Kermit Helps Track Hurricanes

[Columbia University, New York, 20 September 2004]

Photo: U of Washington Applied Physics Laboratory. Senior electrical engineers Jim Carlson and John Dunlap with the EM-APEX float.
    Scientists and engineers at the University of Washington Applied Physics Laboratory (APL) deployed five sensor-laden floats in the Atlantic Ocean in the path of the eye of Hurricane Frances in early September 2004. The floats, programmed to descend and rise repeatedly, measuring initial conditions and then changing conditions as the hurricane passed by, were dropped from aircraft the day before the storm.

The floats, manufactured by Webb Research Corporation and called Electromagnetic-Autonomous Profiling Explorers (EM-APEXs) with instrumentation designed and built at APL, measure the ocean's velocity electromagnetically, and monitor water temperature and salinity, with excursions to depths of 500 meters every 16 hours. Each made more than 100 up-and-down profiles, mostly while under water, and then tranferred back the data they had collected over the Iridium satellite cell phone system, using Columbia University's Embedded Kermit (E-Kermit), which had been embedded in the float's control processor by APL Senior Electrical Engineer John Dunlap (pictured on the right, with float), a longtime Kermit Project contributor, who is leaving Thursday (September 23rd) on a cruise to collect the floats, describes Kermit's role in the project:

Three of our floats were among the 40 scientific packages launched with parachutes from an USAF C-130 in front of Hurricane Frances when she was north of Hispaniola. The data is all good and all parties are pleased. The program is called CBLAST.

The EM-APEX float uses Embedded Kermit with an Iridium data modem, a Motorola model 9522. C-Kermit is used ashore on a Linux box with several serial ports and modems. The floats phone home, log in, then give shell commands to start C-Kermit for data transfer. Mission changes are implemented with a file sent to the float. The integration and testing of E-Kermit 1.5 into the float's Philips XAG30 microprocessor took me several days.

We switched to Kermit because the Xmodem protocol was not recovering reliably when bytes were lost. Ocean waves interupt the signals when they splash over the Iridium antenna which is only about a foot above the ocean surface. We are very pleased to say that all data has arrived intact. This is verified with an end-to-end CRC on each data group which is separate from the Kermit type-3 CRC used used in the actual transfer. The EK addition is fantastic for this application.

Columbia's E-Kermit is ideally suited for applications like this, based on more than 20 years' experience with design and implementation of file-transfer protocols that are resistent to harsh conditions, yet can go fast when sailing is smooth.


[ Top ]

The Kermit Project / Columbia University / / 20 Sep 2004