초록 ▼

A method and apparatus are disclosed for reducing the computational complexity of the RSSE technique. The apparatus and associated method does not assume that the signal energy of a pulse that has gone through a channel is always concentrated primarily in the initial taps, as is true for a minimum p

A method and apparatus are disclosed for reducing the computational complexity of the RSSE technique. The apparatus and associated method does not assume that the signal energy of a pulse that has gone through a channel is always concentrated primarily in the initial taps, as is true for a minimum phase channel. The present invention, however, recognizes that the signal energy is often concentrated in just a few channel coefficients, with the remaining channel coefficients being close to zero. A receiver apparatus and associated method is disclosed for reducing the number of channel coefficients to be processed with a high complexity cancellation algorithm from L to V+K which contain the majority of the signal energy, while processing the L−(K+V) non-selected coefficients with a lower complexity algorithm. By only processing the intersymbol interference caused by a reduced number of channel coefficients (i.e., L−(K+V)) using the tap-selectable TS-RSSE technique, while processing the intersymbol interference caused by the remaining channel coefficients with the tap-selectable decision feedback prefilter TS-DFP technique, a good bit error rate (BER) versus signal-to-noise ratio (SNR) performance is insured for a well-chosen value of V, where V represents the number of channel coefficients processed with the TS-RSSE technique (i.e., high complexity algorithm). No presumption is made apriori concerning which V taps will be processed by the TS-RSSE algorithm, but rather, an a posteriori determination is made in response to a changing channel impulse response.

대표청구항 ▼

1. A method for processing a received signal from a dispersive channel, said channel having a memory length L and being modeled as a filter having (L+1) taps, indexed zero through L, where each said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose a

1. A method for processing a received signal from a dispersive channel, said channel having a memory length L and being modeled as a filter having (L+1) taps, indexed zero through L, where each said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 through L are referred to as (L−K) tap-selectable coefficients, said method comprising the steps of: estimating the (L+1) channel coefficients associated with the (L+1) taps by analyzing said received signal; selecting V channel coefficients from the (L−K) tap-selectable coefficients which satisfy a predetermined criteria, where V is a number of the channel coefficients selected; processing the K non tap-selectable channel coefficients and the selected V channel coefficients with a high complexity cancellation algorithm; and processing L−(K+V) non-selected tap-selectable coefficients with a lower complexity cancellation algorithm. 2. The method according to claim 1, wherein the number V is an independent value in the range 0 to (L−K), and K is an independent value in the range 0 to L.3. The method according to claim 1, wherein the predetermined criteria is determined by selecting channel coefficients having a highest squared value.4. The method according to claim 1, wherein the predetermined criteria is determined by selecting channel coefficients having a highest absolute value.5. The method according to claim 1, wherein the predetermined criteria is determined by selecting channel coefficients whose squared value is above a predetermined coefficient threshold.6. The method according to claim 1, wherein the predetermined criteria is determined by selecting channel coefficients whose absolute value is above a predetermined coefficient threshold.7. The method according to claim 1, wherein the predetermined criteria is determined in accordance with the following steps: establishing a unique predetermined coefficient threshold for each of said (L−K) tap-selectable coefficients; comparing one of an absolute value and a squared signal value for each of said (L−K) tap-selectable channel coefficients with an associated predetermined coefficient threshold; and selecting a channel coefficient when said one of the absolute value and the squared signal value is above said associated predetermined coefficient threshold. 8. The method according to claim 1, wherein the predetermined criteria is determined in accordance with the following steps: establishing a summation threshold; sorting the (L−K) tap selectable channel coefficients in one of a decreasing absolute value order and a squared value order; initializing a variable value of a summation counter to zero; and adding one of the decreasing absolute value and the squared value of said sorted (L−K) tap selectable channel coefficients until it is determined that the summation counter variable value is equal to or greater than a summation threshold. 