http://cartel.csail.mit.edu/ 2009-11-06T16:10:32-05:00 http://cartel.csail.mit.edu/ http://cartel.csail.mit.edu/lib/images/favicon.ico text/html 2009-11-06T16:03:08-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:meeting11-6-09&rev=1257541388&do=diff Meeting Notes for 11/6/09 Status update: The four boxes are experiment-ready. We are left only to finalize the click configuration desired for the experiment (as well a few other minor details). We should begin the logistical planning for the three-node vehicle test. Perhaps next weekend is an appropriate time? text/html 2009-11-06T10:16:03-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:fordexpspec&rev=1257520563&do=diff Ford Experiment Specification This page contains an evolving draft of the specifications for a planned multi-vehicle mesh experiment to be held at Ford. Proposed Date The first two weeks of January, 2010. Vehicle Requirements The experiment requires somewhere between 30 to 50 vehicles with drivers. There is no particular requirement on the type of automobile, if anything diversity is an advantage. text/html 2009-11-06T10:12:20-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:boxstatus&rev=1257520340&do=diff Box Status Box Label Status A Configured to Experiment-Ready (with exception that current click script is not fully functional) B Unknown C Unknown (Unlabeled; in Lenin's office) Unknown text/html 2009-11-06T10:03:04-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:fournodevehicletest&rev=1257519784&do=diff Three Node Vehicle Test The goal of this experiment is to drive three cars, in a real road environment, each running a mesh box. Around thirty minutes of drive time, with a check in on the data every 10 minutes, should be sufficient. The goals of this experiment are two-fold: First, to smoke out any problems or issues that come from longer term deployment of boxes in an experimental setting; Two, these logs gives allow us to begin building the log parsing and analysis framework in the backgrou… text/html 2009-11-06T09:56:25-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:fournodebenchtest&rev=1257519385&do=diff Milestone: Four-Node Bench Test Overview The goal of this experiment is to have four nodes configured to experiment-ready status with the the box configuration document updated to capture the steps this required. These configured boxes will then be activated outside in several positions where their GPS units can get a lock. We will toggle the power off and on non-deterministically. The experiment will be considered a success when we can consistently generate good logs from any such sequences … text/html 2009-11-06T09:53:57-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:clickrequirements&rev=1257519237&do=diff Experiment-Ready Click This page describes the desired features and open questions regarding the click configuration to be used in our 10-node experiment. Desired Features * Logs sent messages, received messages, and received beacons. * Uses unique file names, based on timestamps, for the logs. * Starts new logs on a regular interval to prevent data loss after accidental power down (and to simplify data offload). * Reads GPS position and time directly from gpsd. Logs every entry with… text/html 2009-11-06T09:46:40-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:boxconfig&rev=1257518800&do=diff Configuring the Linux Boxes Overview This page contains notes on how to configure the linux boxes into experiment-ready condition (defined below). Experiment-Ready Condition To be considered in experiment-ready condition, a linux box must have the following behavior on start-up: text/html 2009-11-05T11:08:53-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:issues&rev=1257437333&do=diff Issues and Workarounds This pages captures issues with the boxes and their solutions, as they are discovered. GPSD Starts Automatically During Boot Desc: Even if you configure gpsd to not start at boot (using dpkg-configure as recommended in documentation), it still starts during the boot process. text/html 2009-11-04T10:19:37-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:boxuse&rev=1257347977&do=diff Using the Linux Boxes This document describes information needed to log in and use the linux boxes. Connecting, Logging In, and Initializing To connect through a serial connection: cu -l /dev/ttyS0 -s 38400 (The actual device name may differ from ttyS0 depending on you system setup.) text/html 2009-10-23T09:19:04-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:conncond&rev=1256303944&do=diff Characterizing Connectivity Conditions Useful and concise descriptions of the conditions under which vehicular networks reach various connectivity thresholds. The approach we have in mind is to first gather pairwise connection data from as 10-car experiment and then use this to build and validate a connectivity model. Using this data, combined with large traffic traces, we can overlay car to car connections, and start building connectivity metrics and models. One approach is to first character… text/html 2009-10-22T16:13:02-05:00 cnewport http://cartel.csail.mit.edu/doku.php?id=mobilemesh:fornodevehicletest&rev=1256242382&do=diff Four-Node Vehicle Test Overview This experiment deploys boxes in four cars that drive a diversity of patterns, the goal is to both gather preliminary data that can inform the design of “the game,” as well as shake out deployment-generated bugs.