Iterative interference cancellation using mixed feedback weights and stabilizing step sizes

• US 7,715,508 B2
• Filed: 06/13/2006
• Issued: 05/11/2010
• Est. Priority Date: 11/15/2005
• Status: Active Grant

• Current Assignee: III Holdings 1, LLC (Intellectual Ventures LLC)
• Original Assignee: TensorComm, Inc. (Rambus Inc.)
• Inventors: McCloud, Michael L, Lamba, Gagandeep Singh, Guess, Tommy, Nagarajan, Vijay
• Primary Examiner(s): KIM, KEVIN
• Application Number:
  US11/452,027
• Publication Number:
  US 20070110134A1
• Time in Patent Office: 1,428 Days
• Field of Search: 375/346, 375/347
• US Class Current: 375/346
• CPC Class Codes: H04B 1/7097 Interference-related aspects
• CPC Class Codes: H04B 1/71075 Parallel interference cancellation
• CPC Class Codes: H04B 1/712 Weighting of fingers for combining, e.g. amplitude control or phase rotation using an inner loop
• CPC Class Codes: H04B 2201/70702 Intercell-related aspects

Abstract

A receiver is configured for canceling intra-cell and inter-cell interference in coded, multiple-access, spread-spectrum transmissions that propagate through frequency-selective communication channels. The receiver employs iterative symbol-estimate weighting, subtractive cancellation with a stabilizing step-size, and mixed-decision symbol estimates. Receiver embodiments may be implemented explicitly in software or programmed hardware, or implicitly in standard Rake-based hardware either within the Rake (i.e., at the finger level) or outside the Rake (i.e., at the user or subchannel symbol level).

Claims

1. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal, and

a stabilizing step size module configured for scaling the error signal by a stabilizing step size to produce a scaled error signal.

2. The canceller recited in claim 1, further comprising a combining module configured for adding at least one of the constituent signals to the scaled error signal to produce one or more interference-cancelled signals.

6. The canceller recited in claim 1, further comprising a resolving module configured to resolve the scaled error signal onto the signal basis for all symbol sources in the channel to produce a resolved error signal, and a combining module configured to combine the resolved error signal with the weighted input symbol estimates to produce the interference-cancelled symbol estimates.

7. The canceller recited in claim 1, wherein the stabilizing step size module is configured to produce a scalar stabilizing step size for improving convergence of a plurality of iterations.

3. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal, and

wherein the constituent signals are characterized by at least one of a set of signals, the set comprising received signals associated with a plurality of Rake fingers and received signals associated with a plurality of symbol sources.

4. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal,

the canceller further configured for performing iterative interference cancellation in a receiver, wherein the synthesizing module is configured to synthesize the constituent signals to approximate estimated received signals employing the signal basis for all sources in the channel and the plurality of weighted or unweighted symbol estimates.

5. The canceller recited in claim 4, wherein the synthesizing module is configured to synthesize a subset of the sources for producing the combined signal.

8. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal, and

a decision module configured to produce a decision for a plurality of the symbol estimates, the decision selected from one of a hard decision quantizing at least one of the symbol estimates to a nearby constellation point, and a soft decision scaling, but not quantizing, the at least one of the symbol estimates,

wherein the decision module comprises a de-bias module configured to remove bias, as computed on the symbol estimates, before performing the decision.

9. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal, and

the canceller configured for performing synthesis and resolution onto the signal basis for all symbol sources in the channel in an outer feedback loop, and updating symbol estimates in an inner feedback loop employing an implementation matrix.

10. The canceller recited in claim 9, wherein synthesis and resolution comprise constructing the signal basis from source, channel, and receiver effects.

11. The canceller recited in claim 9, wherein the implementation matrix comprises one of an identity matrix and a transmit-signal correlation matrix.

12. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal, and

the canceller configured for performing symbol-level interference cancellation on despread signals with a direct implementation of a one-step matrix update equation expressing an explicit matrix representation of user correlations, an implementation matrix, a soft-weighting matrix, a de-biasing matrix, and a scalar stabilizing step size.

