CNET también está disponible en español.

Ir a español

Don't show this again

Tech Industry

National lab's software to secure harbors?

The docks have always been a haven for skullduggery. "Orion" is being developed to give authorities a broader, clearer picture.

Oak Ridge National Laboratory is developing software that will let authorities better monitor worldwide ship traffic and, ideally, reduce the risks of disease outbreaks, terrorist attacks and other problems.

Called Orion, the software would essentially crawl the Internet for information about arrival and departure times of ships from various ports, as well as weather reports, satellite images, and cargo and crew information so that security officials can earlier identify possible security risks or problems.

The idea is to find anomalies, said Robert Patton, a research associate at Oak Ridge. A ship arriving an hour or so late might not trigger an alarm. But a ship sending weather reports from a location south of its expected course, in an area known for piracy, might.

"The question is, 'Can we try to predict behavior?,'" Patton said.

Compared to airports, ports seem to be unregulated terrain. Since the Sept. 11, 2001, attacks, security officials have contemplated scenarios in which boats laden with explosives could be pulled into a crowded city center. But ships can also be carriers of contagious diseases such as SARS. In addition, piracy and smuggling remain chronic issues.

Unfortunately, harbor authorities generally don't know what went on at sea. Potential problems such as piracy and disease are further compounded by the fact that different harbors don't always share information with each other. They keep track of individual arrival and departure schedules but might not efficiently track what is happening with the same ship on the other side of the world.

Despite a federal budget crunch, security issues have kept funds flowing to scientists and companies developing detection software, chemical sensors and other devices.

Ultimately, Orion's software agents could scrape data from thousands of sites to fill in the current gaps. By gathering disparate types of data for various sources, Patton and his group will then try to establish more precise patterns of how ships actually behave and how that pattern manifests itself in data logged into various systems. The system relies heavily on Bayesian principles of probability, in which the likelihood of future events can be gauged by past events.

"The first stage is, develop a baseline of normal," he said. "Each port has its own characteristics so it has to be tailored to each port."

Currently, Orion is being fed publicly available data such as ship schedules, data from sensors at the mouth of the port and weather reports beamed from the open ocean to the National Oceanic and Atmospheric Administration. Later, private data, such as the names of the crew and lists of cargo, will be included.

The software has been under development for between one and two years, and Patton hopes to conduct trials within the next year at one or two ports.