9. The method according to claim 1, wherein the estimating and selection steps are performed periodically.10. The method according to claim 1, wherein the estimating and selection steps are performed upon receiving a packet transmission.11. The method according to claim 1, where said lower complexity cancellation algorithm is a tap-selectable decision-feedback prefilter (TS-DFP).12. The method according to claim 11, wherein said TS-DFP reduces the intersymbol interface for said L−(K+V) non-selected tap-selectable coefficients, by calculating ynas an input into said high complexity cancellation algorithm, where ynat time n is calculated as:where {overscore (s)} iis the negation of siand {haeck over (a)}nis a tentative decision, and siis a tap control signal; f iare the channel coeff icients, whose index i is in the range K+1 to L; and znis the received signal.13. The method according to claim 12 wherein the tap control signal, si, is 1 for selected tap-selectable coefficients and 0 for non-selected tap-selectable coefficients.14. The method according to claim 12, where {haeck over (a)}nis obtained by slicinginto a hard value. 15. The method according to claim 12, where {haeck over (a)}nis obtained by slicinginto a soft value. 16. The method according to claim 1, wherein said high complexity cancellation algorithm is performed by a tap-selectable reduced state sequence estimator (TS-RSSE).17. The method according to claim 16, wherein said TS-RSSE reduces the intersymbol interference for the K non tap-selectable channel coefficients, by estimating the intersymbol interference in accordance with the following equation:where f iare channel coefficients, whose index i is in the range 1 to K, and â n−i(τn) is the survivor symbol corresponding to the data symbol an−ifrom the survivor path into state τnat time n.18. The method according to claim 16, wherein said TS-RSSE reduces the intersymbol interference for the V selected channel coefficients, by estimating the intersymbol interference in accordance with the following equation:where f ijare said V selected channel coefficients of index i and whose index j is in the range 1 to V; and â n−i, (τn) is the survivor symbol corresponding to the data symbol an−i, from the survivor path into state τn.19. The method according to claim 16, wherein said TS-RSSE reduces the intersymbol interface for the V selected channel coefficients, by estimating the intersymbol interference in accordance with the following equation:where f iare the channel coefficients, whose index i is in the range K+1 to L; s iare tap control signals for selecting the V channel coefficients; and â n−i(τn) is the survivor symbol corresponding to the data symbol an−i, from the survivor path into state τnat time n.20. The method according to claim 11, wherein said TS-DFP reduces the intersymbol interference for the L−(K+V) non-selected tap-selectable coefficients, by calculating yn, in accordance with the following equation:where {overscore (s)} iis the negation of si; f iare the channel coefficients whose index i is in the range K+1 to L; {haeck over (a)} n−iis the survivor symbol from the survivor path leading into the state with the best path metric, i.e. {haeck over (a)}n−i=ân−i(τ′n) and τ′nis the reduced state associated with a TS-RSSE technique having the best path metric at time n; and y nis an input to the high complexity cancellation algorithm.21. A receiver that receives a signal from a dispersive channel, said channel having a memory length, L, and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those (L+1) channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L+1) channel coefficients whose associated index is in the range K+1 through L are referred to as (L−K) tap-selectable coefficients, said receiver comprising: a channel estimator for determining the (L+1) channel coefficients associated with the (L+1) taps by analyzing a received signal; a tap selector for selecting V of the (L−K) tap-selectable coefficients satisfying a predetermined criteria, where V is a number of coefficients selected; a tap-selectable decision feedback prefilter (TS-DFP) circuit for: receiving (L−K) tap control signals from said tap selector;receiving said (L−K) tap-selectable coefficients from a channel estimator;receiving (L−K) symbols from the TS-RSSE which correspond to said (L−K) tap-selectable coefficients; andprocessing the intersymbol interference associated w ith L−(K+V) channel coefficients from among said (L−K) tap-selectable coefficients; a tap selectable reduced state sequence estimator (TS-RSSE) circuit for: receiving said (L−K) tap control signals and L channel coefficients from among said (L+1) channel coefficients, indexed one through L;processing the intersymbol