METHOD FOR RECOVERING A SPARSE COMMUNICATION SIGNAL FROM A RECEIVE SIGNAL

Information

  • Patent Application
  • 20160248611
  • Publication Number
    20160248611
  • Date Filed
    April 29, 2016
    8 years ago
  • Date Published
    August 25, 2016
    8 years ago
Abstract
The patent application relates to a method for recovering a sparse communication signal from a receive signal, the receive signal being a channel output version of the sparse communication signal, the channel comprising channel coefficients being arranged to form a channel matrix, the method comprising determining a support set indicating a set of first indices of non-zero communication signal coefficients from the channel matrix and the receive signal, determining an estimate of the sparse communication signal upon the basis of the support set, the channel matrix and the receive signal, determining second indices of communication signal coefficients which are not indicated by the support set, and determining the sparse communication signal upon the basis of the support set, the estimate of the sparse communication signal, the second indices and the channel matrix.
Description
TECHNICAL FIELD

The present patent application relates to the field of sparse signal recovery, in particular to performance and quality aspects.


BACKGROUND

Sparse signal recovery problems arise in many different applications. In communication systems, the recovery of sparse communication signals from receive signals is of increasing interest.


An n-dimensional communication signal or vector is called sparse if it comprises only a few non-zero entries among the total number of entries n. Sparse communication signals can be recovered from a fewer number of measurements comparing with the total dimension n. Sparseness can appear in different bases, so communication signals can, for example, be represented as sparse communication signals in time domain or in frequency domain.


The recovery of sparse communication signals is usually more computationally extensive than a matrix inverse or pseudo-inverse involved in least squares computations. Moreover, the matrix is usually not invertible, as the system is under-determined.


Common methods for sparse communication signal recovery suffer from high computational complexity and/or low performance. These methods are usually not suitable for use in real time systems.


In E. J. Candes, and T. Tao, “Decoding by linear programming,” IEEE Trans. on Information Theory, vol. 51, no. 12, pp. 4203-4215, December 2005, a linear programming approach is proposed and analyzed.


In T. Cai, L. Wang, “Orthogonal Matching Pursuit for Sparse Signal Recovery With Noise,” IEEE Trans. on Information Theory, v. 57, N. 7, pp. 4680-4688, the problem of sparse signal recovery from incomplete measurements is considered applying an orthogonal matching pursuit method.


SUMMARY

It is one object of the patent application to provide an efficient scheme for recovering a sparse communication signal from a receive signal.


This object is achieved by the features of the independent claims. Further implementation forms are apparent from the dependent claims, the description and the figures.


The patent application is based on the finding that an estimate of the sparse communication signal can be refined in order to determine the sparse communication signal.


According to a first aspect, the patent application relates to a method for recovering a sparse communication signal from a receive signal, the receive signal being a channel output version of the sparse communication signal, the channel comprising channel coefficients being arranged to form a channel matrix, the method comprising determining a support set indicating a set of first indices of non-zero communication signal coefficients from the channel matrix and the receive signal, determining an estimate of the sparse communication signal upon the basis of the support set, the channel matrix and the receive signal, determining second indices of communication signal coefficients which are not indicated by the support set, and determining the sparse communication signal upon the basis of the support set, the estimate of the sparse communication signal, the second indices and the channel matrix. Thus, the sparse communication signal can be recovered efficiently.


The sparse communication signal can be represented by a vector. The sparse communication signal can comprise communication signal coefficients. The communication signal coefficients can be real numbers, e.g., 1.5 or 2.3, or complex numbers, e.g., 1+j or 5+3j.


The receive signal can be represented by a vector. The receive signal can comprise receive signal coefficients. The receive signal coefficients can be real numbers, e.g., 1.3 or 2.7, or complex numbers, e.g., 2+3j or 1−5j.


The channel can define a linear relationship between the sparse communication signal and the receive signal. The channel can comprise additive noise. The channel coefficients can be real numbers, e.g., 0.3 or 1.5, or complex numbers, e.g., 3−2j or 1+4j.


The support set can indicate a set of indices of non-zero communication signal coefficients.


In a first implementation form according to the first aspect as such, the method further comprises updating the support set upon the basis of the second indices of communication signal coefficients. Thus, the support set can be refined for further processing.


The updating of the support set can relate to adding second indices of communication signal coefficients to the support set, or subtracting second indices of communication signal coefficients from the support set.


