Decoding method and device and systems implementing them

Abstract

In order to decode a sequence α=(α1, . . . , αi, . . . , αn) where αi is the received electrical signal corresponding to a transmitted signal ai representing the ith binary element vi of a word v=(v1, . . . , vn) chosen in a code C of words satisfying v·hT=0, where h is a row n-tuplet on the set {0,1 }, whose number of 1 is denoted w, an item of extrinsic information ρext[A(i,h)]=P[ai=−1|A(i,h)]/P[ai=+1|A(i,h)] is determined on each of the elements vi covered by h, A(i,h) being the set of the received values αj covered by h, with the exception of αi, and P[ai|A(i,h)] being the probability that the ith signal transmitted was ai. This gives ρext[A(i,h)]=[S1(i)+S3(i)+ . . . ]/[1+S2(i)+S4(i)+ . . . ] where the numbers Sr(i) are calculated by applying the recurrence r-1⁢∑i=1w⁢zαi⁢Sr-1⁢(i)-zαj⁢Sr-1⁢(j)=Sr⁢(j) to the numbers S0(i) initialised to 1, with z=exp(−4 E/N), where E is the energy of the transmitted signals ai and N is the spectral power density of the noise on the transmission channel.

Description

FIELD OF THE INVENTION

The present invention relates to a decoding method and device and to systems using them.

BACKGROUND OF THE INVENTION

The situation is considered where a set of information to be transmitted is represented by a sequence of symbols belonging to the set {0,1}. This set is referred to as the binary alphabet and its elements are referred to as binary elements or bits.

In order to transmit these binary elements, they are converted into electrical quantities. For example, the bit 0 is represented by a positive electrical signal and the bit 1 by a negative electrical signal. These electrical signals have the same absolute value, which is here arbitrarily chosen so as to be equal to 1 in order to simplify the description. However, in reality, these signals can take any value deemed appropriate according to the envisaged application, such as, for example, an electrical voltage of ±5 volts.

When these electrical signals are transmitted over a transmission channel impaired by noise, the received values differ from the transmitted values.

In particular, if the transmission channel is impaired by a white Gaussian noise, the received value a corresponding to a transmitted symbol a belonging to the set {−1,+1} is a random variable whose probability density is given by p(α|a)=(σ√{square root over (2π)})⁻¹exp[−(a−α)²/2σ²], where the parameter σ is specified by the signal to noise ratio of the transmission channel: σ=√{square root over (N/2E)} where N is the spectral power density of the noise and E is the energy of the signal transmitted.

The probability that the symbol a has been transmitted, if α is the symbol received, is denoted P(a|α). The value of ρ(α)=P(−1|α)/P(+1|α) can be used to obtain an estimation â of the transmitted symbol a, received in the form of α: if ρ(α)>1 then â is chosen so as to be equal to −1 and if ρ(α)<1 then â is chosen so as to be equal to +1.

For the purpose of limiting the effect of the noise on the efficacy of the transmission of information, it is known that an error correcting encoding can be implemented, which consists of using only a small proportion of all the possible sequences for representing the information.

An example of such an error correcting encoding is block linear encoding: the binary sequence to be transmitted is a sequence of words of n binary elements, n being a positive integer, each of these words being chosen in the subset C of the words v of length n which satisfy v·H^T=0, where H is a matrix of dimension (n−k)×n on the set {0,1}, 0 is the (n−k)-tuplet of zeros and ^T represents the transposition, k being an integer less than n. In addition, the components of the matrix product v·H^T are calculated modulo 2.

It is well known to a person skilled in the art that any word v of the code C satisfies v·h^T=0 for any binary n-tuplet h which is the result of a linear combination of the rows of the matrix H. It should be noted that, in this context, the expression “linear combination” implies that the coefficients which define it are themselves binary elements and the result is always to be reduced modulo 2.

The set of words h thus obtained is referred to as the orthogonal code or dual of the code C and is generally denoted C^⊥. Let us consider a word h of C^⊥ whose weight is w (w being an integer less than n), which means that it contains w binary elements equal to 1 and n−w binary elements equal to 0.

Assume, in order to simplify, that these w binary elements equal to 1 appear in the first w positions of h: h=(1, . . . , 1, 0, . . . , 0). Let v=(v₁, . . . , v_n). The equation v·h^T=0 therefore means $\sum _{i=1}^w{v}_i=0$modulo 2. It implies in particular:v₁=v₂+v₃+ . . . +v_w modulo 2  (1)and, more generally,v_i=v₁+ . . . +v_i−1+v_i+1+ . . . +v_w modulo 2  (2)for any integer i between 1 and w.