interference associated with the V selected channel coefficients and K non tap-selectable channel coefficients; andselecting the best survivor path for each state of a trellis by performing an add-compare-select operation.22. A receiver that receives a signal from a dispersive channel, said channel having a memory length, L, and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 through L are referred to as (L−K) tap-selectable coefficients, said receiver including: a channel estimator for determining the (L+1) channel coefficients by analyzing a received signal; a tap selector for selecting V channel coefficients of the (L−K) tap-selectable coefficients satisfying a predetermined criteria, where V is a number of the channel coefficients selected; a tap-selectable decision feedback prefilter (TS-DFP) circuit for: receiving, from a channel estimator, (L−K) tap control signals and L channel coefficients, indexed 1 through L, from said (L+1) channel coefficients; andreducing the intersymbol interference associated with L−(K+V) coefficients from among said (L−K) tap-selectable coefficients which were not selected by the tap selector; a tap selectable reduced state sequence estimator (TS-RSSE) circuit for: receiving said (L−K) tap control signals and L channel coefficients from among said (L+1) channel coefficients, indexed one though L;processing the intersymbol interference associated with the V selected channel coefficients and K non tap-selectable channel coefficients and;selecting the best survivor path for each state of a trellis by performing an add-compare-select operation.23. The receiver of claim 22, wherein said TS-DFP reduces the intersymbol interference caused by the L−(K+V) non-selected channel coefficients, said TS-DFP comprising: a first selector having (L−K) symbol inputs, said first selector selecting L−(K+V) symbols inputs of said (L−K) symbol inputs; a second selector having (L−K) channel coefficient inputs, said second selector selecting L−(K+V) channel coefficients of said (L−K) tap-selectable coefficients, wherein the L−(K+V) selected channel coefficients are multiplied by corresponding the L−(K+V) selected symbols to yield L−(K+V) multiplied results, wherein said multiplied results are summed in a summer and added to the received signal to be provided as input to the TS-RSSE; a third selector having (L−K) symbol inputs, said third selector selecting V symbol inputs of said L−K symbol inputs; a fourth selector having (L−K) channel coefficient inputs, said fourth selector selecting the V channel coefficients of the (L−K) tap-selectable coefficients, wherein said V selected channel coefficients are multiplied by corresponding outputs of said third selector to yield V multiplied results, wherein said multiplied results are summed and added to said input to the TS-RSSE to form an internal signal. 24. The receiver of claim 23, wherein the TS-DFP further comprises a data slicer for obtaining the (L−K) symbol inputs.25. The receiver of claim 22, wherein said TS-DFP cancels the intersymbol interference caused by the L−(K+V) non-selected channel coefficients, said receiver comprising: a first selector having (L−K) symbol inputs, said first selector selecting L−(K+V) symbol inputs of the(L−K) symbol inputs; a second selector having (L−K) inputs for the (L−K) tap-selectable coefficients, said second selector selecting L−(K+V) channel coefficients of said (L−K) tap-selectable coefficients, wherein the L−(K+V) selected channel coefficients are multiplied by corresponding ones of the L−(K+V) selected symbol inputs, wherein said multiplied results are summed in a summer and added to the received signal provided as an input to the TS-RSSE. 26. The receiver of claim 25, further comprising a data slicer for obtaining the (L−K) symbol inputs.27. The receiver of claim 22, wherein said TS-RSSE includes a plurality of U decision feedback cells (U-DFCs) and a plurality of V decision feedback cells (V-DFCs), each of said plurality of U-DFC cells and V-DFC cells are associated with a respective state of a trellis which is processed by the TS-RSSE, each of said plurality of U-DFC cells cancels the intersymbol interference due to the non-selectable K channel coefficients for its respective state, each of said V-DFC cells cancels the intersymbol interference due to the V selected channel coefficients for its respective state.28. The U-DFCs of claim 27, wherein in each of said plurality of U-DFC cells the non-selectable K channel coefficients are multiplied by corresponding survivor symbols for the respective state.29. The U-DFCs of claim 28, wherein the survivor symbols correspond to one state of the trellis.30. The V-DFCs of claim 27, wherein each of said plurality of V-DFC cells compromise: a first selector for selecting V channel coefficients from among (L−K) channel coefficients; a second selector for selecting V survivor symbols from among (L−K) survivor symbols, wherein the V survivor symbols correspond to said V channel coefficients; and a summer for summing the multiplied result of said V selected channel coefficients with corresponding ones of the V selected survivor symbols. 