13. The canceller recited in claim 12, wherein the one-step matrix update equation is expressed by

{circumflex over (b)}^[i+1]=Ψ

{D^[i](μ

^[i](q−

RΓ

^[i]{circumflex over (b)}^[i])+FΓ

^[i]{circumflex over (b)}^[i])},

wherein {circumflex over (b)}^[i] is a column vector containing the input symbol estimates after an i^th iteration of interference cancellation, q is a vector of the despread received signal, μ

^[i] is a scalar stabilizing step, Γ

^[i] is a weighting matrix used for scaling the input symbol estimates, R is a received-signal correlation matrix, F is an implementation matrix, D^[i] is a de-biasing matrix, and Ψ

returns a mixed (hard- or soft-) decision for each element of its input vector to produce the plurality of updated symbol estimates {circumflex over (b)}^[i+1].

14. An interference canceller for cancelling interference in a received signal from a channel, the canceller configured for performing at least one iteration to produce at least one interference-cancelled symbol estimate, the canceller comprising:

a synthesizing module configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and a subtraction module configured to subtract the combined signal from the received signal to produce an error signal, and

the canceller configured as an internal feedback loop and an external feedback loop;

the external feedback loop configured to multiply a vector of the input symbol estimates by a weighting matrix and a users'"'"' received correlation matrix to produce an external-feedback vector, and subtracting the external-feedback vector from the despread signal vector for producing a difference error vector that is scaled by a scalar stabilizing step size; and

the internal feedback loop configured to multiply a vector of the plurality of weighted symbol estimates by an implementation matrix to produce an internal-feedback vector, the internal-feedback vector being summed with a vector derived from the scaled error signal.

15. The canceller recited in claim 14, wherein the external feedback loop is configured to multiply the input symbol estimate, {circumflex over (b)}^[i], by both the weighting matrix Γ

^[i] and the users'"'"' received correlation matrix R to produce the external-feedback vector, RΓ

^[i]{circumflex over (b)}^[i] which is subtracted from the despread signal vector q in order to produce the difference error vector, q−

RΓ

^[i]{circumflex over (b)}^[i]; and

the internal feedback loop is configured to multiply the weighted symbol estimates, Γ

^[i]{circumflex over (b)}^[i], by the implementation matrix F to produce the internal-feedback vector, FΓ

^[i]{circumflex over (b)}^[i], which is summed with the difference error vector produced by the external feedback loop, q−

RΓ

^[i]{circumflex over (b)}^[i], that has been scaled by the stabilizing step size μ

^[i], to produce FΓ

^[i]{circumflex over (b)}^[i]+μ

^[i](q−

RΓ

^[i]{circumflex over (b)}^[i]).

16. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and providing for subtracting the combined signal from the received signal to produce an error signal, and

providing for scaling the error signal by a stabilizing step size to produce a scaled error signal.

17. The method recited in claim 16, further comprising providing for adding the constituent signals to the scaled error signal to produce one or more interference-cancelled signals.

21. The method recited in claim 16, further comprising providing for combining and resolving configured to resolve the scaled error signal onto the signal basis for all symbol sources in the channel to produce a resolved error signal, and combine the resolved error signal with the weighted input symbol estimates to produce the interference-cancelled symbol estimates.

22. The method recited in claim 16 , wherein providing for scaling is configured to produce a scalar stabilizing step size for improving convergence of a plurality of iterations.

18. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and providing for subtracting the combined signal from the received signal to produce an error signal,

wherein the constituent signals are characterized by at least one of a set of signals, the set comprising received signals associated with a plurality of Rake fingers and received signals associated with a plurality of symbol sources.

19. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and providing for subtracting the combined signal from the received signal to produce an error signal; and

performing iterative interference cancellation in a receiver, wherein providing for employing a signal basis for sources in the channel to synthesize constituent signals is configured to synthesize the constituent signals to approximate estimated received signals employing the signal basis for all sources in the channel and the plurality of weighted or unweighted symbol estimates.

20. The method recited in claim 19, wherein providing for employing a signal basis for sources in the channel to synthesize constituent signals is configured to synthesize a subset of the sources for producing the combined signal.

23. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and

providing for subtracting the combined signal from the received signal to produce an error signal, and

providing for producing a decision for a plurality of symbol estimates, the decision selected from one of a hard decision quantizing at least one of the symbol estimates to a nearby constellation point, and a soft decision, scaling, but not quantizing, the at least one of the symbol estimates,

wherein providing for producing a decision comprises removing bias computed on the symbol estimates before producing the decision.

24. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and

producing a combined signal therefrom, and providing for subtracting the combined signal from the received signal to produce an error signal,

the method configured for performing synthesis and resolution onto the signal basis for all symbol sources in the channel in an outer feedback loop, and updating symbol estimates in an inner feedback loop employing an implementation matrix.

25. The method recited in claim 24, wherein performing synthesis and resolution comprises constructing the signal basis from source, channel, and receiver effects.

26. The method recited in claim 24, wherein the implementation matrix comprises one of an identity matrix and a transmit-signal correlation matrix.

27. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and

producing a combined signal therefrom, and providing for subtracting the combined signal from the received signal to produce an error signal,

the method configured for performing symbol-level interference cancellation on despread signals with a direct implementation of a one-step matrix update equation expressing an explicit matrix representation of user correlations, an implementation matrix, a soft-weighting matrix, a de-biasing matrix, and a scalar stabilizing step size.

28. The method recited in claim 27, wherein the one-step matrix update equation is expressed by

{circumflex over (b)}^[i+1]=Ψ

{D^[i](μ

^[i](q−

RΓ

^[i]{circumflex over (b)}^[i])+FΓ

^[i]{circumflex over (b)}^[i])},

wherein {circumflex over (b)}^[i] is a column vector containing the input symbol estimates after an i^th iteration of interference cancellation, q is a vector of the despread received signal, μ

^[i] is a scalar stabilizing step, Γ

^[i] is a weighting matrix used for scaling the input symbol estimates, R is a received-signal correlation matrix, F is an implementation matrix, D^[i] is a de-biasing matrix, and Ψ

returns a mixed (hard- or soft-) decision for each element of its input vector to produce the plurality of updated symbol estimates {circumflex over (b)}^[i+1].

29. An interference cancellation method for cancelling interference in a received signal from a channel, the method configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the method comprising:

providing for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and

producing a combined signal therefrom, and providing for subtracting the combined signal from the received signal to produce an error signal,

the method configured for employing an internal feedback loop process and an external feedback loop process;

the external feedback loop process comprising multiplying a vector of the input symbol estimates by a weighting matrix and a users'"'"' received correlation matrix to produce an external-feedback vector, and subtracting the external-feedback vector from the despread signal vector for producing a difference error vector that is scaled by a scalar stabilizing step size; and

the internal feedback loop process comprising multiplying a vector of the plurality of weighted symbol estimates by an implementation matrix to produce an internal-feedback vector, and summing the internal-feedback vector with a vector derived from the scaled error signal.

30. The method recited in claim 29, wherein

the external feedback loop process comprises multiplying the input symbol estimate, {circumflex over (b)}^[i], by both the weighting matrix Γ

^[i] and the users'"'"' received correlation matrix R to produce the external-feedback vector, RΓ

^[i]{circumflex over (b)}^[i], which is subtracted from the despread signal vector q in order to produce the difference error vector, q−

RΓ

^[i]{circumflex over (b)}^[i]; and

the internal feedback loop process comprising multiplying the weighted symbol estimates, Γ

^[i]{circumflex over (b)}^[i], by the implementation matrix F to produce the internal-feedback vector, FΓ

^[i]{circumflex over (b)}^[i], and summing the internal-feedback vector with the difference error vector produced by the external feedback loop process, q−

RΓ

^[i]{circumflex over (b)}^[i], that has been scaled by the stabilizing step size μ

^[i], to produce FΓ

^[i]{circumflex over (b)}^[i]+μ

^[i](q−

RΓ

^[i]{circumflex over (b)}^[i]).

31. An interference cancellation system for cancelling interference in a received signal from a channel, the system configured for performing at least one iteration to produce an interference-cancelled symbol estimate, the system comprising:

a synthesizing means configured for employing a signal basis for sources in the channel to synthesize constituent signals from a plurality of weighted or unweighted symbol estimates, and producing a combined signal therefrom, and

a subtraction means configured to subtract the combined signal from the received signal to produce an error signal, and

a stabilizing step size means configured for scaling the error signal by a stabilizing step size to produce a scaled error signal.

32. The system recited in claim 31, further comprising a combining means configured for adding the constituent signals to the scaled error signal to produce one or more interference-cancelled signals.

33. The system recited in claim 31, wherein the constituent signals are characterized by at least one of a set of signals, the set comprising received signals associated with a plurality of Rake fingers and received signals associated with a plurality of symbol sources.