Let a=(a₁, . . . , a_n) be the sequence of electrical signals transmitted belonging, in order to simplify, to {−1,+1} and representing the binary n-tuplet v. Let α=(α₁, . . . , α_n) be the corresponding received sequence. Equations (1) and (2) above show that, given h, there are, for each of the first w binary values v_i, two independent items of information which can be extracted from α. The first is the received value α_i, from which it is possible to calculate ρ(α_i) as explained above. The second is the set, denoted A(i,h), of the values α_j, j=1, . . . , i−1, i+1, . . . , w. This is because, for any i, the values α_j of A(i,h) are a noisy image of the corresponding symbols a_j, which are a faithful translation of the corresponding binary elements v_j.

In order to explain this second item of information, said to be extrinsic, on v_i, the quantity z=exp(−2/σ²)=exp(−4E/N) is introduced, which depends on the signal to noise ratio of the transmission channel in question, and there are defined:S₁(i)=Σz^αj,α_jεA(i,h),S₂(i)=Σz^αj+αk,α_j, α_kεA(i,h),j<k,S₃(i)=Σz^αj+αk+αl,α_j,α_k,α_lεA(i,h),j<k<l, . . . S_w−1(i)=z^{α1+ . . . +αi−1+αi+1+ . . . +αw}

P[a_i|A(i,h)] is defined as being the probability that the i^th signal transmitted was a_i given the symbols α_j of A(i,h). The quantity ρ_ext[A(i,h)]=P[a_i=−1|A(i,h)]/P[a_i=+1|A(i,h)] supplies “additional information” on the value of the transmitted symbol a_i.

It can be shown that the quantities ρ_ext[A(i,h)] have a very simple expression according to the polynomials S_r(i) introduced above:ρ_ext[A(i,h)]=[S₁(i)+S₃(i)+ . . . ]/[1+S₂(i)+S₄(i)+ . . . ]

When methods of the probabilistic type are used for estimating what was the transmitted sequence (or only some of its binary elements), the following problem is posed: it is sought to determine the quantities ρ_ext[A(i,h)] for i=1, . . . , w with a calculation cost as low as possible, on the basis of the w binary elements received represented by the received signals α_j, corresponding to a word h of the code C^⊥.

SUMMARY OF THE INVENTION

The purpose of the present invention is to afford a solution to the aforementioned problem.

To this end, the present invention proposes a method of decoding a received sequence α=(α₁, . . . , α_n) where, for any integer i between 1 and n, n being an integer greater than 1, α_i is the received electrical signal corresponding to the transmission of an electrical signal a_i representing the i^th binary element v_i of a word chosen in a binary code C of words v=(v₁, . . . , v_n) satisfying v·h^T=0, where h is a row n-tuplet on the set {0,1} whose number of 1 is denoted w, where ^T represents the transposition and the scalar product v·h^T is calculated modulo 2,

• • this decoding method including a step consisting of determining extrinsic information on each of the binary elements of v covered by h, the extrinsic information given on the i^th binary element of v, assumed to be covered by h, being the quantity ρ_ext[A(i,h)]=P[a_i=−1|A(i,h)]/P[a_i=+1|A(i,h)], where A(i,h) is the set of received values α_j of α which are covered by h, with the exception of α_i, and where P[a_i|A(i,h)] is the probability, calculated on the basis of the received signals α_j of A(i,h), that the i^th signal transmitted was a_i,
  • this decoding method being remarkable in that the determination of the extrinsic information is effected by means of the formulaρ_ext[A(i,h)]=[S₁(i)+S₃(i)+ . . . ]/[1+S₂(i)+S₄(i)+ . . . ]where the numbers S_r(i), for any integer r between 1 and w−1, are calculated by applying the recurrence $r^{-1}\sum _{i=1}^w{z}^{{\alpha }_i}{S}_{r-1}\left(i\right)-z^{{\alpha }_j}{S}_{r-1}\left(j\right)=S_r\left(j\right)$to the numbers S₀(i) initialised to 1, with z=exp(−4E/N), where E is the energy of the transmitted signal a_i and N is the spectral power density of the noise on the transmission channel.

It is said that the word h covers the index position i (of v, a or α) if the binary element of h in position i is 1.

The present invention thus makes it possible to simplify the calculation of the quantities ρ_ext[A(i,h)], which can thus be effected in a number of steps which can be expressed in polynomial form according to the weight w of h.

According to a particular characteristic, a supplementary item of extrinsic information is determined on each of the binary elements of v covered by h by applying the recurrence $r^{-1}\sum _{i=1}^w\underset{\_}{\rho }\left({\alpha }_i\right)S_{r-1}\left(i\right)-\underset{\_}{\rho }\left({\alpha }_j\right)S_{r-1}\left(j\right)=S_r\left(j\right)$where ρ(α_i) represents the ratio between the probability that a_i is equal to −1 and the probability that a_i is equal to +1, these probabilities taking into account at least part of the extrinsic information calculations already made.

