Version 38 (modified by, 12 years ago) (diff)


WiMAX Campus Deployments and Experiments

Friday, November 4, 2011, 1pm - 3pm
Room: Trianon A/B
Session Leader: Harry Mussman, (GPO, Raytheon BBN Technologies)


This session will first review the GENI WiMAX strategy, and then concentrate on plans for the new WiMAX site deployments and extensions that will be supported by Solicitation 3 projects. Next, goals for GENI WiMAX experiments will be presented, and finally current and planned experiments will be discussed.



This is a tentative agenda, which may change.

NOTE: We have many topics to cover in a short time. PLEASE bring only the designated number of slides, and provide a link to additional information. THANKS!

1) Introductions


2) GENI WiMAX Strategy

Harry Mussman (GPO)

a) Goals for GENI WiMAX site deployments:

Satisfy needs of local research experiments
Support multi-site experiments, e.g., MobilityFirst
Support remote experimenters

Support multi-site applications?

Are we missing something? support for instruction?

b) Sol 2 base station deployments:

Update to include OMF/OML structures, including LoginService

Complete, and update to new software that support multiple VLANs, etc.
Connect via switch to I2 backbone, via OF switches if there

Consider how to setup multi-site slice, and include other GENI resources
Consider federated operation
Consider adding WiFi AP to support dual-homed experiments, e.g., MobilityFirst

c) Sol 3 base station deployments:

Consider 3.6GHz operation?
Study commercial campus deployments, to understand coverage and applications: U Colorado, Boulder; Northern Mich U; UMass Dartmouth
Order Airspan base stations and related items, for delivery 1/2012
Complete installation within Spiral 4! (within 1 year)

Decide on how to provide ASN-GW for handover
Consider open-source ASN-GW? or ASN-GW from Airspan?
How could we use Wisconsin mobility engine?

Airspan Deployment Plan for Univ Colorado at Boulder

d) Mobile stations:

Reference MS is Linux netbook, with Intel WIMAX modem card, internal or external
Reference vehicular MS?
Reference handset MS? Need to be able to unlock Sprint-supported device?

e) Experiment support:

Move to OMF/OML, for better support and for easier sharing
Consider alternate techniques for throughput measurements
Consider how to setup multi-site slice, and include other GENI resources

GIMI project committed to supporting WiMAX sites

How can we make it easy for experimenters?

f) Other topics:

Interworking with carriers (Clemson)
Use of carrier resources, i.e., WiNTeB
How could we add LTE? any match of equipment, frequencies and mobile stations?


Do not have resources to do all of the above! how do we decide?
What is the best way to support campus projects?
What is the best way to support experimenters?

3) New (Sol 3) GENI WiMAX Site Deployments and Extensions


3.1) Airspan Profile-C Base Station, with Associated Hardware and Software

Ivan Seskar (WINLAB) and Harry Mussman (GPO)

a) Choice of Airspan to provide a WiMAX base station for new GENI WiMAX sites
NEC is not able to provide a WiMAX base station for new GENI WiMAX sites
WINLAB and the GPO have decided to purchase base stations from Airspan, a commercial company pursuing target customers such as universities and small carriers.
For example, the Univ of Colorado at Boulder IT Dept has installed an Airspan system with 4 sectors, that covers the campus. They have been pleased with Airspan.
WINLAB has a frim quote from Airspan for their MacroMAXe profile-C WiMAX base station unit, and related items.

Airspan brochure
MacroMAXe Product Specification
Airspan Deployment Plan for Univ Colorado at Boulder

b) Rutgers will provide a kit including:
Airspan MacroMAXe profile-C WiMAX base station unit, for outside mounting

Choice of frequency:
(2510 Lo: 2496MHz to 2570MHz)
(2510 Mid: 2560MHz to 2630MHz)
(2510 Hi: 2620MHz to 2690MHz)
(3650: 3650MHz to 3700MHz)]

NOTE: WINLAB needs to know frequency choices during November (or early December), so that they can place an order with Airspan for delivery late 2011 or early 2012!

