The WiNGS indoor wireless testbed consists of about 60 nodes, deployed in the Department of Computer Sciences building. This wireless testbed is part of the WiNGS lab.
The current status and hardware available at each node of the testbed can be viewed at Testbed Status
The testbed is for research purposes only, and access is restricted. Only students and researchers affiliated to WiNGS lab are allowed.
Contact viveks at cs or shravan at cs, for all questions.
TODO
The controller viveks-unsup.cs is housed in the WiNGS lab, room #3360 All access to the nodes go through the controller. To gain acces to the nodes the user should ssh to the controllder, and then ssh to the nodes. The controller should not be shutdown for any reason. However, if it has to rebooted, please follow these instructions given below to bring the testbed online.
Most of the nodes are Soekris, model 4826. They are placed on the office rooms in the 3rd, 4th, 5th, 6th and 7th floors of the building. Both the main tower, and the condor tower house these nodes. Placing of nodes and configuring is a tedious process that has to go through CSL. See below for contact information. We use Power Over Ethernet (POE) to for power control of these nodes. They are connected to 5 POE switches
All the nodes can be accessed through the IP address or hostname node-. The nodes are fitted with 802.11 abg wireless cards.
There are 9 nodes with Atheros cards that are connected to the testbed. The node numbers are : 70,22,43,54,29,57,89,53,51,50
The mapping of which nodes are mapped to which OS is given in /etc/bootptab so a line like node34:tc=.client:ht=1:ha=000024C711A8:ip=128.105.102.164: means that node 34 is booted in the “client” os, that could be found in /tftpboot/client
Mapping of nodes to particular ports can be found in /usr/sbin/poe_control Please do not change this mapping unless the nodes are specifically moved from one location to another. Then still you should consult Vivek or Shravanme before you change anything.
Any user should try to use the existing setup, and try to work with that, unless, creating a new image is necessary. By default, the node should use the os named expt_client (see inside /tftpboot/). We dont have space to allow everyone to create their own images for experimentation. As I said earlier you can decide which node boots to which OS by changing entries in /etc/bootptab.
To create a new driver in the image “code_client” use the following steps.
chroot /tftpboot/code_client/ /bin/bash make KERNELPATH=”/lib/modules/2.6.16.19-soekris/build/” make KERNELPATH=”/lib/modules/2.6.16.19-soekris/build/” install
then SSH to the node and remove the existing driver using the command “rmmod” and load the new driver module using the following command
mdoprobe ath_pci
TODO