Method for detection of symbols in communication signals

1. A method for detection of symbols carried in received communication signals, the method comprising:

receiving at least one communication signal y transmitted over a radio channel H, wherein said received signal y comprises encoded radio channel symbol inputs x;

receiving a priori Log Likelihood Ratios (LLRs) of said channel inputs x;

computing soft symbol estimates μ and variance D of said channel inputs x based on LLRs of said channel inputs x;

filtering and interference cancelling said received signal y by using a first filter and filtered soft symbol estimates μ to obtain a filtered and interference cancelled received signal, wherein a kth element of said filtered and interference cancelled received signal has a form r k =E(x k |y,μ l ), l∉I k , where E is the expectation operator and I k is a set of indices that depends on k in a pre-determined fashion, and wherein the vector comprising r k , ∀k , is denoted as E(x|y,μ);

estimating the Mean Square Error (MSE) matrix B of said filtered and interference cancelled received signal;

computing an optimal matrix G opt for a Bahl-Cook-Jelinke-Raviv (BCJR) algorithm based on said MSE estimate B, wherein said optimal matrix G opt specifies the BCJR algorithm;

further filtering said filtered and interference cancelled received signal by using a second filter to obtain an input signal r to the BCJR algorithm, wherein the second filter is derived from said optimal matrix G opt ; and

feeding said input signal r into the BCJR algorithm to obtain extrinsic LLRs of said channel inputs x.

2. The method according to claim 1 , wherein the LLRs of said channel inputs x are obtained from a decoder, and wherein the method further comprises:

feeding said extrinsic LLRs into the decoder to obtain updated LLRs of said channel inputs x; and

repeating said computing the soft symbol estimates μ and variance D, filtering and interference cancelling, estimating the MSE matrix B, computing the optimal matrix G opt , filtering said filtered and interference cancelled received signal, feeding said input signal r, and feeding said extrinsic LLRs into the decoder i number of times by using said updated LLRs to compute soft symbol estimates μ and variance D.

3. The method according to claim 1 , wherein the first filter is a modified Wiener filter for said radio channel inputs x having mean μ and variance D.

4. The method according to claim 1 , wherein the second filter has a form (G opt +I), where I is the identity matrix.

5. The method according to claim 1 , wherein said radio channel H is a linear radio channel.

6. The method according to claim 1 , wherein the communication signal is transmitted in a cellular wireless communication system.

7. A method for detection of symbols carried in received communication signals, the method comprising:

receiving at least one communication signal y transmitted over a radio channel H, wherein said received signal y comprises encoded radio channel symbol inputs x;

receiving a priori Log Likelihood Ratios (LLRs) of said channel inputs x;

computing soft symbol estimates μ and variance D of said channel inputs x based on LLRs of said channel inputs x, wherein the LLRs of said channel inputs x are obtained from a decoder;

filtering and interference cancelling said received signal y by using a first filter and filtered soft symbol estimates μ to obtain a filtered and interference cancelled received signal, wherein a kth element of said filtered and interference cancelled received signal has a form r k =E(x k |y,μ l ), l∉I k , where E is the expectation operator and I k is a set of indices that depends on k in a pre-determined fashion, and wherein the vector comprising r k , ∀k , is denoted as E(x|y,μ);

estimating the Mean Square Error (MSE) matrix B of said filtered and interference cancelled received signal;

computing an optimal matrix G opt for a Bahl-Cook-Jelinke-Raviv (BCJR) algorithm based on said MSE estimate B, wherein said optimal matrix G opt specifies the BCJR algorithm;

further filtering said filtered and interference cancelled received signal by using a second filter to obtain an input signal r to the BCJR algorithm, wherein the second filter is derived from said optimal matrix G opt ; and

feeding said input signal r into the BCJR algorithm to obtain extrinsic LLRs of said channel inputs x l ;

feeding said extrinsic LLRs into the decoder to obtain updated LLRs of said channel inputs x; and

repeating said computing the soft symbol estimates μ and variance D, filtering and interference cancelling, estimating the MSE matrix B, computing the optimal matrix G opt , filtering said filtered and interference cancelled received signal, feeding said input signal r, and feeding said extrinsic LLRs into the decoder i number of times by using said updated LLRs to compute soft symbol estimates μ and variance D.