In a second implementation form according to the first aspect as such or the first implementation form of the first aspect, the method further comprises determining a cardinality of the support set or the updated support set, the cardinality indicating a number of elements of the support set or the updated support set. Thus, a sparsity measure for the sparse communication signal can be derived.


In a third implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the sparse communication signal comprises communication signal coefficients and the receive signal comprises receive signal coefficients, and the number of communication signal coefficients is greater than the number of receive signal coefficients. Thus, a reduced number of receive signal coefficients can be used for performing the method.


In a fourth implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the determining of the support set or the determining of the estimate of the sparse communication signal is performed by an orthogonal matching pursuit method. Thus, an efficient method for determining the support set and/or determining the estimate of the sparse communication signal can be applied.


The orthogonal matching pursuit method can be adapted to optimize a linear objective function, subject to linear equality constraints and/or linear inequality constraints. An orthogonal matching pursuit method is described e.g. in T. Cai, L. Wang, “Orthogonal Matching Pursuit for Sparse Siynal Recovery With Noise,” IEEE Trans. on Information Theory, v. 57, N. 7, pp. 4680-4688.


In a fifth implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the determining of second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and the dual linear programming problem is defined by the following equations:





e≦HTz≦e,






y
T
Z→min,


wherein H denotes the channel matrix, y denotes the receive signal, e denotes an all unit vector, and z denotes a dual variable. Thus, the second indices of communication signal coefficients can be determined in an optimized manner.


In a sixth implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the determining of second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and the residual is defined by:





ΔS*,j+=1+HjTzS* or ΔS*,j=1−HjTzS*





with zS*=−HS*(HS*THS*)−1eS*,






e
S*=(eS*,1,eS*,2, . . . ,eS*,l*)T,






e
S*,i=1 if i∈S+*,






e
S*,i=−1 if i∈S*, and






S*=S
+
*∪S
*,


wherein S* denotes the support set being divided into two subsets according to signs of the entries of the vector xS*=(HS*THS*)−1HS*Ty, S+* denotes a subset of the support set S* corresponding to the positive entries of xS*, S* denotes a subset of the support set S* corresponding to the negative entries of xS*, eS* denotes an auxiliary vector, eS*,i denote coefficients of the auxiliary vector eS* with index i, HS* denotes a channel matrix comprising columns indicated by the support set S*, zS* denotes an estimate solution of a dual linear programming problem, Hj denotes a channel matrix comprising a column indicated by index j, and Δ+S*,j or ΔS*,j denote the residual. Thus, the second indices of communication signal coefficients can be determined in an optimized manner.


In a seventh implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the determining of second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and the determining of second indices of communication signal coefficients is performed according to the following equations:








j

(
1
)


=


argmin






Δ


S
*

,
j

+






or






j

(
1
)



=


argmin
j



Δ


S
*

,
j

-




,




wherein ΔS*,j+ or ΔS*,j denote a residual of a dual linear programming problem, and j(1) denotes the second indices of communication signal coefficients. Thus, the second indices of communication signal coefficients can be determined in an optimized manner.


In an eighth implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the determining of the sparse communication signal is performed according to the following equation:






x
S

(1)

=x
S*−α(HS*THS*)−1HS*THj(1),α>0


wherein HS* denotes a channel matrix comprising columns indicated by the support set S*, Hj(1), denotes a channel matrix comprising columns indicated by second indices of communication signal coefficients j(1), α denotes a predetermined value, xS* denotes an estimate of the sparse communication signal, and xS(1) denotes the sparse communication signal. Thus, the sparse communication signal can be determined efficiently.


The predetermined value α can be a real number, e.g. 0.2, or a complex number, e.g. 1+j.


In a ninth implementation form according to the first implementation form of the first aspect, the updating of the support set is performed according to the following equation:






S
(1)
=S*∪{j
(1)
}\{j
i

(1)
},


wherein j(1) denotes second indices of communication signal coefficients, ji(1) denotes indices of communication signal coefficients to be removed from the support set, i(1) denotes a number of an entry to be removed from the support set, S* denotes the support set, and S(1) denotes the updated support set. Thus, the support set can be refined in an optimized manner.


In a tenth implementation form according to the first aspect as such or any one of the preceding implementation forms of the first aspect, the determining of the support set, the determining of the estimate of the sparse communication signal, the determining of the second indices of communication signal coefficients, the determining of the sparse communication signal, the updating of the support set, and the determining of the cardinality are successively repeated until a stopping criterion is met. Thus, the performance of the method can be improved.