Thus the extrinsic information already calculated can be used iteratively.

According to a particular characteristic, the quantity ρ(α_i) is given by ρ(α_i)=ρ(α_i)·ρ_ext[A(i,h)] where ρ(α_i)=P(−1|α_i)/P(+1|α_i), P(a_i|α_i) designating the probability that the i^th signal transmitted was a_i if the i^th signal received is α_i.

According to a particular characteristic, the decoding method of the invention is implemented in a turbodecoding method.

This simplifies this turbodecoding method without weakening its effectiveness.

According to a particular characteristic, the calculations relating to the aforementioned recurrence are effected in multiple precision.

Thus the precision of the calculations is not affected by the calculation of the differences, which is inherent in the proposed decoding method.

For the same purpose as indicated above, the present invention also proposes a device for decoding a received sequence α=(α₁, . . . , α_n) where, for any integer i between 1 and n, n being an integer greater than 1, α_i is the received electrical signal corresponding to the transmission of an electrical signal a_i representing the i^th binary element v_i of a word chosen in a binary code C of words v=(v₁, . . . , v_n) satisfying v·h^T=0, where h is a row n-tuplet on the set {0,1} whose number of 1 is denoted w, where ^T represents the transposition and the scalar product v·h^T is calculated modulo 2,

• • this decoding device having means for determining extrinsic information on each of the binary elements of v covered by h, the extrinsic information given on the i^th binary element of v, assumed to be covered by h, being the quantity ρ_ext[A(i,h)]=P[a_i=−1|A(i,h)]/P[a_i=+1|A(i,h)], where A(i,h) is the set of received values α_j of α which are covered by h, with the exception of α_i, and where P[a_i|A(i,h)] is the probability, calculated on the basis of the received signals α_j of A(i,h), that the i^th signal transmitted was a_i,
  • this decoding device being remarkable in that the determination of the extrinsic information is effected by means of the formulaρ_ext[A(i,h)]=[S₁(i)+S₃(i)+ . . . ]/[1+S₂(i)+S₄(i)+ . . . ]where the numbers S_r(i), for any integer r between 1 and w−1, are calculated by applying the recurrence $r^{-1}\sum _{i=1}^w{z}^{{\alpha }_i}{S}_{r-1}\left(i\right)-z^{{\alpha }_j}{S}_{r-1}\left(j\right)=S_r\left(j\right)$to the numbers S₀(i) initialised to 1, with z=exp(−4E/N), where E is the energy of the transmitted signals a_i and N is the spectral power density of the noise on the transmission channel.

Thus the calculation of the quantities ρ_ext[A(i,h)] requires a device whose complexity is only polynomial in the weight w of h.

In addition, the particular characteristics of the decoding device, which are not explained below, and their advantages, are similar to those of the decoding method and will therefore not be stated here.

In a particular embodiment, the device has:

• • a plurality of multipliers, each of the multipliers receiving, at a first input, the value of S_r−1(i) and, at its second input, the value −z^αi;
  • a plurality of adders, a first input of each of the adders being respectively connected to the output of each of the multipliers;
  • an adding module, whose input is connected to the output of each of the multipliers;
  • an additional multiplication module, a first input of which is connected to the output of the adding module and the second input of which receives the value −1/r, the output of the additional multiplication module being connected to the second input of each of the adders; and
  • a delay introduction module, whose input is connected to the output of each of the adders and whose output is connected to the first input of each of the multipliers,this device being initialised by S₀(i)=1 for any i, so that each of the adders outputs the value S_r(i).

In a particular embodiment, several of the calculations necessary for determining the extrinsic information can be made by circuits put in parallel.

In another embodiment, several of the calculations necessary for determining the extrinsic information can be made by circuits put in series.

The present invention also relates to a digital signal processing apparatus, having means adapted to implement a decoding method as above.

The present invention also relates to a digital signal processing apparatus, having a decoding device as above.

The present invention also relates to a telecommunications network, having means adapted to implement a decoding method as above.

The present invention also relates to a telecommunications network, having a decoding device as above.

The present invention also relates to a mobile station in a telecommunications network, having means adapted to implement a decoding method as above.

The present invention also relates to a mobile station in a telecommunications network, having a decoding device as above.

The present invention also relates to a device for processing signals representing speech, including a decoding device as above.

The present invention also relates to a data transmission device having a transmitter adapted to implement a packet transmission protocol, having a decoding device and/or a device for processing signals representing speech as above.

According to a particular characteristic of the data transmission device, the packet transmission protocol is of the ATM (Asynchronous Transfer Mode) type.

