Marvellworld Trade Ltd.

Marvell International Limited

Pak Hei Matthew Leung

Jacky S Chow

Eugene Yuk-Yin Tang

King-Fuk Luk

Raphael Cendrillon

Nicholas Philip Sands

Peter Tze-Hwa Liu

Francis Man-Lung Ng

Slobodan Nedic

Hungming Chang

Tin-Yau Fung

Nicholas Phllip Sands

A method for optimizing resource allocation for signal conditioning of DSL or other signal lines includes developing a signal quality metric using crosstalk interference and, optionally, signal to noise level. Use of existing pilot tone training selectively assigned to signal lines allows determination of the signal quality metric. A threshold level is selected and all signal lines having a signal quality metric above the threshold level are assigned crosstalk cancellation resources. Because the total number of signal lines requiring crosstalk cancellation is not known at the beginning, a particular threshold level may result in allocation of more than the available resources or allocation of significantly less than the available resources. When this happens the threshold level may be adjusted and the process repeated until crosstalk cancellation resources are assigned up to a usable limit.

Crosstalk canceller initialization

Improved techniques for acquiring symbol boundary timing at a receiver of a multicarrier data transmission system during a training sequence are disclosed. One aspect is symbol boundary determination at a receiver wherein minimal interference is used as a criterion in selecting from a plurality of potential symbol boundary timings. The symbol boundary determination at the receiver can be performed in a time domain or a frequency domain manner. Another aspect pertains to an improved training sequence wherein pairs of identical symbols are transmitted by a transmitter. These symbols can be supplied to the transmitter in a time domain or a frequency domain manner.

Method and system for determining symbol boundary timing in a multicarrier data transmission system

A transform calculator includes a plurality of memories. A memory mapping rules module is configured to apportion points of a discrete time domain sequence among the plurality of memories. A pipelined data path is configured to iteratively process pairs of the points of the discrete time domain sequence received from the plurality of memories. A control module is configured to select the pairs of the points in the plurality of memories for processing by the pipelined data path, wherein only one point of each of the pairs of the points is selected at a given time.

Processing system and method for transform

A system for processing subscriber line data signals uses a coprocessor to receive signals in a pre-determined order from a plurality of line interface cards. The coprocessor uses the pre-determined order to sort the data signals by tone and process all received signals on the same tone at the same time. Because all signals on a tone, or those on a tone with known crosstalk issues, are processed together, the signal improvement is better than prior art methods. When not used, the coprocessor may be replaced with a loop-back circuit to reduce cost when crosstalk is not a significant issue.

Per tone coprocessor architecture

FFT butterfly data sets may be stored in memory in a predetermined order. Such an order may allow a butterfly data set to be read from a single memory address location. The memory addressed may be computed by an address rotary function depending on the butterfly and stage of the FFT. Addressing the memory in such a manner may allow each butterfly data set of a subsequent FFT stage to be stored to a single memory location. Shuffle registers may delay the writing of FFT butterfly results to the memory until most of the data corresponding to a particular butterfly operation has been computed. The shuffle registers may rearrange and combine the results of one or more butterfly operations in a different manner from which they have been computed. Combining the results in this manner may allow a subsequent FFT stage to access data by addressing a single memory location.

Methods and apparatus for providing an efficient FFT memory addressing and storage scheme

An improved training sequence for acquiring symbol boundary timing at a receiver of a data transmission system is disclosed. One aspect pertains to an improved training sequence wherein groups of identical symbols are transmitted by a transmitter. These symbols can be supplied to the transmitter in a time domain or a frequency domain manner. The improved training sequence facilitates symbol boundary determination at a receiver.

Training sequence for symbol boundary detection in a multicarrier data transmission system

Improved techniques for concurrent channel equalization and far-end crosstalk channel compensation (e.g., estimation and/or cancellation) in a multicarrier data transmission system are disclosed. The improved techniques can produce coefficients for an electronic fitter that provide channel equalization and for an electronic filter that provides cancellation of the far-end crosstalk. These coefficients can be initially trained and then subsequently updated during data transmission. Optionally, common coefficient determination resources can be utilized for both far-end crosstalk and near-end crosstalk purposes. These improved techniques are particularly suitable for use with a digital multicarrier communication system having multiple-input multiple-output systems.

Method and system for channel equalization and crosstalk estimation in a multicarrier data transmission system

An improved training sequence for estimating a channel (e.g., channel impulse response) in the time domain is disclosed. The improved training sequence enables time-domain estimation and eliminates the need for correct frequency-domain channel estimation at every frequency. By utilizing the training sequence according to the invention, estimation of each of a plurality of time-domain channel taps can be decoupled from each other. This enables a channel estimate to be performed with not only a higher convergence speed but also lower complexity.

Training sequence for channel estimation in a data transmission system

The invention includes an apparatus and the associated method to digitally process data communicated through a communication channel between a transceiver pair. A global control element and programmable algorithm control elements are used to implement an algorithm using a datapath. The control signal outputs of at least one programmable algorithm control element are coupled to the datapath. The datapath may use the control signals to drive transmission data down a computation path that implements the desired algorithm. The datapath may be duplicated to meet the requirements of a particular device and operate on a larger subset of data using the same control signals that are provided by the programmable algorithm control elements.

Programmable architecture for digital communication systems that support vector processing and the associated methodology

A system includes M memories, wherein a first mapping assigns each point of an N-point input sequence to one of the M memories. A pipelined data path receives an input from each of the M memories, stores an output to each of the M memories, and iteratively processes pairs of points of the N-point input sequence. A control module designates the pairs of points from the M memories for processing by the data path, wherein only one point of each of the pairs is designated at one time.

Improved techniques for concurrent channel equalization and far-end crosstalk channel compensation (e.g., estimation and/or cancellation) in a multicarrier data transmission system are disclosed. The improved techniques can produce coefficients for an electronic filter that provide channel equalization and for an electronic filter that provides cancellation of the far-end crosstalk. These coefficients can be initially trained and then subsequently updated during data transmission. Optionally, common coefficient determination resources can be utilized for both far-end crosstalk and near-end crosstalk purposes. These improved techniques are particularly suitable for use with a digital multicarrier communication system having multiple-input multiple-output systems.

Improved techniques for estimating a channel (e.g., channel impulse response) in the time domain are disclosed. The improved techniques can be used to estimate time-domain channel taps in a digital communication system. An improved training sequence enables time-domain estimation and eliminates the need for correct frequency-domain channel estimation at every frequency. By utilizing the improved training sequence, the estimation of each of the time-domain channel taps can be decoupled from each other. This enables a channel estimate to be performed with not only a higher convergence speed but also lower complexity.

Growing community of inventors

Hong Kong, China

Pak Hei Matthew Leung