Plan for the NSF Demo

---

Update: Demo 0 has been cancelled

Demo 0: Add wireless to Planet Lab experiment

• Wired + wireless service
• Video streaming
• Planet Lab coming into orbit
• No virtualization involved in this demo

Demo 1: Integrated Planet Lab- Orbit (Video demo)

Motivation

• Extending Orbit Framework to PlanetLab
• Modifications to nodehandler script
• Measurement support
• Node addressing and transport protocol for communication

Plan

• Show the earlier PL-ORBIT video demo with the script explaining how we integrated the two together
• No virtualization in this one
• Show video traffic received

Status

• ?

Problems

• ?

For detailed Information on this click here

Demo 2: Planet-Lab orbit integration + Virtual AP

Motivation

• Virtualizing AP (VMAC)
• Integration of Planetlab - Orbit Framework

Plan

• Show four experiments running using the same channel on one AP
• Each experiment may use a different number of orbit nodes
• Each experiment connects to an orbit node

Status (Rajesh - date?)

• Tested successfully for 1 VAP and 4 clients on 4 different networks but the same channel.
• Three Planet Lab Nodes were used in the experiment.
• One node was streamin video and the other 3 nodes were pumping data to the PL nodes using iperf.
• The testing was done with the control framework extended to the PL nodes … so the experiment was fully automated.

Problems

• The OML gives problems. If OML is included in the script, the application does not start on the orbit node.

Break

Move over to virtualizing the wireless space

Demo 3: Virtualization Demo

Motivation

• Enable multiple users to run experiments simultaneously
• SDMA + FDMA + virtual console support

Plan

• Show scheduler (open hours), logging into vconsoles and running experiments
• Four experiments are deployed on each vconsole
• Two experiments must be identical(simple Sender-Receiver exp.)
• Show that similar expts yield similar results (indicating no interference)
• Other two experiments are star-based topologies to signify the increased capabilities in handling multiple different experiments

Status (Updated : Sachin Jan 22 07)

• Tried to use vconsole2. Idea was to imageNodes
• Roadblock : CMC service that does not recognize requests coming in from a vconsole domain and ignores request to power nodes on
• Ivan, Joe to fix this- Until then other vconsoles also need to be checked

Problems

• ?

Demo 4: Splitting a Node (Interfaces)

Motivation

• Show hardware virtualization by assigning full control of one interface to a virtual machine
• Need to add access control in nodehandler for N0/N1 interfaces

Plan

• Run two experiments on one orbit node
• One experiment may communicate to one Planetlab node and one orbit node
  • Topology: Planetlab—Orbit1(AP)—Orbit2(client)
• The second experiment may connect to other Orbit nodes
  • Topology: Orbit(AP)—Orbit(1,2,3)

Status(George-Jan 22)

• Obtained Xen and UML images from the Suman Banerjee's group
• Contacted Thierry Parmentelat working with OneLab​http://onelab-private.inria.fr/
  • Onelab aims to add these capabilities that do not existing in PlanetLab
    • Use multiple interfaces (wireless & wired)
    • Execute monitoring: passive monitoring and topology info components
    • Waiting for response
• Contacted Aki Nakao
  • Requested copy of Xen Slice (Xlice)
  • Waiting for response

Problems

• None at this point

Demo 5a,5b: Sensor Network Application Demo by Dr. Manish Parashar's group.

Motivation

• Show applications that utilize Orbit and/or Planetlab

Plan

• Show two demonstrations
• a)Data collection from oil wells
• b)Data reduction and aggregation from sensors

Status(George-Jan 17)

• Provided a preliminary demo
• Working on improving applications and visualization

Problems

• None at this point

Demo 6a,6b: TDMA virtualization demo by Dr. Suman Banerjee's group

Motivation

• Show an implementation of TDM-based virtualization

Plan

• Show two different experiments
• Each experiment will operate in a different wireless channel
• The first one will be a linear ad-hoc routing topology
• The second one will be a star AP-client topology
• One exp. will use TCP-based applications and the other will use UDP-based applications
• Solution is built using UML (User mode linux)

Status(George-Jan 17)

• Currently testing with 5 Orbit nodes

Problems

• None at this point