a new "gentle_" parameter added to RED.

[Sally Floyd <floyd@aciri.org>]

I have added a new parameter to RED, "gentle_", that implements
the following modification to RED:

 Set to true to increase the packet drop rate slowly from max_p to
 1 as the average queue size ranges from maxthresh to twice maxthresh.
 The default is set to false, and max_p increases abruptly from
 max_p to 1 when the average queue size exceeds maxthresh.

This should appear in tonight's snapshot of the NS distribution.

- Sally
--------------------------------
http://www.aciri.org/floyd/
--------------------------------

The following file tcl/test/test-suite-red.txt describes the new
validation tests for this:

test-suite-red.txt:

Four new tests illustrate and validate the new "gentle_"
parameter added to RED.  These tests can be run with the
following commands:

../../ns test-suite-red.tcl ungentle
../../ns test-suite-red.tcl gentle
../../ns test-suite-red.tcl ungentleBadParams
../../ns test-suite-red.tcl gentleBadParams

The "gentle_" parameter implements a suggestion from
an October 1997 email on "Optimum functions for computing the drop
probability" available from "http://www.aciri.org/floyd/REDfunc.txt".
>From the ns man page on "gentle_":

 Set to true to increase the packet drop rate slowly from max_p to
 1 as the average queue size ranges from maxthresh to twice maxthresh.
 The default is set to false, and max_p increases abruptly from
 max_p to 1 when the average queue size exceeds maxthresh.

The first two tests compare the behavior of RED with "gentle_"
set to false, and to true, with the default RED parameters.
There is very little difference between the two tests.
The next two tests compare the behavior of RED with "gentle_"
set to false, and to true, with a less-optimal setting of
the RED parameters.  (E.g., with "max_p" set to 0.02 instead of
to 0.1.)  These two tests show that with "gentle_" set to true,
RED gives more robust performance with less-optimal settings of
the RED parameters.


I will try to add other proposed RED modifications to the
NS distribution as my spare time permits...