wiki:WiMAX/30/05
close Warning: Can't synchronize with repository "(default)" (/common/SVN/wimax does not appear to be a Subversion repository.). Look in the Trac log for more information.

Version 14 (modified by Nilanjan Paul, 9 years ago) ( diff )

16. Wimax Measurement Experiment: oml2-iperf with variable link attenuation values

Description

This experiment runs oml2-iperf on two nodes and collects bandwidth measurements along with the controlled attenuator values. The RSSI seen by the wimax node interface is varied by controlling this attenuator value. At different attenuator values we can measure the down link throughput for the wimax interface.

This experiment script is geared towards nodes on sand box 4 which has nodes with wireless interfaces connected the base station via an rf attenuator matrix. Using this experiment we also measure the throughout at different modulation-coding schemes (MCS) which is set prior to running the experiment script.

Hardware utilized

  1. Sandbox 4 node1-4: runs oml2-iperf as server. The wimax interface card is used for measurements.
  2. Sandbox 4 node1-9: runs oml2-iperf as client.
  3. RF-attenuator matrix: this used by the script to attenuate the air interface between the wimax base station and node.

Software / scripts requirements worth mentioning

  1. baseline-iperf.ndz: disk image that's loaded onto the nodes. This image is loaded with oml2-iperf.
  2. oml2-server: server running on sandbox 4 which collects measurement data from oml2-iperf and stores in a database file.
  3. sqlite3: command line interface for accessing the contents in a database file.
  4. wimax-iperf-attn.rb: the experiment script
  5. R: a statistic analysis tool.

Set up

Getting started, ssh into the console of sandbox 4.

  • Verify the oml2-server is running: $ ps -elf | grep oml2-server
    nilanjan@console.sb4:~/exp/dl-iperf/5.2$ ps -elf | grep oml2-server
    4 S oml2       948     1  0  80   0 -  2471 poll_s Feb02 ?        00:00:16 /usr/bin/oml2-server -l 3003 --logfile=/var/log/oml2-server.log --user=oml2 --group=oml2
    0 S nilanjan  1259 27360  0  80   0 -  1002 pipe_w 10:59 pts/1    00:00:00 grep oml2-server
    

Before the experiment, the BS parameters can be changed. The following are a few examples commands to control the MCS:

For other details on changing BS parameters please refer to Basic Support Services under http://wimax.orbit-lab.org/wiki

Although the experiment script can vary the attenuator values, the details on manually changing attenuator values can be found in RF Matrix Control Services under http://wimax.orbit-lab.org/wiki

Usage

  • Start clean, turn off all the nodes: omf-5.2 tell offh all
    nilanjan@console.sb4:~/exp/dl-iperf/5.2$ omf-5.2 tell offh all
    
     INFO NodeHandler: init OMF Experiment Controller 5.2.408
     INFO Topology: Loading topology 'system:topo:all'.
    ---------------------------------------------------
     Testbed : sb4.orbit-lab.org - Command: offHard
     Node n_1_2 - Ok
     Node n_1_7 - Ok
     Node n_1_1 - Ok
     Node n_1_6 - Ok
     Node n_1_5 - Ok
     Node n_1_9 - Ok
     Node n_1_4 - Ok
     Node n_1_8 - Ok
     Node n_1_3 - Ok
    ---------------------------------------------------
    
  • Query the status of the nodes: omf-5.2 stat. Wait until all nodes are in powered off state.
    nilanjan@console.sb4:~/exp/dl-iperf/5.2$ omf-5.2 stat
     INFO NodeHandler: init OMF Experiment Controller 5.2.408
    -----------------------------------------------
     INFO Topology: Loading topology 'system:topo:all'.
     Testbed : sb4.orbit-lab.org
     Node n_1_1 - State: POWEROFF
     Node n_1_2 - State: POWEROFF
     Node n_1_3 - State: POWEROFF
     Node n_1_4 - State: POWEROFF
     Node n_1_5 - State: POWEROFF
     Node n_1_6 - State: POWEROFF
     Node n_1_7 - State: POWEROFF
     Node n_1_8 - State: POWEROFF
     Node n_1_9 - State: POWEROFF
    -----------------------------------------------
    
