{array}\right]---\left(4\right)$

The dimension of H is the dimension that jM × kM, M=N/k represent submatrix;

1.3) H is expressed as form:

H＝[A B] (5)

Wherein, the square formation of A to be dimension be jM × jM, the dimension of B is jM × (k-j) M; Then generator matrix G:

G＝[(A ^-1·B) ^TI] (6)

Wherein, square formation A is nonsingular matrix.The check matrix of this method construct, had not both had Fourth Ring not have six rings yet.But in practice, the QC-LDPC code of this method construct can not ensure it is that A is nonsingular, and generator matrix also cannot be obtained by said method.

1.4), according to generator matrix G, signal to be sent is encoded, generate and send signal.

In TRPC system, n information packet can be expressed as: before these data send, adopt LDPC code to encode.Coding needs the generator matrix of known LDPC code, and generator matrix is transformed by parity matrix, when known (4), (5) can be adopted to obtain generator matrix.The key step of LDPC code coding is the parity matrix constructed, and can obtain generator matrix and can avoid becate again.

Index matrix (3) is adopted to construct parity matrix as shown in (4).For guaranteeing the nonsingularity of (5) formula neutron matrix A, have employed again two kinds of methods and check matrix is transformed.First method changes index matrix as (7) formula, and corresponding check matrix such as (8) formula is guaranteed to obtain generator matrix.Second method is by carrying out rank transformation to (5) formula, then by be about to A diagonal below element 1 replace with 0, progressively judge that whether A nonsingular, thus obtain generator matrix and encode.By corresponding output after LDPC coding be: $c_n\left(i\right)\∈{\left\{\mathrm{0,1}\right\}}_{n=0}^{N-1}.$

Here is the two kinds of methods changed this check matrix, and object makes A be nonsingular, then can obtain generator matrix G and carry out coding and decoding.

First method:

2.1), change index matrix P, obtain new index matrix P ₁:

$P_1=\left[\begin{array}{ccccc}0& a& a^2& ...& a^{k-1}\\ z& 0& a^{2}b& ...& a^{k-1}b\\ b^2& z& 0& ...& a^{k-1}{b}^2\\ ...& & & & \\ b^{j-1}& a{b}^{j-1}& a^2{b}^{j-1}& ...& a^{k-1}{b}^{j-1}\end{array}\right]---\left(7\right)$

2.2), in this matrix, what z represented is null matrix, and corresponding check matrix is expressed as:

$H_1=\left[\begin{array}{ccccc}{I}_0& I_a& I_{a^2}& ...& I_{a^{k-1}}\\ 0& I_0& I_{a^{2}b}& ...& I_{a^{k-1}b}\\ I_{b^2}& 0& I_0& ...& I_{a^{k-1}{b}^2}\\ ...& & & & \\ I_{b^{j-1}}& I_{a{b}^{j-1}}& I_{a^2{b}^{j-1}}& ...& I_{a^{k-1}{b}^{j-1}}\end{array}\right]---\left(8\right)$

From H ₁in obtain A ₁

$A_1=\left[\begin{array}{ccccc}{I}_0& I_a& I_{a^2}& ...& I_{a^{j-1}}\\ 0& I_0& I_{a^{2}b}& ...& I_{a^{j-1}b}\\ I_{b^2}& 0& I_{a^2{b}^2}& ...& I_{a^{j-1}{b}^{j-1}}\\ ...& & & & \\ I_{b^{j-1}}& I_{a{b}^{j-1}}& I_{a^2{b}^{j-1}}& ...& I_0\end{array}\right]---\left(9\right)$

By this change, A ₁be nonsingular, then can obtain generator matrix.The check matrix of this method design can ensure do not have Fourth Ring, but still there are six rings.

Second method:

3.1), first each row of H in exchange equation (6), make the diagonal entry of A be all 1;

3.2), when the row of switching matrix H, matrix column weighs and ring property remains unchanged;

3.3), switching matrix H is converted:

The first step: capable to jth M from the 1st row, replaces with 0 by the element 1 below square formation A diagonal, checks that whether square formation A is nonsingular simultaneously; If not nonsingular, then repeat this step, until A is nonsingular;

Second step: obtain generator matrix from square formation A.