As a variant, the packet transmission protocol is of the IP (Internet Protocol) type.

The invention also relates to:

• • an information storage means which can be read by a computer or microprocessor storing instructions of a computer program, making it possible to implement a decoding method as above, and
  • an information storage means which is removable, partially or totally, and which can be read by a computer or microprocessor storing instructions of a computer program, making it possible to implement a decoding method as above. By way of non-limiting example, such a removable storage means may be a floppy disk or a CD-ROM or a DVD-ROM.

The invention also relates to a computer program containing sequences of instructions for implementing a decoding method as above.

The particular characteristics and the advantages of the different digital signal processing apparatus, the different telecommunications networks, the different mobile stations, the device for processing signals representing speech, the data transmission device, the information storage means and the computer program being similar to those of the decoding method according to the invention, they are not stated here.

Other aspects and advantages of the invention will emerge from a reading of the following detailed description of particular embodiments, given by way of non-limitative examples. The description refers to the drawings which accompany it, in which:

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 depicts schematically a circuit for the simultaneous calculation of the quantities S_r(j) using a decoding method according to the present invention, in a particular embodiment;

FIG. 2 depicts schematically a mode of iterative use of the circuit of FIG. 1;

FIG. 3 depicts schematically a more general embodiment of the circuit for calculating the quantities S_r(j), in which this calculation is made using information ρ(α_ii) more general than the simple knowledge of the received symbols α_i;

FIG. 4 illustrates schematically the application of the invention to the field of turbocodes, in a particular embodiment;

FIG. 5 is a flow diagram illustrating steps of a decoding method according to the present invention; and

FIG. 6 depicts schematically an electronic device including a decoding device according to the present invention, in a particular embodiment.

DETAIL DESCRIPTION OF THE PREFERRED EMBODIMENTS

A received sequence α=(α₁, . . . , α_n), which was passed through a transmission channel impaired by white Gaussian noise, is considered below.

For any integer i between 1 and n, n being an integer greater than 1, α_i is the received electrical signal corresponding to the transmission of an electrical signal a_i. The electrical signal a_i represents the i^th binary element v_i of a word v chosen in a code C of words satisfying v·h^T=0, where h is a row n-tuplet of binary elements whose number of 1 is denoted w, where ^T represents the transposition and where the scalar product v·h^T is calculated modulo 2.

The decoding method of the invention includes a step consisting of determining extrinsic information on each of the binary elements v_i of said to be covered by h=(h₁, . . . , h_n), that is to say for any i such that h_i=1. It is said in fact that the word h covers the position of index i of v if the binary element of h in position i is 1.

The method proposed will be illustrated by the example, in no way limitative, where w=4.

From the definitions given in the introduction, the following expressions arise:

 S₁(1)=z^α2+z^α3+z^α4,S₁(2)=z^α1+z^α3+z^α4,S₁(3)=z^α1+z^α2+z^α4,S₁(4)=z^α1+z^α2+z^α3,S₂(1)=z^α2+α3+z^α2+α4+z^α3+α4,S₂(2)=z^α1+α3+z^α1+α4+z^α3+α4,S₂(3)=z^α1+α2+z^α1+α4+z^α2+α4,S₂(4)=z^α1+α2+z^α1+α3+z^α2+α3,S₃(1)=z^α2+α3+α4,S₃(2)=z^α1+α3+α4,S₃(3)=z^α1+α2+α4,S₃(4)=z^α1+α2+α3.

It can be checked that the above expressions satisfy the following equations: $\sum _{i=1}^4{z}^{{\alpha }_i}{S}_1\left(i\right)-2z^{{\alpha }_j}{S}_1\left(j\right)=2S_2\left(j\right),\sum _{i=1}^4{z}^{{\alpha }_i}{S}_2\left(i\right)-3z^{{\alpha }_j}{S}_2\left(j\right)=3S_3\left(j\right).$

These equations are particular cases of more general equations valid for any w. In fact let S₀(i)=1 for any i. For r=1, . . . , w−1: $\begin{array}{cc}\sum _{i=1}^w{z}^{{\alpha }_i}{S}_{r-1}\left(i\right)=r\sum z^{{\alpha }_{j_1}+\dots +{\alpha }_{j_r}},& \left(3\right)\\ {\mathrm{rz}}^{{\alpha }_j}{S}_{r-1}\left(j\right)=r\sum z^{{\alpha }_{j_1}+\dots +{\alpha }_{j_{r-1}}+{\alpha }_j},& \left(4\right)\end{array}$where, in the right-hand member of equation (3), the sum relates to all the r-tuplets (j₁, . . . , j_r) of integers between 1 and w (including delimiters) and satisfying j₁<j₂< . . . <j_r, and where, in the right-hand member of equation (4), the sum relates to all the (r−1)-tuplets (j₁, . . . , j_r−1) of integers which are different and not equal to j, between 1 and w (including delimiters) and satisfying j₁<j₂< . . . <j_r−1.