In an eleventh implementation form according to the tenth implementation form of the first aspect, the sparse communication signal corresponding to the support set with least cardinality is provided as the recovered sparse communication signal. Thus, the sparsity of the recovered sparse communication signal can be optimized.


In a twelfth implementation form according to the tenth implementation form of the first aspect or the eleventh implementation form of the first aspect, the stopping criterion is a positive value or a zero value of all residuals of a dual linear programming problem, and a residual is defined by:





ΔS*,j+=1+HjTzS* or ΔS*,j=1−HjTzS*





with zS*=−HS*(HS*THS*)−1eS*,






e
S*=(eS*,1,eS*,2, . . . ,eS*,l*)T,






e
S*,i=1 if i∈S+*,






e
S*,i=−1if i∈S*, and






S*=S
+
*∪S
*,


wherein S* denotes the support set being divided into two subsets according to signs of the entries of the vector xS*=(HS*THS*)−1HS*Ty, S+* denotes a subset of the support set S* corresponding to the positive entries of xS*, S* denotes a subset of the support set S* corresponding to the negative entries of xS*, eS* denotes an auxiliary vector, eS*,i denote coefficients of the auxiliary vector eS* with index i, HS* denotes a channel matrix comprising columns indicated by the support set S*, zS* denotes an estimate solution of a dual linear programming problem, Hj denotes a channel matrix comprising a column indicated by index j, and ΔS*,j+ or ΔS*,j the residual. Thus, the stopping criterion can be based on a performance criterion.


In a thirteenth implementation form according to the tenth implementation form of the first aspect to the twelfth implementation form of the first aspect, the stopping criterion is an expiry of a length of a predetermined time interval. Thus, the stopping criterion enables an application of the method in real time systems.


The length of the predetermined time interval can be e.g., 1 ms, 5 ms, or 10 ms.


According to a second aspect, the patent application relates to a computer program for performing the method according to the first aspect as such or any one of the preceding implementation forms of the first aspect when executed on a computer. Thus, the method can be performed in an automatic and repeatable manner.


The computer program can be provided in form of a machine-readable code. The computer program can comprise a series of commands for a processor of the computer.


The computer can comprise a processor, a memory, an input interface, and/or an output interface. The processor of the computer can be configured to execute the computer program.


The patent application can be implemented in hardware and/or software.





BRIEF DESCRIPTION OF THE DRAWINGS

Further embodiments of the patent application will be described with respect to the following figures, in which:



FIG. 1 shows a schematic diagram of a method for recovering a sparse communication signal from a receive signal; and



FIG. 2 shows a schematic diagram of a time slot allocated for recovering a sparse communication signal from a receive signal.





DETAILED DESCRIPTION


FIG. 1 shows a schematic diagram of a method 100 for recovering a sparse communication signal from a receive signal.


The receive signal is a channel output version of the sparse communication signal. The channel comprises channel coefficients being arranged to form a channel matrix.


The method 100 comprises determining 101 a support set indicating a set of first indices of non-zero communication signal coefficients from the channel matrix and the receive signal, determining 103 an estimate of the sparse communication signal upon the basis of the support set, the channel matrix and the receive signal, determining 105 second indices of communication signal coefficients which are not indicated by the support set, and determining 107 the sparse communication signal upon the basis of the support set, the estimate of the sparse communication signal, the second indices and the channel matrix.


The sparse communication signal can be represented by a vector. The sparse communication signal can comprise communication signal coefficients. The communication signal coefficients can be real numbers, e.g., 1.5 or 2.3, or complex numbers, e.g., 1+j or 5+3j.


The receive signal can be represented by a vector. The receive signal can comprise receive signal coefficients. The receive signal coefficients can be real numbers, e.g., 1.3 or 2.7, or complex numbers, e.g., 2+3j or 1−5j.


The channel can define a linear relationship between the sparse communication signal and the receive signal. The channel can comprise additive noise. The channel coefficients can be real numbers, e.g., 0.3 or 1.5, or complex numbers, e.g., 3−2j or 1+4j.


The support set can indicate a set of indices of non-zero communication signal coefficients.



FIG. 2 shows a schematic diagram of a time slot allocated for recovering a sparse communication signal from a receive signal.


The problem of sparse signals recovery from incomplete measurements can be solved using OMP method and its modifications. Further, there are linear programming approaches LP. However, there are no attempts to combine the OMP and LP into the one numerical scheme.