In second method, owing to being the row of switching matrix and the element 1 in replacement matrix, not changing ring property, therefore use the check matrix of this method design to there are not six rings.

2), signal to be sent is sent by UWB channel; After being launched by transmitter, become expression is as shown in formula (1).

3), receiving terminal receives signal to be sent, and carries out filtering to the signal to be transmitted received, and obtains Received signal strength;

In TRPC system, signal is through coding and by after UWB channel and filter, the signal received is:

$\begin{array}{c}{r}_n\left(t\right)=\tilde{s}\left(t\right)*h\left(t\right)+n\left(t\right)\\ =\sqrt{\frac{E_b}{2N_f}}\underset{k=0}{\overset{K-1}{\mathrm{\Σ}}}{\mathrm{\α}}_k\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{s}_i(t-{\mathrm{\τ}}_k-i{T}_s)+n\left(t\right)\end{array}---\left(10\right)$

Wherein, * represents linear convolution, and n (t) represents additive white Gaussian noise, and in IEEE802.15.4a standard, multipath channel impulse response is expressed as:

$h\left(t\right)=\underset{k=0}{\overset{K-1}{\mathrm{\Σ}}}{a}_k\mathrm{\δ}(t-{\mathrm{\τ}}_k)---\left(11\right)$

Wherein, α _kand τ _krepresent the multiple fading coefficients of K road multipath component respectively and arrive time delay; In order to ensure there is no intersymbol interference, assuming that T _s>=2N _ft _d+ τ _max, wherein τ _maxrepresent the maximum delay of channel.

4), at receiving terminal adopt BP algorithm to carry out decoding to received signal, calculate the parameters needed for decoding, the Received signal strength required for acquisition.

Described step 4) concrete steps be:

4.1), receiving terminal to the received signal and self Td postpone carry out auto-correlation, the judgment variables y (i) obtained is expressed as:

$y\left(i\right)={\∫}_{i{T}_S+T_1}^{i{T}_s+T_2}{r}_n\left(t\right){r_n}^{*}(t-T_d)\mathrm{dt}---\left(17\right)$

Wherein, T ₁and T ₂represent starting point and the terminal in autocorrelation integral interval respectively, () ^*represent the conjugate operation of plural number;

4.2), obtain prior probability, and adopt BP algorithm to carry out decoding according to prior probability to judgment variables y (i).

Obtain prior probability specifically to comprise the following steps:

5.1) if x is the stochastic variable that equiprobability produces, so can be obtained by Bayesian formula:

$\frac{p(x=u_0/y)}{p(x=u_1/y)}=\frac{p(y/x=u_0)}{p(y/x=u_1)}---\left(18\right)$

Wherein, u ₀representative be sent as " 0 " time bit, u ₁representative be bit when being sent as " 1 ", y representative be the information bit received.P (y/x=u ₁) what represent is when the information sent is u ₁time receive the probability of y, x is the stochastic variable that equiprobability produces;

5.2), u is established ₀represent when bit " 0 " is by the value after the channel that removes noise, u ₁what represent is that bit " 1 " is by the value after this channel; Y is by value that channel receives in real system; Then formula (13) is:

$\frac{p(y/x=u_0)}{p(y/x=u_1)}=\frac{f(y/x=u_0)}{f(y/x=u_1)}=\frac{\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{-\frac{{(y-u_0)}^2}{2{\mathrm{\δ}}^2}}}{\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{-\frac{{(y-u_1)}^2}{2{\mathrm{\δ}}^2}}}---\left(19\right)$

Definition p (y/x=u ₁)=p ₁, can be obtained by abbreviation (14)

$\frac{1-p_1}{p_1}=e^{\frac{u_1^2-u_0^2+2y{u}_0-2y{u}_1}{2{\mathrm{\δ}}^2}}---\left(20\right)$

5.3), carry out initialization with prior probability after, carry out decoding according to BP algorithm, obtain u ₀, u ₁, variance δ ²depend on the characteristic of channel, can be obtained by sample value.