8. The method according to claim 7 , wherein said set of indices I k is I k =I k ={k−L, k−L+1, . . . k+L−1, k+L}∩{1, 2, . . . , N}, where N is the total number of elements in said channel inputs x, and wherein L is the memory of the BCJR algorithm.

9. The method according to claim 8 , wherein said optimal matrix G opt is solely dependent on said MSE estimate B and the memory L.

10. The method according to claim 8 , wherein the memory L is arbitrarily chosen.

11. The method according to claim 7 , wherein said MSE estimate B has a form B(μ)=E [(x−E(x|y,μ))(x−E(x|y,μ))*], where “*” is the Hermitian transpose operator.

12. The method according to claim 7 , wherein said input signal r to said BCJR algorithm is obtained as: r=(G opt +I)E(x|y,μ).

13. A non-transitory computer readable medium having computer-executable instructions stored thereon for detection of symbols carried in received communication signals, the computer executable instructions comprising instructions for:

receiving at least one communication signal y transmitted over a radio channel H, wherein said received signal y comprises encoded radio channel symbol inputs x;

receiving a priori Log Likelihood Ratios (LLRs) of said channel inputs x;

computing soft symbol estimates μ and variance D of said channel inputs x based on LLRs of said channel inputs x;

filtering and interference cancelling said received signal y by using a first filter and filtered soft symbol estimates μ to obtain a filtered and interference cancelled received signal, wherein a kth element of said filtered and interference cancelled received signal has a form r k =E(x k |y,μ l ), l∉I k , where E is the expectation operator and I k is a set of indices that depends on k in a pre-determined fashion, and wherein the vector comprising r k , ∀k , is denoted as E(x|y,μ);

estimating the Mean Square Error (MSE) matrix B of said filtered and interference cancelled received signal;

computing an optimal matrix G opt for a Bahl-Cook-Jelinke-Raviv (BCJR) algorithm based on said MSE estimate B, wherein said optimal matrix G opt specifies the BCJR algorithm;

further filtering said filtered and interference cancelled received signal by using a second filter to obtain an input signal r to the BCJR algorithm, wherein the second filter is derived from said optimal matrix G opt ; and

feeding said input signal r into the BCJR algorithm to obtain extrinsic LLRs of said channel inputs x.

14. A detection device arranged for detection of symbols carried in received communication signals, the detection device comprising a processor arranged to:

receive at least one communication signal y transmitted over a radio channel H, wherein said received signal y comprises encoded radio channel symbol inputs x;

receive a priori Log Likelihood Ratios (LLRs) of said channel inputs x;

compute soft symbol estimates μ and variance D of said channel inputs x based on LLRs of said channel inputs x;

filter and interference cancel said received signal y by using a first filter and filtered soft symbol estimates μ to obtain a filtered and interference cancelled received signal, wherein a kth element of said filtered and interference cancelled received signal has a form r k =E(x k |y,μ l ), l∉I k , where E is the expectation operator and I k is a set of indices that depends on k in a pre-determined fashion, and wherein the vector comprising r k , ∀k, is denoted as E(x|y,μ);

estimate the Mean Square Error (MSE) matrix B of said filtered and interference cancelled received signal;

compute an optimal matrix G opt for a Bahl-Cook-Jelinke-Raviv (BCJR) algorithm based on said MSE estimate B, wherein said optimal matrix G opt specifies the BCJR algorithm;

further filter said filtered and interference cancelled received signal by using a second filter to obtain an input signal r to the BCJR algorithm, wherein the second filter is derived from said optimal matrix G opt ; and

feed said input signal r into the BCJR algorithm to obtain extrinsic LLRs of said channel inputs x.

15. The detection device according to claim 14 , wherein the LLRs of said channel inputs x are obtained from a decoder, and wherein the processor is further arranged to:

feed said extrinsic LLRs into the decoder to obtain updated LLRs of said channel inputs x; and

repeat said compute the soft symbol estimates μ and variance D, filter and interference cancel, estimate the MSE matrix B, compute the optimal matrix G opt , filter said filtered and interference cancelled received signal, feed said input signal r, and feed said extrinsic LLRs into the decoder i number of times by using said updated LLRs to compute soft symbol estimates μ and variance D.