31. The V-DFCs of claim 30, wherein the survivor symbols correspond to one state of the trellis.32. The V-DFCs of claim 27, wherein in each of said plurality of V-DFC cells, (L−K) multiplexers select V channel coefficients for multiplication by corresponding survivor symbols, each one of said plurality of V-DFC cells corresponding to one state from among a number R of states of the trellis.33. The receiver of claim 22, wherein said TS-DFP further includes (L−K) multiplexers for selecting L−(K+V) channel coefficients from among said (L−K) tap-selectable coefficients, the selected L−(K+V) channel coefficients are multiplied by corresponding ones of a group of (L−K) symbols to form L−(K+V) partial products and summing said L−(K+V) partial products and adding said summed L−(K+V) partial products to the received signal to be provided as an input to the TS-RSSE.34. A method for reducing the power consumption in a receiver that receives a signal from a dispersive channel, said channel having a memory length, L, and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 through L are referred to as tap-selectable coefficients, said method comprising the steps of: measuring a channel quality metric of a received signal; comparing the measured channel quality metric with a predetermined threshold to determine whether said channel quality metric is above or below said predetermined threshold; decreasing the number of coefficients to be processed by a tap-selectable reduced state sequence estimator (TS-RSSE) when it is determined that sa id channel quality metric is above said predetermined threshold, indicating a high quality channel; increasing the number of coefficients to be processed by said tap-selectable reduced state sequence estimator (TS-RSSE) when it is determined that said channel quality metric is below said predetermined threshold, indicating a low quality channel; increasing the number of coefficients to be processed by a tap-selectable decision feedback prefilter (TS-DFP) when it is determined that said channel quality metric is above said predetermined threshold, indicating a high quality channel; decreasing the number of coefficients to be processed by a tap-selectable decision feedback prefilter (TS-DFP) when it is determined that said channel quality metric below said predetermined threshold, indicating a low quality channel. 35. The method according to claim 34, wherein the measurement and comparison step are performed periodically.36. The method according to claim 34, wherein the channel quality metric is a signal-to-noise ratio (SNR).37. A method for reducing the power consumption in a receiver that receives a signal from a dispersive channel, said channel having a memory length, L, and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 though L are referred to as tap-selectable coefficients, said method comprising the steps of: measuring a channel quality metric of a received signal; comparing the measured channel quality metric with a predetermined threshold to determine whether said channel quality metric is above or below said predetermined threshold; decreasing the number of coefficients to be processed by a tap-selectable reduced state sequence estimator (TS-RSSE) when it is determined that said channel quality metric is below said predetermined threshold, indicating a high quality channel; increasing the number of coefficients to be processed by said tap-selectable reduced state sequence estimator (TS-RSSE) when it is determined that said channel quality metric is above said predetermined threshold, indicating a low quality channel; increasing the number of coefficients to be processed by a tap-selectable decision feedback prefilter (TS-DFP) when it is determined that said channel quality metric is below said predetermined threshold, indicating a high quality channel; decreasing the number of coefficients to be processed by a tap-selectable decision feedback prefilter(TS-DFP) when it is determined that said channel quality metric is above said predetermined threshold, indicating a low quality channel. 