This algorithm needs the prior probability of known channel, and prior probability formulation process is shown in formula (13)-(16).The visible prior probability that will obtain first need try to achieve channel noise variance.Because actual derivation need consider auto-correlation and the cross-correlation effect of random signal, be therefore difficult to the complete expression drawing noise variance, but can obtain from sample value.In the system of reality, IEEE 802.15.4a channel model standardization program is utilized to set up the template of a noisy template and a removal noise, then allow a training symbol respectively by these two templates, the value obtained just can be used for calculating noise variance δ ².For making desired value, there is generality, adopt different number of times averaged respectively, as 100 times, 1000 times or 10000 times, just close to optimum performance when finding to be averaging for 1000 times in simulation process, therefore choose and average for 1000 times to reach good effect.

Simulation process of the present invention carries out respectively under channel 1 and channel 8 environment.Be illustrated in figure 2 the performance comparison analogous diagram of TRPC system under channel 1 environment of use two kinds of method construct LDPC code.Be illustrated in figure 3 the performance comparison analogous diagram of TRPC system under channel 8 environment of use two kinds of method construct LDPC code.Above two figure all contrast with uncoded system, from simulation curve, adopt the system of LDPC code coding to have larger performance improvement than uncoded system, and adopt the system of second method structure LDPC code to have better performance than adopting the system of first method.Fig. 4 and Fig. 5 uses the system of first method structure LDPC code and uses the system of RS code, the long convolution code of system and the long convolution code of nonsystematic at the performance comparison figure of channel 1 with channel 8.The system of LDPC code coding has best performance as seen from the figure.

The above TRPC-UWB system is only the preferred embodiment of the incoherent UWB communication system of the present invention, be not limited to the present invention, for a person skilled in the art, the present invention can have various modifications and variations, the method can be used in other incoherent UWB communication systems, as TR system, NC-PPM system etc.Within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (7)

1. be applicable to LDPC coding and the interpretation method of non-coherent ultra-broad band communication system, it is characterized in that, comprise the following steps:

1), adopt LDPC code to carry out chnnel coding to signal to be sent, obtain sending signal;

2), signal to be sent is sent by UWB channel;

3), receiving terminal receives signal to be sent, and carries out filtering to the signal to be sent received, and obtains Received signal strength;

4), at receiving terminal adopt BP algorithm to carry out decoding to received signal, calculate the parameters needed for decoding, the Received signal strength required for acquisition.

2. according to claim 1 a kind of be applicable to non-coherent ultra-broad band communication system LDPC coding and interpretation method, it is characterized in that, described step 1) detailed process is:

1.1), arranging signal to be sent is N _findividual compact arranged pulse is to composition, and each pulse is to comprising a reference pulse and a data pulse, according to delay T _dorder sends, and the transmission signal generated at transmitting terminal is:

$\begin{array}{c}{\tilde{s}}_n\left(t\right)=\sqrt{\frac{E_b}{{2N}_f}}\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}\underset{m=0}{\overset{N_f-1}{\mathrm{\Σ}}}[g(t-{\mathrm{iT}}_s-{2\mathrm{mT}}_d)+b_n\left(i\right)g(t-{\mathrm{iT}}_s-(2m+1))]\\ =\sqrt{\frac{E_b}{{2N}_f}}\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{s}_i(t-{\mathrm{iT}}_s)\end{array}---\left(1\right)$

Wherein, E _brepresent the emitted energy of every bit, N represents the bit number of codeword data bag, N _frepresent pulse in each symbol to the number of times repeated, g (t) represents energy normalized ultra-wideband pulse, b _ni { ± 1} is the binary data symbols of launching to () ∈, T _sfor symbol period, T _pfor pulse duration, m is nonnegative integer, and t is variable continuous time, s _it () is defined as:

$s_i\left(t\right)\stackrel{\mathrm{\Δ}}{=}\underset{m=0}{\overset{N_f-1}{\mathrm{\Σ}}}[g(t-{2\mathrm{mT}}_d)+b_n\left(i\right)g(t-(2m+1)T_d)]---\left(2\right)$

1.2), adopt cyclic shift matrices structure QC-LDPC code (N, j, k), N represents code length, and j represents column weight, and k representative row is heavy, first provides index matrix P:

$P=\left[\begin{array}{cccc}1& a& ...& a^{k-1}\\ b& \mathrm{ab}& ...& a^{k-1}b\\ ...& & & \\ b^{j-1}& {\mathrm{ab}}^{j-1}& ...& a^{k-1}{b}^{j-1}\end{array}\right]---\left(3\right)$

Wherein, P is a dimension is the index matrix of j × k, wherein 0≤s≤j-1,0≤t≤k-1 and a, b are prime numbers, and j is less than k; The figure place that the submatrix that the value of each element in index matrix represents check matrix moves right;

Check matrix H is:

$H=\left[\begin{array}{cccc}{I}_1& I_a& ...& I_{a^{k-1}}\\ I_b& I_{\mathrm{ab}}& ...& I_{a^{k-1}b}\\ ...& & & \\ I_{b^{j-1}}& I_{{\mathrm{ab}}^{j-1}}& ...& I_{a^{k-1}{b}^{j-1}}\end{array}\right]---\left(4\right)$

The dimension of H is the dimension that jM × kM, M=N/k represent submatrix;

1.3) H is expressed as form:

H＝[AB] (5)

The wherein square formation of A to be dimension be jM × jM, the dimension of B is jM × (k-j) M; Then generator matrix G:

G＝[(A ^-1·B) ^TI] (6)

Wherein, square formation A is nonsingular matrix;

1.4), according to generator matrix G, signal to be sent is encoded, generate and send signal.

3. a kind of LDPC coding and interpretation method being applicable to non-coherent ultra-broad band communication system according to claim 2, it is characterized in that, described square formation A building method is:

2.1), change index matrix P, obtain new index matrix P ₁:

$P_1=\left[\begin{array}{ccccc}0& a& a^2& ...& a^{k-1}\\ z& 0& a^{2}b& ...& a^{k-1}b\\ b^2& z& 0& ...& a^{k-1}{b}^2\\ ...& & & & \\ b^{j-1}& {\mathrm{ab}}^{j-1}& a^2{b}^{j-1}& ...& a^{k-1}{b}^{j-1}\end{array}\right]---\left(7\right)$

2.2), in this matrix, what z represented is null matrix, and corresponding check matrix is expressed as:

$H_1=\left[\begin{array}{ccccc}{I}_0& I_a& I_{a^2}& ...& I_{a^{k-1}}\\ 0& I_0& I_{a^{2}b}& ...& I_{a^{k-1}b}\\ I_{b^2}& 0& I_0& ...& I_{a^{k-1}{b}^2}\\ ...& & & & \\ I_{b^{j-1}}& I_{{\mathrm{ab}}^{j-1}}& I_{a^2{b}^{j-1}}& ...& I_{a^{k-1}{b}^{j-1}}\end{array}\right]---\left(8\right)$

From H ₁in can obtain A ₁

$A_1=\left[\begin{array}{ccccc}{I}_0& I_a& I_{a^2}& ...& I_{a^{j-1}}\\ 0& I_0& I_{a^{2}b}& ...& I_{a^{j-1}b}\\ I_{b^2}& 0& I_{a^2{b}^2}& ...& I_{a^{j-1}{b}^{j-1}}\\ ...& & & & \\ I_{b^{j-1}}& I_{{\mathrm{ab}}^{j-1}}& I_{a^2{b}^{j-1}}& ...& I_0\end{array}\right]---\left(9\right)$

By this change, A ₁nonsingular.

4. a kind of LDPC coding and interpretation method being applicable to non-coherent ultra-broad band communication system according to claim 2, it is characterized in that, described square formation A building method is:

3.1), first each row of H in exchange equation (5), make the diagonal entry of A be all 1;

3.2), when the row of switching matrix H, matrix column weighs and ring property remains unchanged;

3.3), switching matrix H is converted:

The first step: capable to jth M from the 1st row, replaces with 0 by the element 1 below square formation A diagonal, checks that whether square formation A is nonsingular simultaneously; If not nonsingular, then repeat this step, until A is nonsingular;

Second step: obtain generator matrix from square formation A.