The n-dimensional vector 0 is called sparse if it has only few non-zero entries among total number n. Sparse compressible vectors can be restored from the fewer number of measurements comparing with total dimension n. Sparseness can appear in different bases, so signals can be represented as sparse vectors in time domain or in frequency domain.


Reconstruction of sparse signals is normally more computationally extensive than just measurement matrix inverse or pseudo-inverse involved in the least squares computations. Moreover, the measurement matrix is usually not invertible, as the system is underdetermined.


Numerically efficient reconstruction algorithms should be suitable for use in real time. The linear system






Hx=y,  (1)


where H∈Rm×n, x∈Rn, y∈Rm is under-determined if m<n. More precisely, the system (1) is under-determined if rank(H)<n. This kind of systems has infinitely many solutions. The pattern of the system can be schematically represented as described with the following matrix-vector pattern:








[







































































]

×

[





































]


=

[

















]





The trade-off for reducing the number of measurements is sparseness of the vectors x. Sparseness means that zeros dominate among the entries of the vector x. The methods of search for most sparse solutions among solutions of the system (1) are the key problem considered in the compressive sensing (CS) theory.


The key problem considered in the (CS) theory can be reformulated as reconstruction of the correct support set, corresponding to the most sparse vector, i.e. the set of significant non-zero entries:





Supp(x)={i:xi≠0}.


It has been shown that while





x∥0→min,






Hx=y,  (P0)


is the correct formulation of the problem for determination of the most sparse solution, where ∥x∥0=Supp(x), the relaxed form of this problem





x∥p→min,






Hx=y,  (Pp)


is also sufficient, where









x


p

=


(




i
=
1

n





x


i
p


)


1
p






is the lp-norm of the vector, p>0. Non-convex nature of this problem makes it hard for numerical solution. One of the popular methods for determination of the sparsest solution, i.e. solution having largest number of zero entries, consists in solution of the problem (P1) which is linear programming problem (LP):





x∥l→min,






Hx=y,  (P1), (LP)


where ∥x∥l is the l1-norm of the vector,









x


1

=




i
=
1

n






x
i



.






That is one of the standard methods exploited in the Compressive Sensing (CS) theory.


Its advantages are convexity and applicability of all known convex optimization methods, including linear programming methods such as the simplex method and good performance if compared with the best possible, but NP-hard solution of the problem (P0).


A more efficient solution for real time implementation is the Orthogonal Matching Pursuit (OMP) algorithm. Its key idea is formulated as follows.

    • Step 0: Initialize the residual r(0)=y, and initialize the estimate of the support set S(0)=Ø.
    • Step k=1,2, . . . . Find the column Hj(k) of the matrix H that solves the problem








j

(
k
)


=


argmax
j







H
j
T



r

(

k
-
1

)








H
j






,






    • and update S(k)=S(k-1)∪{j(k)}, r(k)=HS(k)x(k)−y, where HS(k) is the sub-matrix of H consisting of those Hj, with indices j∈S(k). Steps are repeated until stopping criterion is satisfied









r
(k)

T

r
(k)≦σ2.  (SC)


That is a kind of so-called greedy algorithms, the index included into the support set is not taken out of the set. The advantage of this OMP algorithm is its high computational complexity if compared with other methods. The OMP algorithm shows lower performance results when reconstructing the sparse signal if compared with results obtained by (P1).


The mentioned problem can be solved by combining the OMP method and linear programming method (P1) into one numerical algorithm.


A structure of the time slot allocated to the problem solution is shown in FIG. 2. Let the time slot allocated to the sparse recovery problem be TΣ and OMP iterations take TOMP which is less than TΣ as shown. So, we have time TΣ-TOMP for improvements considering the OMP solution as an initial step.


Terminating the process immediately after the time slot TΣ is exhausted, as much as possible is gained compared with conventional OMP making improvements between TOMP and TΣ.


Consider the LP problem induced by the l1 minimization that can be equivalently represented as






Hu−Hv=y,u≧0,v≧0,






e
T
u+e
T
v→min,  (PLP)


where e is the vector consisting of all units, having appropriate size. Here x=u−v.


This linear programming problem will be referred to as the primary LP (PLP). Along with (PLP) consider the dual linear programming problem (DLP):





e≦HTz≦e,






y
T
z→min,  (DLP)


where z stands for dual variable.


Suppose that OMP has completed resulting in:






S*−the last estimate of the support set obtained by OMP,l*=card(S*)






H
S*
∈R
m×l*,






x
S*=(HS*THS*)−1HS*Ty∈Rl*.


