iitis-generator-output
- iitis-generator file-system output
2011.07/
*-- session-name/
*--- 2011.07.26-15:40:25/
*-- 1/
| *-- internal-stats.txt
| *-- linestats.txt
| *-- mon0/
| | *-- interface.txt
| | *-- link-2->1.txt
| *-- mon1/
| | *-- interface.txt
*-- 2/
| *-- internal-stats.txt
| *-- linestats.txt
| *-- mon0/
| | *-- dump.pcap
| | *-- interface.txt
| | *-- link-1->2.txt
*-- generator.conf
*-- traffic.txt
iitis-generator(1) is a tool for distributed, statistical evaluation of wireless networks. Each
network node that takes part in an evaluation experiment runs the iitis-generator
program. This
manual page documents the structure of its file-system output and format of constituent files.
iitis-generator
has a concept of root directory for the program output. This property can be set
using either command-line or configuration file option named "root".
On each run of iitis-generator
, after the synchronization phase, a five-level directory structure
is initialized:
origin-Y.M/ (1)
*-- session-name/ (2)
*--- origin-Y.M.D-H:M:S/ (3)
*-- node-id/ (4)
*-- interface ... (5)
With the following meaning:
origin-Y.M
is the year and month of the origin - e.g. "2007.07"session-name
is the session nameorigin-Y.M.D-H:M:S
is the full date and time of the origin - e.g. "2011.07.26-15:40:25"node-id
is the ID number of the current nodeinterface
is the name of network interface, e.g. "mon0"See iitis-generator(1) for the concepts of session name and origin.
In case session name is not set, level (2) is skipped. Several directories on level (5) are created
- each test network interface has its own directory. Also, each node will create its own directory
on level (4). Running iitis-generator
with the same session name many times will result in many
directories on level (3).
Both the traffic and the config files are copied as-is to level (3).
On level (4), following files are created:
internal-stats.txt
: statistics of the iitis-generator
internalslinestats.txt
: aggregated statistics of all lines from the traffic fileOn level (5), following files may be created:
interface.txt
: statistics of network interfacelink-X->Y.txt
: statistics for traffic coming from node X to Ydump.pcap
: PCAP file with dumped interface trafficStatistics are stored in text files. Data is organized in a tabular form. First line holds column names and always starts with "#time ...". Below is a short example:
#time rcv_ok rcv_ok_bytes rcv_dup rcv_dup_bytes rcv_lost rssi rate antnum
1 1 1500 0 0 0 -7 4.90909 0
2 2 3000 0 0 0 -17.3388 12.2119 0
3 1 1500 0 0 0 -21.1863 14.9006 0
4 1 1500 0 0 0 -24.3343 17.1005 0
5 1 1500 0 0 0 -26.9099 18.9004 0
First column gives time since the origin. Besides, there are two kinds of columns:
counter
: an integer, counts occurances, bytes, etc.; it is set back to 0 after writing its
value to a statistics file, so actually the values found in statistics are their first derivatives
in timegauge
: a real number, simply gives the current value; in case column value is an aggregate
constructed off several other gauges, an EWMA value is giveniitis-generator
was written by Pawel Foremski pjf@iitis.pl. Copyright (C) 2011 IITiS PAN Gliwice
http://www.iitis.pl/.
iitis-generator(1), iitis-generator-conf(5), iitis-generator-traffic(5), IITiS WiFi lab