It is verified that the difference between the right-hand members of equations (3) and (4) is given by:rΣz^αj₁^{+ . . . +αj}_r−rΣz^αj₁^{+ . . . +αj}_r−1^+αj=rS_r(j),from which: $r^{-1}\sum _{i=1}^w{z}^{{\alpha }_i}{S}_{r-1}\left(i\right)-z^{{\alpha }_j}{S}_{r-1}\left(j\right)=S_r\left(j\right).$

The interpretation of the terms of this recurrence is as follows. The sum $r^{-1}\sum _{i=1}^w{z}^{{\alpha }_i}{S}_{r-1}\left(i\right)$includes once and only once each monomial obtained as the product of r different factors z^αi. The term z^αjS_r−1(j) includes once and only once each monomial obtained as the product of r different factors z^αi, under the constraint that the monomial z^αj is a factor of each of these monomials. Consequently the difference between these two expressions is indeed the sum of all the monomials obtained as a product of r different factors z^αi amongst which the monomial z^αj does not appear, that is to say S_r(j).

This recurrence makes it possible, for r>0, to determine all the S_r(j) values for j=1, . . . , w as soon as all the S_r−1(j) values are known for j=1, . . . , w.

Given that the use of this recurrence entails the calculation of a difference, it is advantageous to make the calculations in multiple precision, in order to avoid the conventional problem of lack of precision in the result, which can be posed when this result is small compared with the terms whose difference is being calculated.

To this end, it is for example possible to represent the quantities manipulated by a number of binary elements twice as great as the minimum necessary.

The present invention lends itself particularly well to implantations of the “serial” or “parallel” type, using a very simple base circuit, illustrated in FIG. 1. This circuit is in fact adapted to the iterative calculation of the w·(w−1) values S_r(i) for i=1, . . . , w and r=1, . . . , w−1.

It has a series of multipliers 10₁, 10₂, 10₃, . . . , 10_w, each multiplier 10_i receiving, at a first input, the value of S_r−1(i) and, at its second input, the value −z^αi.

The circuit also includes a series of adders 12₁, 12₂, 12₃, . . . , 12_w. The output of each multiplier 10_i is connected to a first input of the adder 12_i.

The output of each multiplier 10_i is also connected to the input of a single adding module 14. The output of the adding module 14 is connected to a first input of a multiplier 16, which receives at its second input the value −1/r.

The output of the multiplier 16 is connected to the second input of each adder 12_i.

Thus each adder 12_i outputs the value S_r(i).

It suffices to initialise the inputs of the circuit by S₀(i)=1 for any i. For r=1, the circuit calculates in one step all the values S₁(i), i=1, . . . , w. Using these values S₁(i) at the input of the circuit with r=2, the values S₂(i) are obtained; by continuing in this way, the values S_r(i) for any r are obtained. FIG. 2 illustrates schematically this iterative functioning mode.

FIG. 2 is identical to FIG. 1 and will therefore not be described again here, except that the outputs of the adders 12₁, 12₂, 12₃, . . . , 12_w are respectively connected to the inputs of the multipliers 10₁, 10₂, 10₃, . . . , 10_w by means of a matrix of delay elements 18, which makes it possible to use the recurrence giving the values S_r(i) according to the values S_r−1(i).

It is also possible to form a global circuit by disposing in cascade w−1 circuits like the one in FIG. 2 and initialising the first of them by S₀(i)=1 for any i. In this case, the w values S_r(i), i=1, . . . , w appear as outputs of the r^th stage of the global circuit.

It is also possible to dispose a predetermined number of circuits like the one in FIG. 1 in parallel, in an iterative version (like the one in FIG. 2) or in cascade, in order to simultaneously calculate several different items of extrinsic information corresponding to parity relationships specified by different words h of the dual code C^⊥.

The invention has a particular application for turbodecoding.

In the above, the case of a transmission channel with white Gaussian noise was considered, along with the determination of extrinsic information obtained via one or more words h of the code C^⊥ orthogonal to the code C used for transmission.