38. The method according to claim 37, wherein the channel quality metric is one of a bit error rate (BER), a mean squared error, a packet error rate, and a path metric of the best state.39. The method according to claim 37, wherein the measurement and comparison step are performed periodically.40. A method for reducing the power consumption in a receiver that receives a signal from a dispersive channel, said channel having a memory length, L, and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 through L are referred to as (L−K) tap-selectable coefficients, said method comprising the steps of: measuring a channel quality metric of a received signal; comparing the measured channel quality metric with a predetermined threshold to determine whether said channel quali ty metric is above or below said predetermined threshold; sorting the (L−K) tap-selectable coefficients by one of a decreasing absolute signal value and a squared value order; selecting either i) a number V of channel coefficients of the (L−K) tap-selectable coefficients, where the number V is less than a number of the (L−K) tap-selectable coefficients selected in a previous selection interval when it is determined that said channel quality metric is below said predetermined threshold, indicating a high quality channel; or ii) the number V of channel coefficients of the (L−K) tap-selectable coefficients, where the number V is greater than a number of the (L−K) tap-selectable coefficients selected in a previous selection interval when it is determined that said channel quality metric is above said predetermined threshold, indicating a low quality channel; processing intersymbol interference (ISI) associated with said V selected channel coefficients and K non-selectable channel coefficients in tap-selectable reduced state sequence estimator (TS-RSSE); and processing the ISI associated with the L−(K+V) non-selected channel coefficients in tap-selectable decision feedback prefilter (TS-DFP). 41. The method according to claim 40, wherein the measurement and comparison step are performed periodically.42. The method according to claim 40, wherein the measurement and comparison step are performed at one of a reception of a packet and subsequent to the reception of a packet.43. The method according to claim 40, where the quality metric is one of a bit error rate (BER), a mean squared error, a packet error rate, and a path metric of the best state.44. A method for reducing the power consumption in a receiver that receives a signal from a dispersive channel, said channel having a memory length, L, and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 through L are referred to as (L−K) tap-selectable coefficients, said method comprising the steps of: measuring a channel quality metric of a received signal; determining whether the measured channel quality metric is above or below a predetermined metric threshold; raising a coefficient threshold when it is determined that the measured channel quality metric is below said predetermined metric threshold, indicating a high quality channel; lowering said coefficient threshold when it is determined that the measured channel quality metric is above said predetermined metric threshold; comparing one of a squared signal value and an absolute signal value associated with (L−K) channel coefficients with said coefficient threshold to determine whether the (L−K) tap-selectable coefficients are equal to or greater than said coefficient threshold; processing those channel coefficients which are determined to be equal to or greater than said coefficient threshold with a higher complexity cancellation algorithm; and processing those channel coefficients which are determined to be less than said coefficient threshold with a lower complexity cancellation algorithm. 45. A method according to claim 44, wherein the measurement and comparison step are performed periodically.46. The method according to claim 44, wherein the measurement and comparison step are performed at one of a reception of a packet and subsequent to the reception of a packet.47. The method according to claim 44, wherein the channel quality metric is one of a bit error rate (BER), a mean squared error, and a path metric of the best state.48. A method for processing a signal received from a dispersive channel, said channel having a memory length L and being modeled as a filter having (L+1) taps, indexed zero through L, where each of said (L+1) taps has associated therewith a channel coefficient, wherein those channel coefficients whose associated index is in the range one through K are referred to as non tap-selectable coefficients and those (L−K) channel coefficients whose associated index is in the range K+1 through L are referred to as (L−K) tap-selectable coefficients, said method comprising: means for estimating the (L+1) channel coefficients associated with the (L+1) taps by analyzing said received signal; means for selecting V channel coefficients from the (L−K) tap-selectable coefficients which satisfy a predetermined criteria, where V is a number of the channel coefficients selected; first processing means for processing K non tap-selectable channel coefficients and the selected V channel coefficients with a higher complexity cancellation algorithm; and second processing means for processing the L−(K+V) non-selected tap-selectable channel coefficients with a lower complexity cancellation algorithm.