COMMUNICATION SYSTEM

• US 20110142181A1
• Filed: 11/09/2010
• Published: 06/16/2011
• Est. Priority Date: 11/09/2009
• Status: Abandoned Application

• Current Assignee: BAR ILAN University
• Original Assignee: BAR ILAN University
• Inventors: Goldberger, Jacob, Leshem, Amir
• Application Number:
  US12/942,352
• Publication Number:
  US 20110142181A1
• Time in Patent Office: Days
• US Class Current: 375/341
• CPC Class Codes: H04L 2025/03426 transmission using multiple-input and multiple-output channels
• CPC Class Codes: H04L 25/03318 Provision of soft decisions
• CPC Class Codes: H04L 25/03968 mean-square error [MSE]

Abstract

A communication system having a program of machine-readable instructions for solving an ILS problem, tangibly embodied on a computer readable memory and executable by a digital data processor, to perform actions directed toward outputting a set of a-posteriori probability vectors.

Claims

1. A communication system having a program of machine-readable instructions for solving an ILS problem, tangibly embodied on a computer readable memory and executable by a digital data processor, to perform actions directed toward outputting a set of a-posteriori probability vectors, the actions comprising steps of:

a. receiving an input vector x on a plurality of channels;

b. estimating a channel matrix H via a channel tracking unit;

c. computing matrices H_m,n by a projection generation unit;

d. projecting said input vector x onto a low-dimensional subspace via a vector projecting unit, said vector projecting unit;

e. computing the projections z₁, . . . , z_d via a one-dimensional unit;

f. computing a prior probability vectors for each coordinate of a solution vector s via a prior probability computation unit; and

then

g. calculating said set of a-posteriori probability vectors for each coordinate by performing iterations with low-dimensional computations via a belief propagation unit.

2. The communication system of claim 1, wherein said low-dimensional subspace is a two-dimensional subspace.

3. The communication system of claim 1, wherein said final estimated vector is calculated by choosing the maximum a-posteriori probability among all the alphabet symbols.

4. The communication system of claim 1, wherein said prior probability vectors are calculated according to a solution technique selected from the group consisting of:

a linear solution, L-MMSE, zero-forcing, V-BLAST, and any combination thereof.

5. The communication system of claim 1, wherein said the actions further comprise a step of sphere decoding said set of a-posteriori probability vectors.

6. The communication system of claim 1, wherein said actions are performed independently for each tone.

7. The communication system of claim 1, wherein said communication system is a MIMO communication system.

8. A method for solving an ILS problem in a MIMO communication system having a program of machine-readable instructions, tangibly embodied on a computer readable memory and executable by a digital data processor, comprising steps of:

a. receiving an input vector x on a plurality of channels;

b. estimating a channel matrix H via a channel tracking unit;

c. computing matrices H_m,n by a projection generation unit;

d. projecting said input vector x onto a low-dimensional subspace via a vector projecting unit, said vector projecting unit;

e. computing the projections z₁, . . . , z_d via a one-dimensional unit;

f. computing a prior probability vectors for each coordinate of a solution vector s via a prior probability computation unit; and

then

g. calculating a set of a-posteriori probability vectors for each coordinate by performing iterations with low-dimensional computations via a belief propagation unit.

9. The method of claim 8, further comprising a step of providing said low-dimensional subspace in a two-dimensional subspace.

10. The method of claim 8, further comprising a step of calculating said final estimated vector by choosing the maximum a-posteriori probability among all the alphabet symbols.

11. The method of claim 8, further comprising a step of calculating said prior probability vectors according to a solution technique selected from the group consisting of:

a linear solution, L-MMSE, zero-forcing, V-BLAST, and any combination thereof.

12. The method of claim 8, further comprising a step of sphere decoding said set of a-posteriori probability vectors.

13. The method of claim 8, further comprising a step of performing said actions independently for each tone.

14. The method of claim 8, further comprising a step of configuring said communication system as a MIMO communication system.