However, after having determined, on the basis of one or more words h of C^⊥, one or more items of extrinsic information concerning the values of the transmitted symbols, it may be effective, as shown by recent works with regard to turbocodes (see for example the article by R. Pyndiah entitled “Near optimum decoding of product codes: block turbo codes”, in IEEE Transactions on Communications, 46, No. 8, pages 1003 to 1010, August 1998), to calculate new items of extrinsic information, on the basis not only of the values ρ(a)=P(−1|α)/P(+1|α), but on the basis of values ρ(α) taking into account all or part of the extrinsic information calculations already made. This remark is at the basis of the iterative decoding of error correcting codes. As indicated in the article by R. Pyndiah, the product codes lend themselves particularly well to this type of decoding. In particular, if, on the basis of one or more words h of C^⊥, quantities ρ_ext[A(i,h)] have been calculated, it is possible to calculate new extrinsic information items, on the basis no longer only of the values ρ(α_i)=P(−1|α_i)/P(+1|α_i), but on the basis of the products ρ(α_i)·ρ_ext[A(i,h)].

The circuit described above with the help of FIG. 1 can easily be used for this purpose. This is because, in this circuit, the ρ(α_i) values are represented by the parameters z^αi: this gives ρ(α_i)=z^αi . Thus, if, at any time in the decoding, there is extrinsic information represented globally by new values ρ(α_i) of the values ρ(α_i), it suffices to replace in FIG. 1 the multiplicative coefficients z^αi by these new coefficients ρ(α_i). The recurrence used is then $r^{-1}\sum _{i=1}^w\underset{\_}{\rho }\left({\alpha }_i\right)S_{r-1}\left(i\right)-\underset{\_}{\rho }\left({\alpha }_j\right)S_{r-1}\left(j\right)=S_r\left(j\right)$where ρ(α_i) represents the ratio between the probability that a_i is equal to −1 and the probability that a_i is equal to +1.

Whatever the way in which the changes in the n values ρ(α_i) corresponding to the n binary elements v_i transmitted, or, equivalently, to the n electrical signals as which represent them, are managed, the estimation of a_i will be −1 if the final value of ρ(α_i) is strictly positive and will be +1 if the final value of ρ(α_i) is strictly negative.

FIG. 3 depicts schematically this type of use of the circuit of FIG. 1, in the case of an implantation of the iterative type. It will not be described in any further detail given that the elements making up the circuit are the same and bear the same reference numbers as the elements already described with the help of FIGS. 1 and 2.

A use of the invention for decoding data using turbocodes is depicted schematically in FIG. 4. In order to simplify, the notations set out in the figure concern only one iteration of the decoding.

From the received sequence (α_i, . . . , α_n), from extrinsic information already calculated ρ(α_i) and from a word h of C^⊥, new extrinsic information ρ_ext[A(i,h)] is calculated for all the positions i covered by h. This is illustrated by an extrinsic information determination block 40 in FIG. 4. Next this extrinsic information ρ_ext[A(i,h)] is recombined with that already calculated ρ(α_i) in order to update this extrinsic information already calculated ρ(α_i). This is illustrated by a recombination and updating block 42.

Concerning the choice of the word h, by way of in no way limitative example, it will for example be possible to use one and the same word h in several non-consecutive iterations. Nevertheless, neither the optimisation of the choice of h, nor the mode of recombining the extrinsic information ρ_ext[A(i,h)] with the information ρ(α_i), when the latter is updated, are within the scope of the present invention. These elements are therefore not described in detail here.

The flow diagram in FIG. 5 illustrates steps of a decoding method according to the present invention.

As shown in FIG. 5, it is assumed that electrical signals a representing respectively binary elements v_i of a word v have previously been transmitted. As described above, the word v is chosen from a binary code C of words v=(v₁, . . . , v_n) satisfying v·h^T=0, where h is a row n-tuplet on the set {0,1} whose number of 1 is denoted w, where ^T represents the transposition and where the scalar product v·h^T is calculated modulo 2.

It is also assumed that a sequence α=(α₁, . . . , α_n) of electrical signals has been received, where the received signal α_i, 1≦i≦n corresponds respectively to the transmitted signal a_i, 1≦i≦n.

A step 500 of initialising the decoding method according to the invention then consists of initialising a variable denoted r to the value 0 and initialising the quantities S_r(i) defined above to the value 1 for any i between 1 and w, that is to say S₀(i)=1∀i, 1≦i≦w.

Then the recurrence on r is effected as follows:

• • an incrementation step 502 consists of increasing the value of r by one unit,
  • an extrinsic information determination step 504 consists of obtaining S_r(i) from the knowledge of S_r−1(i), and this for all the values of i between 1 and w, as described above, and
  • a test 506 consists of determining whether or not the value of r has reached the value w−1.

If test 506 is negative, steps 502, 504 and 506 are reiterated.

If test 506 is positive, this means that r=w−1; S_w−1(i) has then been obtained for i between 1 and w.