Sector antenna, with x4 RF jumper cables
2x4 MIMO, supported by base station and antenna
GPS antenna, with surge suppressor and cable
DC power cable, from base station unit (outside) to testbed (inside)
Network management software license, 1 per base station

Figure 3-1 Airspan MacroMAXe Base Station with Antenna

c) Campus will need to provide:
Pole for mounting
DC power supply (-48v, approx 300W)
Ethernet cable, from base station unit (outside) to testbed (inside); consider copper (may be a problem with lightning surge) or fiber (requires MM fiber cable, and fiber interface on switch)
Windows server to host network management software
Linux server to host OMF software
Ethernet switches, and other network equipment, for access (configuration to be provided)

d) Additional kits can be purchased from Airspan

Price is approximately $15,000.

e) It is expected that additional base stations kits will be purchased for some (or all) Sol 3 GENI sites in Spiral 5.

Thus, campus site might be able to add additional sectors, at no cost.
Planning for next year will be done later this year.
Consider including additional sectors within your technical plan right away.

f) Note: Airspan indicates that they expect to introduce a new base station unit within the next year.

Somewhat reduced performance/range.
Approximately half the cost

g) Airspan indicates that their sector antenna provides:

A 90degree pattern, which is consistent with proper coverage over a 120degree sector.
Four internal elements, to support 2x4 MIMO.
Airspan can provide provide x4 omni-directional antennas, but the cost is much higher, and the range is signifcantly reduced.

h) Airspan shows that their profile-C base station unit can be configured to operate without an ASN GW and without a RADIUS (authentication) server.

In this mode, a list of allowed base stations must be configured into the base station, and groups of these base stations are mapped to a VLAN on the ethernet port.
This appears to be consistent with the current NEC base station unit operated with the ASN GW/WiMAX RF AggMgr code that has been provided by Rutgers/WINLAB
MacroMAXe Configuration Guide

i) Airspan shows that if handover operation is desired, an ASN GW must be provided, plus typically an associated RADIUS (authentication) server.

In this case, the base station unit operates with IP at the interface.

3.2) WINLAB/Rutgers University Site Extension Plan

Ivan Seskar (WINLAB)

At WINLAB: Replace NEC BS on roof with Airspan BS (currently at WINLAB)
Retain at WINLAB: NEC BS in sandbox

At Bush Campus: Install Airspan BS
Choice of frequency: 2610MHz

x1 (2510 Mid: 2560MHz to 2630MHz)

3.2) UCLA Site Extension Plans

Giovanni Pau (UCLA) and Mario Gerla (UCLA)


Add x1 Airspan BS near UCLA dormitories; possibly other BSs.
Choice of frequency: 2572MHz

x1 (2510 Mid: 2560MHz to 2630MHz)

3.3) Clemson University Site Deployment Plan

K-C Wang (Clemson)

Add x1 Airspan BS near campus
Add x2 Airspan BS near Greenville; connect via DOT fibers.

Want handover near I85/I385 intersection.

Utilize Clemson existing Educational Broadband Services (EBS) license?

3.4) University of Wisconsin Site Extension Plan

Suman Banerjee (Wisconsin)


Considering expansion to two possible locations: Madison Metro; and Dept of ? near Capitol.

x1 (2510 Mid: 2560MHz to 2630MHz)

Connect via Madison unified fiber project
Outside contributions for power and poles.

3.5) University of Michigan Site Deployment Plan

Morley Mao (Michigan)


No EBS license; working with Wayne State on licenses.

Identified several roof-top locations; aesthetics approval required.

Will signals penetrate into buildings?

3.6) Wayne State University Site Deployment Plan

Hongwei Zhang (Wayne State)


Considering: one site with three sectors; three separated sites.

