Systems and methods for mitigating the effects of upstream far-end cross talk
A method is described for reducing self-induced far end crosstalk (self-FEXT) in a multiple input multiple output (MIMO) digital subscriber line (xDSL) system. An initial value for an off-diagonal multiple input multiple output canceller (ODMC) is derived while the ODMC is inactive and while in data mode. The method includes activating the initial ODMC and converging towards an steady-state value for the ODMC by performing an adaptive algorithm to maximize the Shannon's capacity of the system and to reduce upstream self-FEXT. Bit loading is performed and a frequency domain equalizer (FEQ) is updated.
1. A method comprising:
deriving an initial value for an off-diagonal multiple input multiple output canceller (ODMC) while the ODMC is inactive;
activating the initial ODMC and converging towards a steady-state value for the ODMC by performing an adaptive algorithm to maximize a distributed Shannon's capacity of the system and to reduce upstream self-induced far end crosstalk (self-FEXT); and
performing bit loading and updating a frequency domain equalizer (FEQ).
2. The method of claim 1 , wherein reducing self-FEXT is performed for a discrete tone q and for a discrete multi-tone (DMT) time instant.
3. The method of claim 2 , wherein the adaptive algorithm is expressed by the following:
f pc,m [q,t nK+i ]=f pc,m [q,t nK+i−1 ]+μ[t nK+i ]( y bc,m [q,t nK+i ]−r −m T [q,t nK ]y bc,−m [q,t nK+i ]) e pc,m *[q,t nK+i /t nK ,t nK+i−1 ] for n≧0 and 1≦i≦K
and
r −m [q,t (n+1)K ]=r −m [q,t nK ]−μ[t (n+1)K ]f pc,m [q,t (n+1)K ]y bc,−m [q,t (n+1)K ]e* pc,m [q,t (n+1)K /t nK ,t (n+1)K ] for n≧0,
where f pc,m represents a post-cancellation frequency domain equalizer (FEQ), y bc,−m represents equalized received vectored data, and r −m represents the steady-state ODMC.
4. The method of claim 1 , wherein deriving an initial value for an off-diagonal multiple input multiple output canceller (ODMC) is performed while in data mode.
5. The method of claim 1 , wherein deriving an initial value for an off-diagonal multiple input multiple output canceller (ODMC) is performed while in warm-up mode.
6. The method of claim 1 , wherein deriving an initial value for the ODMC comprises determining a cross-correlation between error samples and received data comprising a known learning sequence.
7. The method of claim 1 , wherein deriving an initial value for the ODMC comprises:
setting the initial value to zero; and
performing an adaptive algorithm to converge towards an initial value for the ODMC.
8. The method of claim 1 , wherein deriving an initial value for the ODMC comprises determining a cross-correlation between error samples and detected transmit data based on received samples.
9. A method comprising:
performing a learning phase to derive both an initial off-diagonal multiple input multiple output canceller (ODMC) and a steady-state ODMC for a given tone frequency, wherein performing a learning phase comprises:
deriving an initial value for the ODMC while in data mode and while the ODMC is inactive; and
converging toward the steady-state ODMC while in data mode and the ODMC is active to maximize a distributed Shannon's capacity of the system, wherein converging comprises performing an adaptive algorithm to derive a steady state ODMC; and
passing upstream data through the steady-state ODMC to reduce self-induced far end crosstalk (self-FEXT).
10. The method of claim 9 , wherein performing a learning phase further comprises performing bit loading and updating a frequency domain equalizer (FEQ) when the SNR crosses a predetermined threshold for a predetermined duration.
11. The method of claim 9 , wherein performing a learning phase further comprises performing bit loading and updating a frequency domain equalizer (FEQ) when the steady state ODMC is reached.
12. The method of claim 9 , further comprising maintaining the steady-state ODMC by performing the adaptive algorithm in order to maintain the maximum distributed Shannon's capacity of the system.