As already described, for all the positions i covered by h and for all the values of i between 1 and w, there is then derived therefrom extrinsic information on v_i denoted ρ_ext[A(i,h)] and defined as being equal to P[a_i=−1|A(i,h)]/P[a_i=+1|A(i,h)], where A(i,h) is the set of the received values α_j of α which are covered by h, with the exception of α_i, and where P[a_i|A(i,h)] is the probability, calculated on the basis of the received signals α_j of A(i,h), that the i^th signal transmitted was a_i.

As described above, the extrinsic information is determined by means of the formulaρ_ext[A(i,h)]=[S₁(i)+S₃(i)+ . . . ]/[1+S₂(i)+S₄(i)+ . . . ]where the numbers S_r(i), for any integer r between 1 and w−1, are calculated by applying the recurrence $r^{-1}\sum _{i=1}^w{z}^{{\alpha }_i}{S}_{r-1}\left(i\right)-z^{{\alpha }_j}{S}_{r-1}\left(j\right)=S_r\left(j\right)$to the numbers S₀(i) initialised to 1, with z=exp(−4E/N), where E is the energy of the transmitted signals a_i and N is the spectral power density of the noise on the transmission channel.

FIG. 6 illustrates schematically the constitution of a network station or computer decoding station, in the form of a block diagram.

This station has a keyboard 311, a screen 309, an external information destination 310, and a radio receiver 306, conjointly connected to an input/output port 303 of a processing card 301.

The processing card 301 has, connected together by an address and data bus 302:

• • a central processing unit 300;
  • a random access memory RAM 304;
  • a read only memory ROM 305; and
  • the input/output port 303.

Each of the elements illustrated in FIG. 6 is well known to a person skilled in the art of microcomputers and transmission systems and, more generally, information processing systems. These common elements are therefore not described here. It should however be noted that:

• • the information destination 310 is, for example, an interface peripheral, a display, a modulator, an external memory or other information processing system (not shown), and is advantageously adapted to receive sequences of signals representing speech, service messages or multimedia data, in the form of binary data sequences, and that
  • the radio receiver 306 is adapted to implement a packet transmission protocol on a non-cabled channel, and to receive packets over such a channel.

It should also be noted that the word “register” used in the description designates, in each of the memories 304 and 305, both a memory area of low capacity (a few binary data) and a memory area of large capacity (making it possible to store an entire program).

The random access memory 304 stores data, variables and intermediate processing results, in memory registers bearing, in the description, the same names as the data whose values they store. The random access memory 304 contains notably:

• • a register “received_data”, in which there are stored the binary data received, in their order of arrival on the bus 302 coming from the transmission channel,
  • a register “extrinsic_inf”, in which there are stored, at a given moment, the extrinsic information corresponding to each binary element v_i of a word v, and
  • a register “S_r(i)”, in which there are stored the values of the numbers S_r(i) as they are calculated.

The read only memory 305 is adapted to store the operating program of the central processing unit 300, in a register “Program”.

The central processing unit 300 is adapted to implement the embodiments illustrated by FIGS. 1 to 3 and/or the application to the turbocodes illustrated by FIG. 4 and/or the flow diagram of FIG. 5.

As a variant, the invention could be implemented not only by software method but possibly by using hard-wired or programmable logic.

Claims

1. A method of decoding a received sequence α=(α1, . . . , αn) where, for any integer i between 1 and n, n being an integer greater than 1, αi is the received electrical signal corresponding to the transmission of an electrical signal ai representing the ith binary element vi of a word chosen in a binary code C of words v=(v1, . . . , vn) satisfying v·hT=0, where h is a row n-tuplet on the set {0,1} whose number of 1 is denoted w, where T represents the transposition and the scalar product v·hT is calculated modulo 2, said decoding method including a step consisting of determining extrinsic information on each of the binary elements of v covered by h, said extrinsic information given on the ith binary element of v, assumed to be covered by h, being the quantity ρext[A(i,h)]=P[ai=−1|A(i,h)]/P[ai=+1|A(i,h)],

2. A method according to claim 1, wherein a supplementary item of extrinsic information is determined on each of the binary elements of v covered by h by applying the recurrence r-1⁢∑i=1w⁢ρ_⁢(αi)⁢Sr-1⁢(i)-ρ_⁢(αj)⁢Sr-1⁢(j)=Sr⁢(j)

3. A method according to claim 2, wherein the quantity ρ(αi) is given by ρ(αi)=ρ(αi)·ρext[A(i,h)] where ρ(αi)=P(−1|αi)/P(+1|αi), P(ai|αi) designated the probability that the ith signal transmitted was ai if the ith signal received is αi.

4. A method according to claim 1, 2 or 3, wherein it is implemented in a turbodecoding method.

5. A method according to 1, 2 or 3, wherein the calculations relating to said recurrence are made in multiple precision.

6. Digital signal processing apparatus, having means adapted to implement a decoding method according to claim 1, 2 or 3.