15. A communication system having a program of machine-readable instructions for solving a detection problem according to a PPBP algorithm, tangibly embodied on a computer readable memory and executable by a digital data processor, to perform actions directed toward outputting a set of a-posteriori probability vectors, the actions comprising steps of:

a. receiving a tap gain matrix H_mxn;

b. performing a MMSE linear detection;

c. providing a max-product initialization m<sub>j->

i</sub>(x_i)=prior_i(x_i);

d. calculating the belief propagation for generating probabilities per symbol for each transmitted symbol, thereby providing belief_i(x_i); and

,

e. calculating the PPBP solution vector x_i by choosing the maximum a-posteriori probability among all alphabet symbols according to the equation;

x_i=argmax<sub>xi∈

A</sub>belief_i(x_i).

16. The communication system of claim 15, further adapted to perform at least one action selected from a group consisting of:

calculating the MMSE_SIC solution;

comparing the performance of said MMSE_SIC solution to said PPBP solution; and

selecting the preferred solution from said MMSE_SIC solution and said PPBP solution or any combination thereof.

17. The communication system of claim 15, wherein said actions further comprise a step of sphere decoding to prune the search according to a-posteriori symbol probabilities.

18. The communication system of claim 15, wherein said actions are performed independently for each tone.

19. The communication system of claim 15, wherein said actions are performed for the calculation of log-likelihood probabilities for each bit.

20. The communication system of claim 15, wherein said communication system is a MIMO communication system.

21. A method for solving a MIMO detection problem according to a PPBP algorithm in a MIMO communication system having a program of machine-readable instructions, tangibly embodied on a computer readable memory and executable by a digital data processor, comprising steps of:

a. receiving a tap gain matrix H_mxn and a measured vector y;

b. performing a MMSE linear detection;

c. providing a max-product initialization m<sub>j->

i</sub>(x_i)=prior_i(x_i);

d. calculating the belief propagation for generating probabilities per symbol for each transmitted symbol, thereby providing belief_i(x_i); and

then

e. calculating the PPBP solution vector x_i by choosing the maximum a-posteriori probability among all alphabet symbols according to the equation;

x_i=argmax<sub>xi∈

A</sub>belief_i(x_i).

22. The method of claim 21, further comprising a step of performing actions selected from a group consisting of:

calculating the MMSE_SIC solution;

comparing the performance of said MMSE_SIC solution to said PPBP solution; and

selecting the preferred solution from said MMSE_SIC solution and said PPBP solution or any combination thereof.

23. The method of claim 21, further comprising a step of sphere decoding to prune the search according to a-posteriori symbol probabilities.

24. The method of claim 21, further comprising a step of performing said actions independently for each tone.

25. The method of claim 21, further comprising a step of performing said actions for the calculation of log-likelihood probabilities for each bit.

26. The method of claim 21, further comprising a step of providing said communication system as a MIMO communication system.

27. A communication system having a program of machine-readable instructions for solving an ILS problem with a Belief Propagation (BP) paradigm, tangibly embodied on a computer readable memory and executable by a digital data processor, to perform actions directed toward outputting a set of a-posteriori probability vectors, the actions comprising steps of:

a. receiving a tap gain matrix H_mxn and a measured vector y;

b. applying a Chow-Liu algorithm on the distribution corresponding to the unconstrained linear system;

c. applying a finite-set constraint and utilizing the Gaussian tree distribution to form a discrete loop free approximation of p(x|y); and

then

d. applying BP on the loop free Markov Random Field (MRF).

28. The communication system of claim 27, wherein said actions further comprise a step of calculating:

29. The communication system of claim 27, wherein said step (d) applying BP on the loop free Markov Random Field (MRF) is performed on:

30. The communication system of claim 27, wherein said communication system is a MIMO communication system.

31. A method for solving an ILS problem with a Belief Propagation (BP) paradigm in a MIMO communication system having a program of machine-readable instructions, tangibly embodied on a computer readable memory and executable by a digital data processor, comprising steps of:

a. receiving a tap gain matrix H_mxn and a measured vector y;

b. applying a Chow-Liu algorithm on the distribution corresponding to the unconstrained linear system;

c. applying a finite-set constraint and utilizing the Gaussian tree distribution to form a discrete loop free approximation of p(x|y); and

then

d. applying BP on the loop free Markov Random Field (MRF).

32. The method according to claim 31, further comprising a step of calculating said actions according to a formula:

33. The method of claim 31, wherein said step (d) applying BP on the loop free Markov Random Field (MRF) is performed on:

34. The method of claim 31, further comprising a step of providing said communication system as a MIMO communication system.