Browse Prior Art Database

Tree-Pruned Sequential Method for Generating Signals

IP.com Disclosure Number: IPCOM000105660D
Original Publication Date: 1993-Aug-01
Included in the Prior Art Database: 2005-Mar-20
Document File: 2 page(s) / 52K

Publishing Venue

IBM

Related People

Feig, E: AUTHOR [+3]

Abstract

Disclosed is a method for generating a constrained class of signals used in magnetic recording of data using a sequential dynamic process together with tree pruning techniques.

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

Tree-Pruned Sequential Method for Generating Signals

      Disclosed is a method for generating a constrained class of
signals used in magnetic recording of data using a sequential dynamic
process together with tree pruning techniques.

     The invention generates signals whose outputs by the read head
closely approximate signals of the form

      s(t) %% = %% sum from <k=1> to K % a sub k % f ( t - k T )
where a sub k % = % pm 1, f(t) is some prescribed fixed pulse shape,
T is some fixed time interval, and K is a fixed integer.
Furthermore, the input to the write head generating these signals is
of the form

     w(t) %% = %% sum from <k=1> to NK % a sub k % f ( t - k T/N )
where N is an integer, typically 6 to 10.  Clearly there are 2 sup
<NK> possible input signals.  These may all be generated , written
and read, their corresponding read signal denoted by r(t), then
compared to the desired output signal s(t) relative to an appropriate
norm, and the one closest chosen.  But this is too burdensome a
procedure.

     Two norms will be used here for comparison.  Let M be an integer
such that the memory of the media is at most MT; that is, a signal
after a pulse decays completely after time period MT.  The first is
the least squares norm given by

                  integral from <T(1-M)> to <T(K+M)> %
                  vbar % s(t) - r(t) % vbar sup 2 % d t
and the second is the maximum correlation norm given by

         < integral from <T(1-M)> to <T(K+M)> % s(t) % r(t) % d t >
           ...