The set S* can be further divided into two subsets according to signs of the entries of the vector xS*:






S*=S
+
*∪S
*





Let define:






e
S*=(eS*,1,eS*,2, . . . ,eS*,l*)T,






e
S*,i=1 if i∈S+*,






e
S*,i=−1 if i∈S*.


Generally it is l*≦m and so, the matrix HS* is not square.


The current estimate of the DLP solution, corresponding to the basis S*, can be calculated as the minimum norm solution to the under-determined system






H
S*
T
z=−e
S*
,∥z∥
2
2→min,


which has the closed form solution






z
S*
=−H
S*(HS*THS*)−1eS*  (2)


This vector corresponds to the problem (DLP) that it can be either feasible or not feasible to the system of linear inequalities. It can be either optimal or not optimal to (DLP). Optimality conditions hold, as the set S* is obtained at the last step of the first stage (OMP). So, feasibility conditions of the vector x* obtained at the last iteration of OMP are ‘almost satisfied’ due to (SC). But (PLP) is dual to (DLP) and feasibility conditions for (PLP) are optimality conditions to (DLP).


So, only feasibility conditions of the vector zS* are checked. Check residuals of (DLP) for inequalities j∉JS* to see if there are violated conditions or, what is the same, residuals with incorrect sign.


Residuals of (DLP) are




ΔS*,j+=1+HjTzS*,





ΔS*,j=1−HjTzS*.  (3)


Both ΔS*,j+ and ΔS*,j can be non-negative for all j. If this condition is satisfied, both (PLP) and (DLP) are solved with solutions xS* and zS* respectively. Let one of the residuals be negative. Without loss of generality, let











j

(
1
)


=


argmin
j



Δ


S
*

,
j

+








and




(
4
)







Δ


S
*

,

j

(
1
)



+

<
0.




(
5
)







Then a new solution






x
S

(1)
)=xS*−α(HS*THS*)−1HS*THj(1),α>0  (6)


can be generated with the following l1 improvement property:





xS(1)1<∥xS*1,






S
(1)
=S*∪{j
(1)
}\{j
i

(1)
},  (7)


i(1) is the number of the entry to be removed from the basis (the serial number in the series 1,2, . . . , l*).


The Enhanced OMP algorithm can be summarized as follows:














 1) k = 0;


 while(allocated time slot is not exhausted)


 {










1)
perform OMP (or any of its modifications,  10);



2)
Start  an  l1 improvement  step:  performing  calculations







 according to (2)-(4);










3)
Check if  the condition (5) is satisfied. If not,







terminate the algorithm as









(DLP) is feasible and so, (PLP) has an optimal







solution. Otherwise, perform











next step;



4)
correct set of ‘active’ indices and recalculate the







 solution according to (6),









(7); The index removed from the basis is not inserted







back, so the











modification (7) is ‘greedy’;



5)
Keep the record value of card(S(k)) (or l1-norm): solution







 corresponding to









the least cardinality (or least l1) is recorded to be







 reported as a result;










6)
k := k + 1







 }









An n-dimensional vector x is sparse if it has only few non-zero entries among total number n.


A sparse compressible vector can be restored from fewer number of measurements comparing with total dimension n.


Sparseness can appear in different bases, so signals can be represented as sparse vectors in time domain or in frequency domain.


Reconstruction of sparse signals is normally more computationally extensive than just measurement matrix inverse. Moreover, the measurement matrix is usually not invertible, as the system is underdetermined.


It is desirable for numerically efficient reconstruction algorithms to be suitable for use in real time.


Recovery of sparse signals from under-determined or incomplete measurement model is can be performed. The linear system






Hx=y


where H∈Rm×n,y∈Rm,x∈Rn and m<n is under-determined. Having the pattern








[







































































]

×

[





































]


=

[

















]





it has infinitely many solutions: Null(H)+x0 is a set of solutions together with every single solution x0. To find the sparsest solution, i.e. solution having largest number of zero entries, can be found as l1 minimizer:





x∥1→min subjected to constraints Hx=y


which is a linear program (LP). That is standard problem setup exploited in the Compressive


Sensing (CS) theory.


LP is computationally extensive. Much easier for real time implementation is the Orthogonal Matching Pursuit (OMP) algorithm. Its key idea is formulated as follows.


Step 0: Initialize the residual r0=y and estimate of the support set S0=Ø.


