Columbia University WiMAX Campus Deployment and Installation

1. People

  • Henning Shculzrinne (PI)
  • SungHoon Seo (postdoctoral researcher)
  • Jan Janak (visiting scholar)
  • Marcus Knuepfer (visiting scholar)

2. TODOs

  1. Contact Facilities to mount pole mask and ODU on outdoor site
  2. Request permission from Dean's office, facilities, and real-estate committee
  3. Preliminary base station testing with Wi-Fi antenna by using signal power attenuator (it should be set with reduced (lowest) TX power 0 dBm) See Section 7. Indoor testing for more details.
  4. Purchase additional necessary items: a pole mask (not necessary due the new location)
  5. Ethernet activation (1402B) when all equipments are ready via Patrick (CUIT staff) It will be replaced to the other one because the antenna site was changed.
  6. Check rack space (smaller size) for IDU and ASN-GW

3. History

  1. Obtained FCC licence on Apr 14, 2010
  2. Site survey on Jun 24 and Sep 30, 2010
    • We chose Fairchild building rooftop (12 fl) for mounting WiMAX antenna + ODU and HAM radio station (14 fl) to install IDU+server(s).
  3. The first WiMAX package arrived on Jun 16, 2010
    • shipment from Rutgers University
      1. Outdoor unit x 1 EA
      2. GPS receiver antenna x 1 EA
      3. Fiber cable x 1 ROLL (?? feet)
      4. Sector antenna x 1 EA
      5. Mounting brackets for ODU and the antenna x 1 SET
      6. Power connector for the ODU x 1 EA
      7. Ligthning surge protector for the WiMAX antenna x 1 EA
  4. Buying required items (via Elias Tesfaye, officer in CS department)
  5. Visiting WINLAB, Rutgers University on Oct 8, 2010
    • Obtained the rest of the equipments:
      1. WiMAX indoor unit (IDU) x 1 EA
      2. ASN-GW Server x 1 EA
      3. IDU/ODU power supplies x 2 EA
      4. AWB US210 USB clients x 10 EA
      5. Intel Link 6250 mini PCI module x 2 EA
  6. Project request to the Facilities on Oct 9, 2010
  7. Space Unit measurement on Oct 13, 2010 @ Fairchild building
  8. Complete setting of client testbed platform on Oct 29, 2010
  9. Meeting with Facilities on Nov 29, 2010
    • At CS office: an additional committe (real-estate) admission required.
  10. Site survery for a new outdoor site (indoor site as well) on 09, 2010
    • At Fairchild and Mudd building: Site change is necessary.
  11. Meeting with electrician for the antenna mount on Dec 16, 2010
    • At Mudd 14fl and IRT
  12. Teleconf with Ivan at Rutgers Univ. on Feb 18, 2011
    • Changed several parameters on BS based on the guidance of Ivan but the BS didn't work properly. The parameters we tried to change are as follows:
      • odu_noise_floor: 4 -> 40
      • dlul_ratio: 0x02 -> 0x00
      • authgw_id: 0000000041534E47 -> 0x41534E4757303030
      • bs_tx_power: 0 -> 20
    • Planned to visit WINLAB to figure out the problems on both BS and clients
  13. Visiting WINLAB at Rutgers on Feb 24, 2011
    • Contacted Ivan Seskar and Nilanjan Paul
    • Solved every problems on both BTS and clients
    • Upgraded OMF aggregation manager with the latest version and checked out that wimaxrf works
  14. Indoor testing at IRT on Feb 25, 2011
    • Succeeded to scan and connect one of our WiMAX clients (Linux OS) to BS
  15. Purchase of a server rack cabinet and a PDU (power distribution unit) via Diasy Nguyen (CRF) and Elias Tesfaye (CS Department) on May 19, 2011
    • Triplite 12U rack ($444,73)
    • APC PDU for remote power ($425.45)
  16. Meeting with Donald A. Schlosser (Assistant Vice President of Campus Opers-Custodial Services, Columbia Facilities) on Jun 09, 2011
    • Site survey at 16th floor, Mudd building
  17. Started to mount and install WiMAX equipment on site on Jun 14, 2011
    • Technicians from Facilities and Nathan Schulzrinne (CRF intern) helped us for the deployment, including: Drilling holes on railings, fixing the mask and the antenna, mounting ODU on wall inside the building, and connecting the coaxial cable between antenna and ODU
    • Wired opticla fiber and power cables
    • Activation of Ethernet by CUIT (Two ports of Giga-bit Ethernet for Internet connection and for PDU management)
  18. Finished installation of the WiMAX BS on Jun 23, 2011
  19. First singnal measurement survey on Jun 26, 2011
  20. Measurement survey to create the coverage map on Jun 30, 2011

