Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
1998-05-27
2001-02-27
Pham, Chi H. (Department: 2731)
Pulse or digital communications
Equalizers
Automatic
Reexamination Certificate
active
06195386
ABSTRACT:
TECHNICAL FIELD OF THE INVENTION
This invention relates in general to the field of digital signal processing circuits and systems. More particularly, the present invention relates to a parallel decision feedback equalizer.
BACKGROUND OF THE INVENTION
With the imminent introduction of digital television and high definition television (HDTV), the signal processing requirements of digital video delivery systems have recently received much attention. The current trend in the industry is to digitize the analog signal as soon as possible at a low rate and process the signal digitally. Digital signal processing algorithms such as adaptive filtering and equalization are critical for such systems and have long been used in telephone modem applications. However, the bit rate requirement for compressed video data is an order of magnitude greater than that for telephone modems.
For a HDTV system, the symbol rate at the receiver is approximately 6 MHz. An adaptive equalizer, used to reduce the intersymbol interference of the incoming signal, requires approximately a 100-tap filter for the equalizer. Even when using the simplest Least Mean Square (LMS) algorithm for updating the filter coefficients at every sample, the computational requirements are beyond the capabilities of any programmable digital signal processor in the market today. The complexity of the LMS algorithm is known to be 3N where N is number of taps. Therefore, if the coefficients of the equalizer is to be updated at every sample, the required processing power would be approximately 3N*6.0×10
6
, i.e., 1,800 Million for N=100, instructions per second.
The conventional approach to resolve this problem is to assume that the channel characteristics are changing much slower than the symbol rate. This assumption permits filter coefficient updates at a much slower rate. The resulting system architecture consists of dedicated hardware for the filter section and a programmable digital signal processor (DSP) for error computation and new coefficient generation. Current application specific integrated circuit (ASIC) technology enables us to develop dedicated filter chips for the equalizer which can operate at high input sample rates. However, depending on the application, a conventional custom chip cannot provide an adequate level of performance when both a wide input data width, e.g. greater than 16 bits and high sample rates are required.
To overcome this limitation, various approaches have been proposed to parallelize the equalizer structure. Previously, parallelization of the decision feedback equalizer has been developed based on block updates using an extended LMS algorithm for filter update. In this scheme, the input data samples are broken into multiple blocks with the same number of samples each. They are then processed by multiple decision feedback equalizers in parallel. Also, another parallel scheme uses a double-row decision feedback equalizer algorithm. These schemes have the disadvantage of requiring more hardware and are not suited to applications where fast coefficient updates are required.
Accordingly, a need has arisen for a decision feedback equalizer which can accommodate a wide input data width and high sampling rates, and does not involve more complex hardware.
SUMMARY OF THE INVENTION
In accordance with the present invention, a parallel decision feedback equalizer is provided which eliminates or substantially reduces the disadvantages associated with prior systems.
In one aspect of the invention, a parallel decision feedback equalizer using residue number system (RNS) is provided. The binary input data is converted to the residue number system, in particular the quadratic residue number system for digital video systems, prior to filtering by a plurality of parallel adaptive filters. The filtered data is then converted back to the binary system by a converter prior to the decision making operation by a slicer. The output from the slicer is also provided as a feedback in the residue number system to be summed with the output from the parallel adaptive filters.
In another aspect of the invention, a method for performing decision feedback equalization is provided. The input binary data is first converted to the residue number system, and the converted data is then provided to the decision feedback equalizer which includes a number of adaptive filters in parallel. The filtered data is converted back to binary data before supplying it to a slicer. The output of the slicer is also fed back to be converted to the residue number system, adaptive filtered, and combined with the output of the adaptive filters in the forward direction.
The parallel decision feedback equalizer in accordance with the present invention results in less hardware and faster data processing even for large data widths and fast symbol rates.
REFERENCES:
patent: 5008668 (1991-04-01), Takayama et al.
patent: 5619535 (1997-04-01), Alvarez, Jr.
patent: 5640423 (1997-06-01), Archer
patent: 5872817 (1999-02-01), Wei
Oh et al. Implementation of a parallel DFE using residue number system, IEEE, pp. III-237 through III-240, 1994.
Brady III Wade James
Corrielus Jean B.
Pham Chi H.
Telecky , Jr. Frederick J.
Texas Instruments Incorporated
LandOfFree
Parallel decision feedback equalizer does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Parallel decision feedback equalizer, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Parallel decision feedback equalizer will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2595789