Step k=1, . . .: Find the column Hjk of the matrix H that solves the problem







j
k

=


argmax
j







H
j
T



r

k
-
1






H
j











and update Sk=Sk-1∪{jk}, rk=y−Hxk, where xk=(HSkTHSk)−1HSkTy, HSk denotes submatrix of the matrix H corresponding to the set of indices sk.


Steps are repeated until stopping criterion is satisfied.


Suppose the measurement model is described by the equation y=Hx+∈, x∈Rn, y∈Rm with number of measurements m significantly less than the length of the binary word x.


Suppose that not more than k entries of the vector x are non-zero. In other words n,






x
i=0 for i∉supp(x),xi≠0 for i∈supp(x),card(supp(x))=k.


Moreover, the matrix H satisfies the Restricted Isometry Property (RIP) property and the following conditions hold








m
~
k







log


(

n
k

)



,


E


(


ɛ
T


ɛ

)


=


σ
2

.






Stopping criterion for OMP can be chosen like






r
k
T
r
k≦σ2.  (SC)


The key idea is to combine advantages of OMP and LP approaches, considering OMP as a first stage of the Enhanced OMP (EOMP), and making several iterations of modified simplex method after (SC) is met, considering Sk as an initial basis, attempting to sequentially modify its content.


The combined algorithm is greedy: the basis is only extended at the first OMP stage; the index removed from the basis at the second stage is not inserted again. Matrix pseudoinverse is modified using low rank modifications of matrix factorizations when performing the second stage.


Consider the LP problem induced by the l1 minimization. It can be equivalently represented as






Hu−Hv=y,u≧0,v≧0,






e
T
u+e
T
v→min  (PLP)


where e is the vector consisting of all units, having appropriate size. Here x=u−v. This LP will be referred to as the Primary LP (PLP). Along with (PLP) consider the Dual LP (DLP)





e≦HTz≦e,






y
T
z→min  (DLP)


where z stands for dual variable.


Suppose that OMP has completed resulting in:






S*—the last estimate of the support set obtained by OMP,l*=card(S*)






H
S*
∈R
m×l*,






x
S*=(HS*THS*)−1HS*Ty∈Rl*.


The set S* can be further divided into two subsets according to signs of the entries of the vector xS*.






S*=S
+
*∪S
*





Let define






e
S*=(eS*,1,eS*,2, . . . ,eS*,l*)T,






e
S*,i=1 if i∈S+*,






e
S*,i=−1 if i∈S*.


The current estimate of the DLP solution, corresponding to the basis S*, can be calculated as the minimum norm solution to the under-determined system






H
S*
z=−e
S*
,∥z∥
2→min,


which has the closed form solution






z
S*
=−H
S*(HS*THS*)−1eS*.


This vector corresponds to the problem (DLP) in that it can be either feasible or not feasible to the system of linear inequalities. It can be either optimal or not optimal to (DLP).


Optimality conditions hold, as the set S* is obtained at the last step of the first stage (OMP). So, feasibility conditions are ‘almost satisfied’ due to (SC). But (PLP) is dual to (DLP) and feasibility conditions for (PLP) are optimality conditions to (DLP). So, only feasibility conditions can be checked.


Let JS* be the set of columns indices of the matrix corresponding to the basis S*. Check residuals of (DLP) for inequalities j∉JS*.


Residuals of (DLP) are




ΔS*,j+=1+HjTzS*,





ΔS*,j=1−HjTzS*.


Both ΔS*,j+ and ΔS*,j can be non-negative. If so both (PLP) and (DLP) are solved with solutions xS* and zS* respectively. Let one of residuals be negative. Without loss of generality, let ΔS*,j(1)+<0. Then new solution






x
S

(1)

=x
S*−α(HS*THS*)−1HS*THj(1),α>0


can be generated with the following l1 improvement property:





|xS(1)|<|xS*|,






S
(1)
=S*∪{j
(1)
}\{j
i

(1)
}


i(1) is the number of the entry to be removed from the basis.


Finally, the algorithm can be summarized as follows:

















while(DLP constraints are violated and number of iterations did



not exceed a limit)



{



  perform OMP;



  make an l1 improvement step;



  correct set of ‘active’ indexes: index removed from the



basis is not inserted back;



  keep record value of card(S(l)): solution corresponding to the



least cardinality is  reported.



}










In an implementation form, the patent application relates to an enhanced orthogonal matching pursuit for sparse signal recovery.