34. The system recited in claim 31, further configured for performing iterative interference cancellation in a receiver, wherein the synthesizing means is configured to synthesize the constituent signals to approximate estimated received signals employing the signal basis for all sources in the channel and the plurality of weighted or unweighted symbol estimates.

35. The system recited in claim 34, wherein the synthesizing means is configured to synthesize a subset of the sources for producing the combined signal.

36. The system recited in claim 31, further comprising a resolving means configured to resolve the scaled error signal onto the signal basis for all symbol sources in the channel to produce a resolved error signal, and a combining means configured to combine the resolved error signal with the weighted input symbol estimates to produce the interference-cancelled symbol estimates.

37. The system recited in claim 31, wherein the stabilizing step size means is configured to produce a scalar stabilizing step size for improving convergence of a plurality of iterations.

38. The system recited in claim 31, further comprising

a decision means configured to produce a decision for a plurality of symbol estimates, the decision selected from one of a hard decision quantizing at least one of the symbol estimates to a nearby constellation point, and a soft decision, scaling, but not quantizing, the at least one of the symbol estimates,

wherein the decision means comprises a de-biasing means configured to remove bias, as computed on the symbol estimates, before producing the decision.

39. The system recited in claim 31, configured for performing synthesis and resolution onto the signal basis for all symbol sources in the channel in an outer feedback loop, and updating symbol estimates in an inner feedback loop employing an implementation matrix.

40. The system recited in claim 39, wherein synthesis and resolution comprise constructing the signal basis from source, channel, and receiver effects.

41. The system recited in claim 39, wherein the implementation matrix comprises one of an identity matrix and a transmit-signal correlation matrix.

42. The system recited in claim 31, configured for performing symbol-level interference cancellation on despread signals with a direct implementation of a one-step matrix update equation expressing an explicit matrix representation of user correlations, an implementation matrix, a soft-weighting matrix, a de-biasing matrix, and a scalar stabilizing step size.

43. The system recited in claim 42, wherein the one-step matrix update equation is expressed by

{circumflex over (b)}^[i+1]=Ψ

{D^[i](μ

^[i](q−

RΓ

^[i]{circumflex over (b)}^[i]) +FΓ

^[i]{circumflex over (b)}^[i])},

wherein {circumflex over (b)}^[i] is a column vector containing the input symbol estimates after an i^th iteration of interference cancellation, q is a vector of the despread received signal, μ

^[i] is a scalar stabilizing step, Γ

^[i] is a weighting matrix used for scaling the input symbol estimates, R is a received-signal correlation matrix, F is an implementation matrix, D^[i] is a de-biasing matrix, and Ψ

returns a mixed (hard- or soft-) decision for each element of its input vector to produce the plurality of updated symbol estimates {circumflex over (b)}^[i+1].

44. The system recited in claim 31, configured as an internal feedback loop and an external feedback loop;

the external feedback loop configured to multiply a vector of the input symbol estimates by a weighting matrix and a users'"'"' received correlation matrix to produce an external-feedback vector, and subtracting the external-feedback vector from the despread signal vector for producing a difference error vector that is scaled by a scalar stabilizing step size; and

the internal feedback loop configured to multiply a vector of the plurality of weighted symbol estimates by an implementation matrix to produce an internal-feedback vector, the internal-feedback vector being summed with a vector derived from the scaled error signal.

45. The system recited in claim 44, wherein

the external feedback loop is configured to multiply the input symbol estimate, {circumflex over (b)}^[i], by both the weighting matrix Γ

^[i] and the users'"'"' received correlation matrix R to produce the external-feedback vector, RΓ

^[i]{circumflex over (b)}^[i], which is subtracted from the despread signal vector q in order to produce the difference error vector, q−

RΓ

^[i]{circumflex over (b)}^[i]; and

the internal feedback loop is configured to multiply the weighted symbol estimates, Γ

^[i]{circumflex over (b)}^[i], by the implementation matrix F to produce the internal-feedback vector, FΓ

^[i]{circumflex over (b)}^[i], which is summed with the difference error vector produced by the external feedback loop, q−

RΓ

^[i]{circumflex over (b)}^[i], that has been scaled by the stabilizing step size μ

^[i], to produce FΓ

^[i]{circumflex over (b)}^[i]+μ

^[i](q−

RΓ

^[i]{circumflex over (b)}^[i]).