7. A telecommunications network, having means adapted to implement a decoding method according to claim 1, 2 or 3.

8. A mobile station in a telecommunications network, having means adapted to implement a decoding method according to claim 1, 2 or 3.

9. A device for decoding a received sequence α=(α1, . . . , αn) where, for any integer i between 1 and n, n being an integer greater than 1, αi is the received electrical signal corresponding to the transmission of an electrical signal ai representing the ith binary element vi of a word chosen in a binary code C of words v=(v1, . . . , vn) satisfying v·hT=0, where h is a row n-tuplet on the set {0,1} whose number of 1 is denoted w, where T represents the transposition and the scalar product v·hT is calculated modulo 2, said decoding device having means for determining extrinsic information on each of the binary elements of v covered by h, said extrinsic information given on the ith binary element of v, assumed to be covered by h, being the quantity ρext[A(i,h)]P[ai=−1|A(i,h)]/P[ai=+1|A(i,h)],

10. A device according to claim 9, further comprising means for determining an additional item of extrinsic information on each of the binary elements of v covered by h by applying the recurrence r-1⁢∑i=1w⁢ρ_⁢(αi)⁢Sr-1⁢(i)-ρ_⁢(αj)⁢Sr-1⁢(j)=Sr⁢(j)

11. A device according to claim 10, wherein the quantity ρ(αi) is given by ρ(αi)=ρ(αi)·ρext[A(i,h)] where ρ(αi)=P(−1|αi)/P(+1|αi), P(ai|αi) designating the probability that the ith signal transmitted was ai if the ith signal received is αi.

12. A device according to claim 9, 10 or 11, having: a plurality of multipliers, each of said multipliers receiving, at a first input, the value of Sr−1(i) and, on its second input, the value −Zαi; a plurality of adders, a first input of each of said adders being respectively connected to the output of each of said multipliers; adding means, whose input is connected to the output of each of said multipliers; additional multiplication means, a first input of which is connected to the output of said adding means and the second input of which receives the value −1/r, the output of said additional multiplication means being connected to the second input of each of said adders; and delay introduction means, whose input is connected to the output of each of said adders and whose output is connected to the first input of each of said multipliers, said device being initialised by S0(i)=1 for any i, so that each of said adders outputs the value Sr(i).

13. A device according to claim 9, 10 or 11, said device being used in a turbodecoder.

14. A device according to claim 9, 10 or 11, wherein the calculations relating to said recurrence are effected in multiple precision.

15. A device according to claim 9, 10 or 11, wherein the implementation of several of the calculations necessary for determining the extrinsic information is made by circuits put in parallel.

16. A device according to claim 9, 10 or 11, wherein the implementation of several of the calculations necessary for determining the extrinsic information is made by circuits put in series.

17. Digital signal processing apparatus, having a decoding device according to claim 9, 10, or 11.

18. A telecommunications network, having a decoding device according to claim 9, 10 or 11.

19. A mobile station in a telecommunications network, having a decoding device according to claim 9, 10 or 11.

20. A device for processing signals representing speech, including a decoding device according to claim 9, 10 or 11.

21. An information storage means which can be read by a computer or microprocessor, the storage means storing instructions of a computer program that, when implemented, causes the computer to execute a method of decoding a received sequence α=(α1, . . . , αn) where, for any integer i between 1 and n, n being an integer greater than 1, αi is the received electrical signal corresponding to the transmission of an electrical signal ai representing the ith binary element vi of a word chosen in a binary code C of words v=(v1, . . . , vn) satisfying v·hT=0, where h is a row n-tuplet on the set {0,1} whose number of 1 is denoted w, where T represents the transposition and the scalar product v·hT is calculated modulo 2, said decoding method including a step consisting of determining extrinsic information on each of the binary elements of v covered by h, said extrinsic information given on the ith binary element of v, assumed to be covered by h, being the quantity ρext[A(i,h)]=P[ai=−1|A(i,h)]/P[ai=+1A(i,h)],

22. An information storage means according to claim 21, wherein the storage means is partially or totally removable.

23. An information storage means according to claim 22, wherein a supplementary item of extrinsic information is determined on each of the binary elements of v covered by h by applying the recurrence r-1⁢∑i=1W⁢ρ_⁢(αi)⁢Sr-1⁢(i)-ρ_⁢(αj)⁢Sr-1⁢(j)=Sr⁢(j)

24. An information storage means according to claim 22, wherein the quantity ρ(αi) is given by ρ(αi)=ρ(αi)·ρext[A(i,h)] where ρ(αi)=P(−1|αi)/P(+1|αi), P(ai|αi) designated the probability that the ith signal transmitted was ai if the ith signal received is αi.