Wayne State had EBS spectrum in 2.6GHz band, but leased out to Clearwire.
Talking with Clearwire about loan of frequencies; concerns about their need to change frequencies, and radiation into Canada.

Possibly consider 3.65GHz band, with Intel 6250 modem, where no interference? Concerns about: coverage; lack of MS devices.

3.7) Discussion of Deployment Issues and Suggestions from Current Sites

Moderator: Harry Mussman (GPO)

Round-robin discussion, with a focus on deployment issues.

NOTE: Sol 3 (and Sol 2) projects: FORWARD your issues ahead of time, and they will be summarized here.

4) GENI WiMAX Experiments


4.1) Goals

Harry Mussman (GPO)

4.2) Throughput Experiment Using iperf

Manu Gosain (GPO) and Harry Mussman (GPO)

Overview of basic throughput experiment, using bidirectional iperf, both TCP and UDP

Basic throughput experiment using OMF/OML

The basic throughput experiments we have done utilize iperf, both TCP and UDP.

In iperf, the data is generated in the client, and flows to the server.
We put the client in the Mobile Station, and the server in the Base Station.
We used -d dualtest (bidirectional mode), where test is initiated at the client, data begins to flow to the server, and then a second data flow starts at the server; at the end of the test, results are available at the client.
Because of this, all tests can be initiated at the Mobile Station, and then results are available there.

iperf results in the TCP mode depend upon buffer sizes; overall delay; and lost packets.

Because of wireless propagation conditions, lost packets are common, and slight changes can significantly affect the measured throughput.
Thus, TCP results are highly variable; we took multiple measurements at each point, and identified the best and worst results.
It would certainly be good to have a better way to evaluate available channel bandwidth.
On the other hand, most apps use TCP and the variable results are typical of how these apps would see the channel.

iperf results in the UDP mode, counts % packets received, for a given (fixed) transmit rate.

If the rate is set below the available bandwidth, typically 100% of the packets are received.
If the rate is set above the available bandwidth, typically % of the packets are received typically equals available bandwidth divided by offered bandwidth, but there is no way to understand how many packets are actually lost.
An extended test that ramped up the offered bandwidth in multiple tests, could actually verify the available bandwidth; this could then be repeated to see real variations in available bandwidth.


iperf tutorial slides
iperf tutorial web site

4.3) Throughput Experiment Using Bit Torrent

Fraida Fund (NYU Poly)


Overview of throughput experiment, using bit torrent.

More information & source code
Use the NYU-Poly WiMAX testbed

4.4) Raw IP and UDP Traffic Generators

Surat (Au) Teerapittayanon (MIT)

Overview of new raw IP and UDP traffic generators, to accurately gauge available channel bandwidth.

4.5) Using OMF and OML in Your Experiment

Christoph Dwertmann (NICTA)


Basic throughput experiment using OMF/OML

OML'ified apps we know of include:

otg and otr
Yantt (Yet another network testing tool)
SNMP wrapper

Repository for OML'ified apps

List of available modules with OMF and OML for use in your experiment
Approach for adding OMF and OML interfaces to additional modules

4.6) Multiple Site Experiment: Mobility First

Kiran Nagaraja (WINLAB)


Overview of current Mobility First multiple-site experiment.

What additions are needed?
1) Add WiFi AP to each site.
2) L2 path from MS to core

Experiment done without human intervention, by using x2 MAC addresses, and emulating packet loss to fake mobility.
Consider these as possible sites by GEC13: WINLAB; BBN; NYU Poly; UCLA; Colorado.

4.7) Experiment Plans and Issues

Moderator: Harry Mussman (GPO)

Round-robin discussion, with a focus on experiment design and issues.

NOTE: Sol 3 (and Sol 2) projects: FORWARD your issues ahead of time, and they will be summarized here.

5) GENI WiMAX Strategy Issues

Moderator: Harry Mussman (GPO)

Intend to add: central OML Server to accept data from all sites.
Issue: need to maintain anonymity?

6) Adjourn


Attachments (14)