Browse Prior Art Database

Hybrid Analog/ Digital Computation of Digital Data with Prescribed Distortion Tolerance for Transmission on Constrained Channel

IP.com Disclosure Number: IPCOM000108835D
Original Publication Date: 1992-Jun-01
Included in the Prior Art Database: 2005-Mar-23
Document File: 3 page(s) / 130K

Publishing Venue

IBM

Related People

Carter, JL: AUTHOR [+2]

Abstract

This invention is concerned with the transmission of data at very high rates over a channel that has a priori known signal-to-noise characteristics, as a function of frequency. Prescribed tolerances of distortion of the data are also known a priori .

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Hybrid Analog/ Digital Computation of Digital Data with Prescribed Distortion Tolerance for Transmission on Constrained Channel

       This invention is concerned with the transmission of data
at very high rates over a channel  that has a priori known
signal-to-noise characteristics, as a function of frequency.
Prescribed tolerances of distortion of the data are also known a
priori .

      A motivating example is provided by HDTV (High Definition TV),
where the transmission of digital data is over a channel constrained
by legal requirements.  The most common constraint is that the
channel reside (in frequency) between two presently existing channels
whose frequency supports are separated by about 6 MHz.  The signals
on the two surrounding channels leak onto this separation band.  Thus
transmission is limited to what we have called a constrained channel.
Furthermore, the eye tolerates various distortions of visual images
without noticing any degradation.  The eye's tolerance is a function
of the spatial frequencies of the image, and also of the temporal
frequency when a sequence of images forms a continuous video.

      The method disclosed here is to use analog circuits for certain
portions of the computations required for processing signals for
transmission over a constrained channel.  This is accomplished
without introducing more distortion than can be tolerated.  There are
other portions of the computation that are best performed with
digital circuitry.

      A preferred embodiment uses analog and digital circuits in
conjunction with Fourier transform coded modulation (FTCM) techniques
for video.  FTCM, in turn, involves previous work (1-3).
Conceptually, the computation comprises three phases.  In the first
phase, the image data (actual image data for initial image, and
difference image for subsequent images until reinitialization) is
passed through a 2-dimensional filter bank of K filters, yielding K
strings of numbers.

      In the second phase, each of the K strings is encoded by
methods, that achieve the desired level of reliability.  (In general,
strings corresponding to lower frequencies in the image are encoded
with greater reliability.)  The strings may be further manipulated by
digital techniques, including but not limited to the following:
.   If a string is all 0, then it need not be encoded at all.
Instead, extra digital information (encoded according to some
protocol) identifies which strings are zero.
.   Several strings may be combined in a single subchannel; for
example, a constellation can be designed so that the x-coordinate
labels one string and the y-coordinate another.
.   Two strings can be encoded along a single axis by mapping the
pair of integers (a,b) to the real number ax + b, where ax < b
for all expected values of a and b.

      In the third phas...