In an implementation form, the patent application relates to sparse signal detection, especially aspects of its performance quality.


In an implementation form, the patent application relates to a method for recovery of sparse signal combining OMP with one or several l1 improvement steps comprising, performing OMP, computation of solution of the problem (DLP) according to equations (2), (3), checking residuals (4), (5), making improvement (6), (7), and keeping the record solution.


In an implementation form, the record solution is the best in terms of the cardinality.


In an implementation form, the record solution is the best in terms of the l1 criterion value.


In an implementation form, the termination condition is feasibility of (DLP) (5).


In an implementation form, the termination condition is exhausting of the time allocated to the problem.


In an implementation form, the patent application allows for better performance if compared to OMP and/or LP.


In an implementation form, low rank matrix factorizations can be applied at each step of the algorithm to reduce the computational load.


In an implementation form, the patent application relates to a method for recovery of sparse signals combining advantages of OMP and LP approaches.

Claims
  • 1. A method for recovering a sparse communication signal from a receive signal, the receive signal being a channel output version of the sparse communication signal, the channel comprising channel coefficients being arranged to form a channel matrix, the method comprising: determining a support set indicating a set of first indices of non-zero communication signal coefficients from the channel matrix and the receive signal;determining an estimate of the sparse communication signal upon the basis of the support set, the channel matrix and the receive signal;determining second indices of communication signal coefficients which are not indicated by the support set; anddetermining the sparse communication signal upon the basis of the support set, the estimate of the sparse communication signal, the second indices and the channel matrix.
  • 2. The method according to claim 1, further comprising: updating the support set upon the basis of the second indices of communication signal coefficients.
  • 3. The method according to claim 1, further comprising: determining a cardinality of the support set or the updated support set, the cardinality indicating a number of elements of the support set or the updated support set.
  • 4. The method according to claim 1, wherein the sparse communication signal comprises communication signal coefficients and the receive signal comprises receive signal coefficients, and wherein the number of communication signal coefficients is greater than the number of receive signal coefficients.
  • 5. The method according to claim 1, wherein determining the support set or determining the estimate of the sparse communication signal is performed by an orthogonal matching pursuit method.
  • 6. The method according to claim 1, wherein determining the second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and wherein the dual linear programming problem is defined by the following equations: −e≦HTz≦e, yTZ→min,wherein H denotes the channel matrix, y denotes the receive signal, e denotes an all unit vector, and z denotes a dual variable.
  • 7. The method according to claim 1, wherein determining the second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and wherein the residual is defined by: ΔS*,j+=1+HjTzS* or ΔS*,j−=1−HjTzS* with zS*=−HS*(HS*THS*)−1eS*,eS*=(eS*,1,eS*,2, . . . ,eS*,l*)T,eS*,i=1 if i∈S+*,eS*,i=−1 if i∈S−*, andS*=S+*∪S−*,wherein S* denotes the support set being divided into two subsets according to signs of the entries of the vector xS*=(HS*THS*)−1HS*Ty, S+* a subset of the support set S* corresponding to the positive entries of xS*, S−* denotes a subset of the support set S* corresponding to the negative entries of xS*, eS*, denotes an auxiliary vector, eS*,i denote coefficients of the auxiliary vector eS* with index i, HS* denotes a channel matrix comprising columns indicated by the support set S*, zS* denotes an estimate solution of a dual linear programming problem, Hj denotes a channel matrix comprising a column indicated by index j, and ΔS*,j+ or ΔS*,j− denote the residual.
  • 8. The method according to claim 1, wherein determining the second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and wherein the determining of second indices of communication signal coefficients is performed according to the following equations:
  • 9. The method according to claim 1, wherein determining the sparse communication signal is performed according to the following equation: xS(1)=xS*−α(HS*THS*)−1HS*THj(1),α>0wherein HS* denotes a channel matrix comprising columns indicated by the support set S*, Hj(1) denotes a channel matrix comprising columns indicated by second indices of communication signal coefficients j(1), α denotes a predetermined value, xS* denotes an estimate of the sparse communication signal, and xS(1) denotes the sparse communication signal.
