Frequency compensated communications reception
View Patent ↗Frequency compensated communications reception includes compensating for frequency offset in a received signal by constructing a reference signal for comparison with a training sequence in a received signal. The reference signal is formed from basis functions and the training sequence. It is obtained by minimising a cost function J constructed from an adaptively weighted combination of basis functions, the training sequence, the received signal and a constraint requiring non-zero signal power. Multi-element antenna signals are weighted with a beamforming weight vector w in J given by formula (I), where X is a matrix of received signal samples, C is a diagonal matrix containing elements of the training sequence, F is a matrix having columns defining basis functions, v is a vector of adaptive weights, index H indicates complex conjugate transpose and λ is a Lagrange multiplier constraining beamformer power. A single element antenna signal x is scaled in J given by formula (II), where α is a scaling factor, * indicates a complex conjugate, and x is a vector of received signal samples.
1. A method of compensating for frequency offset in a received signal comprising:
constructing a reference signal comprising a training sequence of the received signal and one or more basis functions;
minimizing a cost function associated with the reference signal, wherein the cost function comprises the training sequence, the one or more basis functions, and the received signal; and
acquiring a desired frequency-shifted signal when the cost function indicates a predetermined degree of correlation between the reference signal and the received signal.
2. The method according to claim 1 wherein a minimum cost function indicates a maximum degree of correlation between the reference signal and the received signal.
3. The method according to claim 1 wherein the cost function further comprises a constraint to obtain non-trivial solutions.
4. The method according to claim 3 , wherein the constraint requires non-zero signal power.
5. The method according to claim 1 , wherein the cost function is J given by:
J=∥Xw−CFv∥ 2 +λ(w H X H Xw−1), wherein X is a matrix of received signal samples, wherein w is a vector of beamforming weights configured to minimize J, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, wherein v is a vector of weights configured to minimize J, wherein superscript index H indicates a complex conjugate transpose, and wherein λ is a Lagrange multiplier for a term to constrain beamformer output power to be non-zero.
6. The method according to claim 5 , further comprising:
determining the vectors w and v at intervals from estimates of a correlation matrix determined from multiple data vectors and from inverses of such estimates recursively updated to reflect successive new data vectors which are rows of the matrix X.
7. The method according to claim 6 , wherein the inverse correlation matrices are recursively updated by:
forming a vector u(n) having a first element u 1 (n) equal to √{square root over (U 1,1 (n))} and other elements u p (n) (p=2 to M) which are respective ratios U p,1 (n)/u 1 (n), wherein U p,1 (n) is a pth element of a first column of a matrix U(n), wherein the matrix U(n)=u(n)u H (n)=x(n)x H (n)−x(n−K+1)x H (n−K+1), wherein x(n) is a most recent data vector, wherein x(n−K+1) is a least recent data vector involved in updating, and wherein x(n)x H (n) and x(n−K+1)x H (n−K+1) are correlation matrices;
premultiplying a previous inverse correlation matrix P(n−1) by vector u H (n) and postmultiplied by the vector u(n) to form a product;
adding the product to a forget factor to form a sum;
postmultiplying the previous inverse correlation matrix P(n−1) by the vector u(n) to form a second product;
dividing the second product by the sum to form a quotient; and
subtracting the quotient from the previous inverse correlation matrix P(n−1) to provide a difference.
8. The method according to claim 1 , wherein the received signal is acquired by a single antenna element.
9. The method according to claim 8 , wherein the cost function is J given by:
J=∥x−CFv∥ 2 , wherein x is a vector of received signal samples, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, and wherein v is a vector of weights configured to minimize J.
10. The method according to claim 8 , wherein the cost function is J given by:
J=∥αx−Gv∥ 2 +λ(α*x H xα−1), wherein α is a scaling factor, wherein x is a vector of received signal samples, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, wherein G is a matrix equal to CF, wherein v is a vector of weights configured to minimize J, wherein superscript index H indicates a complex conjugate transpose, and wherein λ is a Lagrange multiplier for a term to constrain beamformer output power to be non-zero.
11. An apparatus configured to compensate for frequency offset in a received signal, wherein the apparatus comprises:
means for constructing a reference signal comprising one or more basis functions and a training sequence, wherein the means for constructing the reference signal is configured to minimize a cost function comprising the one or more basis functions, the training sequence, and the received signal; and means for acquiring a desired frequency-shifted signal when the cost function indicates a predetermined degree of correlation between the reference signal and the received signal.
12. The apparatus according to claim 11 , wherein the means for acquiring comprises multiple antenna elements.
13. The apparatus according to claim 12 , wherein the cost function further comprises a constraint to obtain a non-trivial solution, and wherein the constraint requires non-zero signal power.
14. The apparatus according to claim 11 , wherein the cost function is J given by:
J=∥Xw−CFV∥ 2 +λ(w H X H Xw−1),wherein X is a matrix of received signal samples, wherein w is a vector of beamforming weights configured to minimize J, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, wherein v is a vector of weights configured to minimize J, wherein superscript index H indicates a complex conjugate transpose, and wherein λ is a Lagrange multiplier for a term to constrain beamformer output power to be non-zero.
15. The apparatus according to claim 14 , further comprising:
means for determining the vectors w and v at intervals from estimates of a correlation matrix determined from multiple data vectors and from inverses of such estimates recursively updated to reflect successive new data vectors which are rows of the matrix X.