13. The method of claim 9 , wherein deriving the initial value of the ODMC comprises determining a cross-correlation between error samples and received data comprising a known learning sequence.
14. The method of claim 9 , wherein deriving the initial value of the ODMC comprises:
setting the initial value to zero; and
performing an adaptive algorithm to converge towards a steady state initial value for the ODMC.
15. The method of claim 9 , wherein deriving the initial value of the ODMC comprises determining a cross-correlation between error samples and detected transmit data based on received samples.
16. A method comprising:
performing a learning phase to derive both an initial off-diagonal multiple input multiple output canceller (ODMC) and a steady-state ODMC for a given tone frequency, wherein performing a learning phase comprises:
deriving an initial value for the ODMC while in warm-up mode and while the ODMC is inactive; and
converging toward the steady-state ODMC while in data mode and the ODMC is active to maximize a distributed Shannon's capacity c the system, wherein converging comprises performing an adaptive algorithm to derive a steady state ODMC; and
passing upstream data through the steady-state ODMC to reduce self-induced far end crosstalk (self-FEXT).
17. The method of claim 16 , wherein performing a learning phase further comprises performing bit loading.
18. The method of claim 16 , wherein performing a learning phase further comprises performing bit loading and updating a frequency domain equalizer (FEQ) when the steady state ODMC is reached.
19. The method of claim 16 , further comprising maintaining the steady-state ODMC by performing the adaptive algorithm in order to maintain the maximum distributed Shannon's capacity of the system.
20. The method of claim 16 , wherein deriving the initial value of the ODMC comprises determining a cross-correlation between error samples and received data comprising a known learning sequence.
21. The method of claim 16 , wherein deriving the initial value of the ODMC comprises:
setting the initial value to zero; and
performing an adaptive algorithm to converge towards a steady state initial value for the ODMC.
22. The method of claim 16 , wherein deriving the initial value of the ODMC comprises determining a cross-correlation between error samples and detected transmit data based on received samples.
23. A system comprising:
initialization logic stored on a memory and executed by a processor for deriving an initial off-diagonal multiple input multiple output canceller (ODMC), wherein the logic for deriving an initial ODMC comprises logic for determining a cross-correlation between error samples and received data comprising a known learning sequence;
logic stored on the memory for deriving a steady-state ODMC from the initial ODMC to reduce self-induced far end crosstalk (self-FEXT) by utilizing an adaptive algorithm, wherein the ODMC is active; and
logic stored on the memory for performing bit loading and updating a frequency domain equalizer (FEQ) when a signal to noise ratio (SNR) of the system rises above a pre-determined value for a predetermined duration in order to maximize channel capacity.
24. A system comprising:
initialization logic stored on a memory and executed by a processor for deriving an initial off-diagonal multiple input multiple output canceller (ODMC),
logic stored on the memory for deriving a steady-state ODMC from the initial ODMC to reduce self-induced far end crosstalk (self-FEXT) by utilizing an adaptive algorithm, wherein the ODMC is active, Wherein the logic for deriving an initial ODMC comprises logic for determining a cross-correlation between error samples and detected transmit data based on received samples; and
logic stored on the memory for performing bit loading and updating a frequency domain equalizer (FEQ) when a signal to noise ratio (SNR) of the system rises above a pre-determined value for a predetermined duration in order to maximize channel capacity.
25. A system comprising:
initialization logic stored on a memory and executed by a processor for deriving an initial off-diagonal multiple input multiple output canceller (ODMC);
logic stored on the memory for deriving a steady-state ODMC from the initial ODMC to reduce self-induced far end crosstalk (self-FEXT) by utilizing an adaptive algorithm, wherein the ODMC is active;
logic stored on the memory for performing bit loading and updating a frequency domain equalizer (FEQ) when a signal to noise ratio (SNR) of the system rises above a pre-determined value for a predetermined duration in order to maximize channel capacity;
logic stored on the memory for maintaining the steady-state ODMC by performing the adaptive algorithm in order to maintain the maximum channel capacity.