  • 10. The method according to claim 2, wherein updating the support set is performed according to the following equation: S(1)=S*∪{j(1)}\{ji(1))},wherein j(1) denotes second indices of communication signal coefficients, ji(1) denotes indices of communication signal coefficients to be removed from the support set, i(1) denotes a number of an entry to be removed from the support set, S* denotes the support set, and S(1) denotes the updated support set.
  • 11. The method according to claim 1, wherein determining the support set, determining the estimate of the sparse communication signal, determining the second indices of communication signal coefficients, determining the sparse communication signal, the updating of the support set, and determining the cardinality are successively repeated until a stopping criterion is met.
  • 12. The method according to claim 11, wherein the sparse communication signal corresponding to the support set with least cardinality is provided as the recovered sparse communication signal.
  • 13. The method according to claim 11, wherein the stopping criterion is a positive value or a zero value of all residuals of a dual linear programming problem, and wherein a residual is defined by: ΔS*,j+=1+HjTzS* or ΔS*,j−=1−HjTzS* with zS*=−HS*(HS*THS*)−1eS*,eS*=(eS*,1,eS*,2, . . . ,eS*,l*)T,eS*,i=1 if i∈S+*,eS*,i=−1 if i∈S−*, andS*=S+*∪S−*,wherein S* denotes the support set being divided into two subsets according to signs of the entries of the vector xS*=(HS*THS*)−1HS*Ty, S+* a subset of the support set S* corresponding to the positive entries of xS*, S−* denotes a subset of the support set S* corresponding to the negative entries of xS*, eS* denotes an auxiliary vector, eS*,i denote coefficients of the auxiliary vector eS* with index i, HS* denotes a channel matrix comprising columns indicated by the support set S*, zS* denotes an estimate solution of a dual linear programming problem, Hj denotes a channel matrix comprising a column indicated by index j, and ΔS*,j+ or ΔS*,j− denote the residual.
  • 14. The method according to claim 11, wherein the stopping criterion is an expiry of a length of a predetermined time interval.
  • 15. A computer program for performing the method according to claim 1 when executed on a computer.
  • 16. The method according to claim 2, further comprising: determining a cardinality of the support set or the updated support set, the cardinality indicating a number of elements of the support set or the updated support set.
  • 17. The method according to claim 2, wherein the sparse communication signal comprises communication signal coefficients and the receive signal comprises receive signal coefficients, and wherein the number of communication signal coefficients is greater than the number of receive signal coefficients.
  • 18. The method according to claim 2, wherein determining the second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and wherein the dual linear programming problem is defined by the following equations: −e≦HTz≦e, yTz→min,wherein H denotes the channel matrix, y denotes the receive signal, e denotes an all unit vector, and z denotes a dual variable.
  • 19. The method according to claim 1, wherein determining the second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and wherein the residual is defined by: ΔS*,j+=1+HjTzS* or ΔS*,j−=1−HjTzS* with zS*=−HS*(HS*THS*)−1eS*,eS*=(eS*,1,eS*,2, . . . ,eS*,l*)T,eS*,i=1 if i∈S+*,eS*,i=−1 if i∈S−*, andS*=S+*∪S−*,wherein S* denotes the support set being divided into two subsets according to signs of the entries of the vector xS*=(HS*THS*)−1HS*Ty, S+* denotes a subset of the support set S* corresponding to the positive entries of xS*, S−* denotes a subset of the support set S* corresponding to the negative entries of xS*, eS* denotes an auxiliary vector, eS*,i denote coefficients of the auxiliary vector eS* with index i, HS* denotes a channel matrix comprising columns indicated by the support set S*, zS* denotes an estimate solution of a dual linear programming problem, Hj denotes a channel matrix comprising a column indicated by index j, and ΔS*,j+ or ΔS*,j− denote the residual.
  • 20. The method according to claim 2, wherein determining the second indices of communication signal coefficients comprises minimizing a residual of a dual linear programming problem, and wherein the determining of second indices of communication signal coefficients is performed according to the following equations:
  • 21. The method according to claim 2, wherein determining the sparse communication signal is performed according to the following equation: xS(1)=xS*−α(HS*THS*)−1HS*THj(1),α>0wherein HS* denotes a channel matrix comprising columns indicated by the support set S*, Hj(1) denotes a channel matrix comprising columns indicated by second indices of communication signal coefficients j(1), α denotes a predetermined value, xS* denotes an estimate of the sparse communication signal, and xS(1) denotes the sparse communication signal.
CROSS-REFERENCE TO RELATED APPLICATIONS

This application is a continuation of International Application No. PCT/RU2013/000974, filed on Nov. 1, 2013, which is hereby incorporated by reference in its entirety.

Continuations (1)
Number Date Country
Parent PCT/RU2013/000974 Nov 2013 US
Child 15142731 US