16. The apparatus according to claim 14 , further comprising means for recursively updating inverse correlation matrices by:
forming a vector u(n) having a first element u 1 (n) equal to √{square root over (U 1,1 (n))} and other elements u p (n) (p=2 to M) which are respective ratios U p,1 (n)/u 1 (n), wherein U p,1 (n) is a pth element of a first column of a matrix U(n), wherein the matrix U(n)=u(n)u H (n)=x(n)x H (n)−x(n−K+1)x H (n−K+1), wherein x(n) is a most recent data vector, wherein x(n−K+1) is a least recent data vector involved in updating, and wherein x(n)x H (n) and x(n−K+1)x H (n−K+1) are correlation matrices;
premultiplying a previous inverse correlation matrix P(n−1) by vector u H (n) and postmultiplied by the vector u(n) to form a product;
adding the product to a forget factor to form a sum;
postmultiplying the previous inverse correlation matrix P(n−1) by the vector u(n) to form a second product;
dividing the second product by the sum to form a quotient; and
subtracting the quotient from the previous inverse correlation matrix P(n−1) to provide a difference.
17. The apparatus according to claim 11 , wherein the means for acquiring comprises a single antenna element, and wherein the single antenna is configured to product a single output signal for any given sample time.
18. The apparatus according to claim 11 , wherein the cost function is J given by:
J=∥x−CFv∥ 2 , wherein x is a vector of received signal samples, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, and wherein v is a vector of weights configured to minimize J.
19. The apparatus according to claim 11 , wherein the cost function is J given by:
J=∥αx−GV∥ 2 +λ(α*x H xα−1), wherein α is a scaling factor, wherein x is a vector of received signal samples, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, wherein G is a matrix equal to CF, wherein v is a vector of weights configured to minimize J, wherein superscript index H indicates a complex conjugate transpose, and wherein λ is a Lagrange multiplier for a term to constrain beamformer output power to be non-zero.
20. A computer-readable medium having stored thereon computer-executable instructions that, in response to execution by a system, cause the system to perform operations comprising:
constructing a reference signal comprising an original training sequence and a plurality of sinusoidal basis functions;
minimizing a cost function associated with the reference signal, wherein the cost function comprises the original training sequence, the plurality of sinusoidal basis functions, and a received signal; and
acquiring a desired frequency-shifted signal when the cost function indicates a predetermined degree of correlation between the reference signal and the received signal.
21. The computer-readable medium according to claim 20 , wherein the operations further comprise:
constructing a comparison training sequence that is an adaptively formed combination of the plurality of sinusoidal basis functions and the original training sequence.
22. The computer-readable medium according to claim 20 , wherein the cost function further comprises constraint to obtain non-trivial solutions, and wherein the constraint requires non-zero signal power.
23. The computer-readable medium according to claim 20 , wherein the cost function is J given by:
J=∥Xw−CFv∥ 2 +λ(w H X H Xw−1), wherein X is a matrix of received signal samples, wherein w is a vector of beamforming weights configured to minimize J, wherein C is a diagonal matrix having elements of the training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, wherein v is a vector of weights configured to minimize J, wherein superscript index H indicates a complex conjugate transpose, and wherein λ is a Lagrange multiplier for a term to constrain beamformer output power to be non-zero.
24. The computer-readable medium according to claim 23 , wherein the operations further comprise:
determining the vectors w and v at intervals from estimates of a correlation matrix determined from multiple data vectors and from inverses of such estimates recursively updated to reflect successive new data vectors which are rows of the matrix X.
25. The computer-readable medium according to claim 23 , wherein the operations further comprise recursively updating inverse correlation matrices by:
forming a vector u(n) having a first element u 1 (n) equal to √{square root over (U 1,1 (n))} and other elements u p (n) (p=2 to M) which are respective ratios U p,1 (n)/u 1 (n), wherein U p,1 (n) is a pth element of a first column of a matrix U(n), wherein the matrix U(n)=u(n)u H (n)=x(n)x H (n)−x(n−K+1)x H (n−K+1), wherein x(n) is a most recent data vector, wherein x(n−K+1) is a least recent data vector involved in updating, and wherein x(n)x H (n) and x(n−K+1)x H (n−K+1) are correlation matrices;
premultiplying a previous inverse correlation matrix P(n−1) by vector u H (n) and postmultiplied by the vector u(n) to form a product;
adding the product to a forget factor to form a sum;
postmultiplying the previous inverse correlation matrix P(n−1) by the vector u(n) to form a second product;
dividing the second product by the sum to form a quotient; and
subtracting the quotient from the previous inverse correlation matrix P(n−1) to provide a difference.
26. The computer-readable medium according to claim 20 , wherein the received signal is acquired by a receiver comprising a single antenna element.
27. The computer-readable medium according to claim 26 , wherein the cost function is J given by:
J=∥x−CFv∥ 2 , wherein x is a vector of received signal samples, wherein C is a diagonal matrix having elements of the original training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, and wherein v is a vector of weights configured to minimize J.
28. The computer-readable medium according to claim 26 , wherein the cost function is J given by:
J=∥αx−Gv∥ 2 +λ(α*x H xα−1), wherein α is a scaling factor, wherein x is a vector of received signal samples, wherein C is a diagonal matrix having elements of the original training sequence on its diagonal, wherein F is a matrix having columns defining respective basis functions, wherein G is a matrix equal to CF, wherein v is a vector of weights configured to minimize J, wherein superscript index H indicates a complex conjugate transpose, and wherein λ is a Lagrange multiplier for a term to constrain beamformer output power to be non-zero.
29. The method according to claim 1 , wherein the one or more basis functions comprise a sinusoid.
30. The apparatus according to claim 11 , wherein the one or more basis functions span a subspace in which a complex sinusoid associated with the frequency offset lies.