Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Class 1 and Class 2 Maximum Likelihood Detection for PPM Channels

IP.com Disclosure Number: IPCOM000123328D
Original Publication Date: 1998-Sep-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 3 page(s) / 71K

Publishing Venue

IBM

Related People

Hutchins, RA: AUTHOR [+2]

Abstract

A method for detecting optical data written using pulse-position modulation (PPM) is disclosed. The method uses a class 1 or class 2 target channel with maximum likelihood detection to detect the data. Modulation code constraints and multiple difference metrics are used to simplify the detector structure.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 53% of the total text.

Class 1 and Class 2 Maximum Likelihood Detection for PPM Channels

   A method for detecting optical data written using
pulse-position modulation (PPM) is disclosed.  The method uses a
class 1 or class 2 target channel with maximum likelihood detection
to detect the data.  Modulation code constraints and multiple
difference metrics are used to simplify the detector structure.

   In detecting pulse-position modulation (PPM) data from an
optical disk, a maximum-likelihood detector with a class 1 or class 2
target channel lbracket 1 rbracket  can be used (referred to as PR1ML
for a class 1 target channel with maximum likelihood detection and
PR2ML for a class 2 target channel with maximum likelihood
detection).  PR1ML and PR2ML detectors reduce the bit error rate of a
PPM channel by allowing more intersymbol interference (ISI) and
requiring less bandwidth than conventional peak detectors.

   A block diagram of the PPM channel for PR1ML or PR2ML is
shown in the figure.  The input to the channel is a pulse of
duration T when a '1' is is to be written on the disk and the
absence of a pulse when a '0' is to be written.  The output of the
equalizer is a class 1 or class 2 target response, depending upon the
equalization method used in the read-detect channel.  The equalizer
output is fed into a maximum likelihood detector for data recovery.

   For a PR1ML channel, y sub k = a sub k + a sub <k - 1>
where  y sub k  equals the digitized output of the equalizer at
sample k, a sub k  equals the input at sample k, and  a sub <k - 1>
equals the input at sample k - 1.  The following equation for the
mean-squared error must be minimized:
   sum from k of lbrace y sub k - ( a sub k + a sub <k - 1> )
   rbrace sub 2

   After some algebraic manipulation, the problem can be described
in terms of a survivor metric.
   J sub k (a sub k) = max lbrace J sub <k - 1> ( a sub <k - 1> )
   + ( a sub k + a sub <k - 1> ) ( 2 y sub k - a sub k - a sub
   < k - 1 > ) rbrace
   where a sub <k - 1> memberof 0,1 and  a sub k memberof 0,1.
   A difference metric  DJ sub k  can be defined such that:
   DJ sub k = 1 over 2 lbrace J sub k ( 1 ) - J sub k ( 0 ) rbrace
   If the d-constraint of the modulation code is at least 2, the
   algorithm can be simplified to:
   'If ' DJ sub <k - 1> lt - y sub k + 1 over 2 , ' then '
...