  • Load image onto node1-4 and node1-9: omf-5.2 load 1,4],[1,9 baseline-iperf.ndz
    nilanjan@console.sb4:~/exp/dl-iperf/5.2$ omf-5.2 load [[1,4],[1,9]] baseline-iperf.ndz ; sleep 360 ; omf-5.2 tell on [[1,4],[1,9]] ; sleep 360 ; omf-5.2 exec dl-iperf.rb
    Imaging nodes: '[[1,4],[1,9]]' with image 'baseline-iperf.ndz'
    (Domain:  default from hostname)
    (Timeout:  800 sec.)
     INFO NodeHandler: init OMF Experiment Controller 5.2.408
     INFO NodeHandler: init Experiment ID: sb4.orbit-lab.org_2012_02_09_10_22_14
     INFO NodeHandler: Web interface available at: http://10.14.0.10:4000
     INFO Experiment: load system:exp:stdlib
     INFO property.resetDelay: value = 100 (Fixnum)
     INFO property.resetTries: value = 1 (Fixnum)
     INFO Experiment: load system:exp:imageNode
     INFO property.nodes: value = [[1, 4], [1, 9]] (Array)
     INFO property.image: value = "baseline-iperf.ndz" (String)
     INFO property.domain: value = nil (NilClass)
     INFO property.outpath: value = "/tmp" (String)
     INFO property.timeout: value = 800 (Fixnum)
     INFO stdlib: Waiting for nodes (Up/Down/Total): 0/2/2 - (still down: n_1_9,n_1_4)
     INFO stdlib: Waiting for nodes (Up/Down/Total): 0/2/2 - (still down: n_1_9,n_1_4)
     INFO stdlib: Waiting for nodes (Up/Down/Total): 0/2/2 - (still down: n_1_9,n_1_4)
     INFO stdlib: Waiting for nodes (Up/Down/Total): 0/2/2 - (still down: n_1_9,n_1_4)
     INFO exp: Progress(0/0/2): 0/0/0 min(n_1_9)/avg/max (46) - Timeout: 750 sec.
     INFO whenAll: *: 'status[@value='UP']' fires
     INFO exp: Progress(0/0/2): 20/20/20 min(n_1_9)/avg/max (46) - Timeout: 740 sec.
     INFO exp: Progress(0/0/2): 40/40/40 min(n_1_9)/avg/max (46) - Timeout: 730 sec.
     INFO exp: Progress(0/0/2): 60/60/60 min(n_1_9)/avg/max (46) - Timeout: 720 sec.
     INFO exp: Progress(0/0/2): 70/75/80 min(n_1_9)/avg/max (46) - Timeout: 710 sec.
     INFO exp: Progress(0/0/2): 90/90/90 min(n_1_9)/avg/max (46) - Timeout: 700 sec.
     INFO exp: Progress(2/0/2): 100/100/100 min()/avg/max (46) - Timeout: 690 sec.
     INFO exp:  -----------------------------
     INFO exp:  Imaging Process Done
     INFO exp:  - 2 node(s) successfully imaged - See the topology file: '/tmp/sb4.orbit-lab.org_2012_02_09_10_22_14_topo_active.rb'
     INFO exp:  -----------------------------
     INFO Experiment: DONE!
     INFO NodeHandler: Shutting down experiment, please wait...
     INFO NodeHandler: Shutdown flag is set - Turning Off the resources
     INFO run: Experiment sb4.orbit-lab.org_2012_02_09_10_22_14 finished after 1:51
    
    
  • After loading the image, node1-4 & node1-9 are turned off. Turn these nodes back on: omf-5.2 tell on 1,4],[1,9
    nilanjan@console.sb4:~/exp/dl-iperf/5.2$ omf-5.2 tell on [[1,4],[1,9]]
    
     INFO NodeHandler: init OMF Experiment Controller 5.2.408
    ---------------------------------------------------
     Testbed : sb4.orbit-lab.org - Command: on
     Node n_1_9 - Ok
     Node n_1_4 - Ok
    ---------------------------------------------------
    
  • Query the status of the nodes until node1-4 & node1-9 are in POWERON state: omf-5.2 stat
  • Now run the experiment script: omf-5.2 exec wimax-iperf-attn.rb

Post processing data

After the experiment successfully executes a database file (values.db) is created. This file is populated with the following parameters for each iperf run:

  1. atten ——→ rf matrix attenuation value
  2. nBytes ——> number of bytes transferred from iperf client to server.
  3. time ———> total time taken for iperf connection.
  4. kBperS ——> throughput in kilobytes per second

To view the contents of the database file: sqlite3 -header -column values.db "select * from t1"

nilanjan@console.sb4:~/exp/dl-iperf/5.2$ sqlite3 -header -column values.db "select * from t1"
t1key       atten       nBytes      time              kBperS
----------  ----------  ----------  ----------------  ----------------
1           0           10878976    10.0952429920435  8419.01478418953
2           5           11141120    10.2252359986305  8512.2729697053
3           10          11010048    10.1255509853363  8494.94512689406
4           15          11010048    10.1354229897261  8486.67096451635
5           20          2097152     12.675085991621   1292.61450461408

In this simple example we can use R (statistical analysis tool) to analyze the data. This tool can be easily installed: apt-get install r-base
So first dump the contents out to a file: sqlite3 -header values.db "select * from t1" > data.tmp

nilanjan@console.sb4:~/exp/dl-iperf/5.2$ sqlite3 -header values.db "select * from t1" > data.tmp
nilanjan@console.sb4:~/exp/dl-iperf/5.2$ more data.tmp
t1key|atten|nBytes|time|kBperS
1|0|10878976|10.0952429920435|8419.01478418953
2|5|11141120|10.2252359986305|8512.2729697053
3|10|11010048|10.1255509853363|8494.94512689406
4|15|11010048|10.1354229897261|8486.67096451635
5|20|2097152|12.675085991621|1292.61450461408

Next in another window run R from the prompt. From inside the R command prompt, read data.tmp file:

R> data<-read.table("data.tmp",header=TRUE,sep="|")

Check the data that was read:

R> data
   t1key atten   nBytes     time    kBperS
1      1     0 10878976 10.09524 8419.0148
2      2     5 11141120 10.22524 8512.2730
3      3    10 11010048 10.12555 8494.9451
4      4    15 11010048 10.13542 8486.6710
5      5    20  2097152 12.67509 1292.6145

Now parse the columns into separate variables:

R> tp<-data[,5]
R> at<-data[,3]

View both variables:

R> tp
[1] 8419.015 8512.273 8494.945 8486.671 1292.615
R> at
[1]  0  5 10 15 20

Now plot throughput against attenuation. See below for example plot.

R> d <- bw ~ at
R> plot(d)

Trouble shooting & miscellaneous info

Attachments (2)

Download all attachments as: .zip

Note: See TracWiki for help on using the wiki.