Max-log stack decoder

1. A method for demodulating signals in a multi-input multi-output (MIMO) receiver, comprising:

a. obtaining a transformed vector by a coordinate transformation of a received observation vector using a unitary matrix determined through QR decomposition of an estimated channel matrix;

b. maintaining a list containing nodes along with a cost metric for each node;

c. using the list to generate soft-outputs in the form of log-likelihood ratios (LLRs) for selected symbols of interest, based on the transformed vector and a lower triangular matrix determined through QR decomposition of the estimated channel matrix and the constellations to which the input symbols belong.

2. The method of claim 1 , wherein the QR decomposition of the estimated channel matrix comprises H=QL where Q is semi unitary (Q*Q=I); and

L

=

[

L

1

,

1

0

…

0

L

2

,

1

L

2

,

2

⋱

0

⋮

⋮

⋱

0

L

K

,

1

L

K

,

2

…

L

K

,

K

]

is the lower triangular matrix with positive diagonal elements.

3. The method of claim 1 , comprising determining the transformed vector z as: z=Q*y, where y is the vector of received observations.

4. The method of claim 1 , comprising selecting a node with a minimum metric among all nodes in the list and removing the selected node from the list.

5. The method of claim 4 , comprising determining whether the selected node is a leaf node.

6. The method of claim 4 , comprising examining the selected node using an LLR module, if the selected node is a leaf node.

7. The method of claim 4 , comprising determining whether the selected node is the first leaf node to be examined by the LLR module and if so, determining the bits corresponding to the selected node and otherwise determining first the bits corresponding to the selected node and then determining an LLR for each index for which the bit of the selected node is different from the bit of the first examined leaf node.

8. The method of claim 1 , comprising expanding the descendents of a selected partial or non-leaf node and adding the descendents and their metrics to the list.

9. The method of claim 1 , comprising removing all nodes and their metrics from the list which have the same values for symbols of interest as those in a selected leaf node.

10. The method of claim 1 , comprising analyzing at-most a predetermined number of leaf nodes using an LLR module.

11. The method of claim 10 , comprising removing non-leaf nodes from the list to leave only leaf-nodes and their metrics in a pruned list.

12. The method of claim 10 , comprising generating a pruned list if more than a predetermined number of LLRs have been generated or if more than a predetermined number of leaf nodes have been analyzed.

13. The method of claim 1 , comprising forcing the list to only contain nodes whose associated metrics are less than a specified input parameter C.

14. The method of claim 13 , comprising adding only children to the list whose associated metrics are no greater than C.

15. The method of claim 13 , wherein the input parameter C is incremented if the list becomes empty before an LLR analysis of D leaf nodes, where D is also a predetermined selection.

16. The method of claim 1 , comprising determining whether all required LLRs have been generated by checking if a set Ω is empty.

17. The method of claim 12 , wherein if the set Ω is not empty, comprising determining the magnitude of the remaining LLRs as a large positive number.

18. The method of claim 17 , wherein the large positive number equals the magnitude of the largest generated LLR.

19. The method of claim 1 , comprising generating soft-outputs in the form of LLRs for a specified subset of the input quadrature amplitude modulation (QAM) symbols.

20. The method of claim 19 , comprising:

identifying a root node and the child of the root node having the smallest metric; and

selecting from the children of all nodes in a list, a best child having the smallest metric.

21. The method of claim 19 , comprising determining a best valid node among all nodes in a list and removing the best valid node from the list.

22. The method of claim 21 , comprising determining for each selected leaf node a valid sibling with the smallest metric and adding the sibling to a list.

23. The method of claim 22 , comprising determining for each selected non-leaf node a valid sibling with the smallest metric and a valid child with the smallest metric and adding them to a list while adding the selected node to another list.

24. A demodulator in a multi-input multi-output (MIMO) receiver, comprising:

a. means for obtaining a transformed vector by a coordinate transformation of a received observation vector using a unitary matrix determined through QR decomposition of an estimated channel matrix;

b. means for maintaining a list containing nodes along with a cost metric for each node;

c. means for using the list to generate soft-outputs in the form of log-likelihood ratios (LLRs) for selected symbols of interest, based on the transformed vector and a lower triangular matrix determined through QR decomposition of the estimated channel matrix and the constellations to which the input symbols belong.

25. A wireless system, comprising:

a multi-input multi-output (MIMO) base station; and

a MIMO receiver adapted to communicate with the MIMO base station, comprising:

a. means for obtaining a transformed vector by a coordinate transformation of a received observation vector using a unitary matrix determined through QR decomposition of an estimated channel matrix;

b. means for maintaining a list containing nodes along with a cost metric for each node;

c. means for using the list to generate soft-outputs in the form of log-likelihood ratios (LLRs) for selected symbols of interest, based on the transformed vector and a lower triangular matrix determined through QR decomposition of the estimated channel matrix and the constellations to which the input symbols belong.