4. Campus WiMAX radio coverage map (expected)

a. 2D map view (by Google map) b. Aerial view (by Bing map)
Campus WiMAX radio coverage Campus WiMAX radio coverage-aerial view

5. Conceptual diagram for deployment

Diagram for Columbia Univ. Campus WiMAX deployment Diagram (ODU+antenna)

6. Site survey

Becuase of the various reaons, on Dec. 16 2010, we decide to change the outdoor and indoor sites to the southwest corner of Mudd building, 16 fl, rooftop.

New outdoor site at Mudd 16th floor

a. Wide view b. front view c. SW corner d. side view e. detail
(new) outdoor site 1 (new) outdoor site 5 (new) outdoor site 6 (new) outdoor site 3 (new) outdoor site 4

line-of-sight view 1 LOS view 2
line-of-sight line-of-sight 2

New Indoor site at Mudd 14th floor (neer outdoor site, same level)

(new) rack space
  • Need to check the space availabilitye: 5U + pwr suplies.

See Old outdoor/indoor sites.

7. Indoor testing

This section describes a note for the purpose of preliminary base station test at indoor environment.

Important notes

  • It is strongly recommended that connecting to the IDU (CHC's serial port) is allowed through `serial cable' by using any model control application (e.g., minicom) for the monitoring purpose. However, it can be substituted with ssh connection.
  • Preliminary base station testing with the sector-antenna: Signal power attenuator should be located between ODU and the antenna (bs_tx_power should be set in 0 dBm)

Testing topology

  1. ASN-GW
    • eth0: -> connected to the monitoring console machine
    • eth1: -> connected to the IDU
    • eth2: dhclient -> connected to the CU networks
  2. IDU
    1. CHC:
    2. NW INTFC:
  3. Monitoring console machine
    • eth0: -> directly connected to the eth0@ASN-GW

Booting up sequence

First time, whenever we tried to boot up the ASN-GW server, IDU, and ODU, NW INTFC and CHC at IDU indicated that several warning and alram lights were turning on.

a. ALM light (red) at NW INTFC: It's occurred by GPS.
b. WARN light (orange) at NW INTFC: It indicates that "Real time clock seems stuck!" which is caused from time sync issues because GPS receiver is not connected at this time of testing.
c. ODU ALM light (red) at CHC: Setting up ODU fails.
d. ODU LINK light (green) at CHC: When it's on, it means the connection between IDU and ODU is successfully established.

We found that there is a turn-on sequence for these machines to work properly, which is as follows:

  1. Turn on the power of ASN-GW server
    • Send ping to (IP address of ASN-GW) from monitoring console to check the server alive.
    • ssh
  2. Turn on the power of NW INTFC
    • Send ping to (IP address of NW INTFC) from a console at ASN-GW to check the NW INTFC alive.
    • telnet
  3. Turn on the power of CHC
    • Send ping to (Ip address of CHC) from a console at ASN-GW to check the CHC alive.
    • telnet
  4. Restart the asn-gw module at a console of ASN-GW
    • sudo /usr/bin/asn-gw stop
    • sudo /usr/bin/asn-gw start // this command will reboot the CHC (internally bs_restart is called)
    • Wait until CHC is ready by sending ping to the (IP address of CHC)
    • Check the ODU LINK light on at the CHC panel!!
  5. For the logging information, manually synchronize the time between machines.
    • We turn on ntpd at asn-gw for the testing purpose
    • ntpdate (IP address of the ASN-GW)
    • export TZ=EST
    • NOTE: At CHC, TZ variable is automatically recovered to the JST.