5. according to claim 2 a kind of be applicable to non-coherent ultra-broad band communication system LDPC coding and interpretation method, it is characterized in that, described step 3) concrete steps be:

In TRPC system, signal is through coding and by after UWB channel and filter, the signal received is:

$\begin{array}{c}{r}_n\left(t\right)=\tilde{s}\left(t\right)*h\left(t\right)+n\left(t\right)\\ =\sqrt{\frac{E_b}{{2N}_f}}\underset{k=0}{\overset{K-1}{\mathrm{\Σ}}}{\mathrm{\α}}_k\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{s}_i(t-{\mathrm{\τ}}_k-{\mathrm{iT}}_s)+n\left(t\right)\end{array}---\left(10\right)$

Wherein, * represents linear convolution, and n (t) represents additive white Gaussian noise, and in IEEE 802.15.4a standard, multipath channel impulse response is expressed as:

$h\left(t\right)=\underset{k=0}{\overset{K-1}{\mathrm{\Σ}}}{a}_k\mathrm{\δ}(t-{\mathrm{\τ}}_k)---\left(11\right)$

Wherein, α _kand τ _krepresent the multiple fading coefficients of K road multipath component respectively and arrive time delay; In order to ensure there is no intersymbol interference, if T _s>=2N _ft _d+ τ _max, wherein τ _maxrepresent the maximum delay of channel.

6. according to claim 2 a kind of be applicable to non-coherent ultra-broad band communication system LDPC coding and interpretation method, it is characterized in that, described step 4) concrete steps be:

4.1), receiving terminal to the received signal and the T of self _dpostpone to carry out auto-correlation, the judgment variables y (i) obtained is expressed as:

$y\left(i\right)={\∫}_{{\mathrm{iT}}_s+T_1}^{{\mathrm{iT}}_s+T_2}{r}_n\left(t\right){r_n}^{*}(t-T_d)\mathrm{dt}---\left(3\right)$

Wherein, T ₁and T ₂represent starting point and the terminal in autocorrelation integral interval respectively, () ^*represent the conjugate operation of plural number;

4.2), obtain prior probability, and adopt BP algorithm to carry out decoding according to prior probability to judgment variables y (i).

7. a kind of LDPC coding and interpretation method being applicable to non-coherent ultra-broad band communication system according to claim 6, is characterized in that, obtain prior probability and specifically comprise the following steps:

5.1) if x is the stochastic variable that equiprobability produces, so can be obtained by Bayesian formula:

$\frac{p(x=u_0/y)}{p(x=u_1/y)}=\frac{p(y/x=u_0)}{p(y/x=u_1)}---\left(4\right)$

Wherein, u ₀representative be sent as " 0 " time bit, u ₁representative be bit when being sent as " 1 ", y representative be the information bit received, p (y/x=u ₁) what represent is when the information sent is u ₁time receive the probability of y, x is the stochastic variable that equiprobability produces;

5.2), u is established ₀represent when bit " 0 " is by the value after the channel that removes noise, u ₁what represent is that bit " 1 " is by the value after this channel; Y is by value that channel receives in real system; Then formula (13) is:

$\frac{p(y/x=u_0)}{p(y/x=u_1)}=\frac{f(y/x=u_0)}{f(y/x=u_1)}=\frac{{\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}e}^{-\frac{{(y-u_0)}^2}{{2\mathrm{\δ}}^2}}}{{\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}e}^{-\frac{{(y-u_1)}^2}{{2\mathrm{\δ}}^2}}}---\left(5\right)$

Definition p (y/x=u ₁)=p ₁, can be obtained by abbreviation (14)

$\frac{1-p_1}{p_1}=e^{\frac{u_1^2-u_0^2{+2\mathrm{yu}}_0-{2\mathrm{yu}}_1}{{2\mathrm{\δ}}^2}}---\left(6\right)$

$p_1=\frac{1}{1+e^{\frac{u_1^2-u_0^2+{2\mathrm{yu}}_0-{2\mathrm{yu}}_1}{{2\mathrm{\δ}}^2}}}$

5.3), carry out initialization with prior probability after, carry out decoding according to BP algorithm, obtain u ₀, u ₁, variance δ ²depend on the characteristic of channel, can be obtained by sample value.