  • Scanning test (02/10/11)
    • As soon as we checked the BTS works properly, we did a wide range of test for scanning procedures from linux based WiMAX client, but failed to see the GENI id'ed base station (NSP=51).
    • Tried to change the bs_tx_power to the 10, but nothing changed.

8. Client Setting

  • Platform: Dell Latitude D830 (IRT0045)
  • NIC: Intel Centrino Advanced-N WiMAX 6250 - Mini-PCI express with half size slot
  • OS: Ubuntu 10.10 Maverick with patched linux kernel 2.7.36-rc3

NIC installation

See network interface card installation for more detail.

Software installation

  • Required software (from  http://linuxwimax.org)
    1. NIC firmware: i2400m-fw-1.5.0
    2. Driver:compiled and built by patched linux kernel
    3. WiMAX tool: wimax-tools-1.4.4
    4. Open source supplicant: wpa_supplicant-0.7.2 with patch to build libeap.so library
    5. WiMAX network service: wimax-1.5.1, Intel's WiMAX daemon
  • History
    1. Initial compilation and installation (10/28/10)
      • It seems like working properly.
    2. NSP definition does not detect the geni NSP id 51 (01/07/11)
      • This issue has been discussed on linuxwimax.org mailing list
    3. Patch for GENI NSP ID '51' (02/02/11)
      • Update WPA supplicant (wpa_supplicant-0.7.3 + wpa_supplicant-0.7.3-generate-libeap-peer.patch
      • Replace XML file on wimax-1.5.1 (append geni NSP into ~/InfraStack/OSAgnostic/WiMax/Agents/NDnS/XML_Files/NDnSAgentConfig_forDriver.xml)
        • Note: Don't miss running ldconfig after the compilation completes.
      • Replace GENI definition and database files (WiMAX_Def.bin and WiMAX_DEF.bin obtained from orbit-lab) on ~/InfraStack/OSAgnostic/Product/AppSrvInfra)
        • Installation completed and checked working properly
        • Note: The file of /var/lib/wimax/WiMAX_DB.bin is changed whenever performing 'wimaxcu scan'

9. Signal Strength Measurements

We measured the signal quality such as Received Signal Strength Indication (RSSI) and Carrier to Interference plus Noise Ratio (CINR) across campus using the 'wimax_gps_oml2' tool. The tool writes the current location obtained from a GPS received connected via USB and various WiMAX signal parameters and writes them into an sqlite based database. The data from the database was then used to plot the results using  http://gpsvisualizer.com.

The base station TX power was set to 40 dBm. We also had to fix the 'wimax_gps_oml2' tool in the process to make sure that the measured WiMAX signal strength is paired with correct GPS location.

a. Coverage map of measured RSSI b. Coverage map of measured CINR
RSSI coverage map CINR coverage map

The WiMAX base station is in the top right corner of the map. The location of the antenna is shown marked with a small red arrow. The antenna is aimed towards the building labelled as "Columbia University" in the center of the northern part of the campus.

Since the antenna was installed in a location we could reach easily, rather than the perfect location, we were trying to cover only the northern third of the campus, above Low Memorial Library. It turned out that even from that antenna location most of the campus is covered, which is much better than what we had hoped for. We used a laptop with an Intel 5150 WiMAX adapter. The laptop could connect to the base station from all yellow, green, and blue areas. With default values for most parameters on the base station, the download speed measured in selected green spots was about 8 Mbit/s. Columbia campus, being located in Manhattan, is comprised mostly of high-rise buildings. Nevertheless, we can connect to the base station from most publicly accessible areas.

Last updated on

Error: Failed to load processor Timestamp
No macro or processor named 'Timestamp' found