The ActComm project is in its final year of technology development and integration. In previous years the ActComm project team has demonstrated progress through testbed demonstrations modeled after military warfare scenarios. The overall goal is to show how technologies developed by the MURI institutions can be integrated into a transportable wireless communication system that could be used in the context of realistic military operations.
In 1998 the demonstration scenario was an intelligence gathering mission in which platoon soldiers in the field used laptops with wireless communication devices to send and retrieve data from a Battalion Headquarters (BN HQ). Database types, structures and content were modeled after actual systems used by the military. Technologies highlighted in the demonstration were Harvard’s Any Path Routing without Loops (APRL) ad hoc wireless routing algorithm, Dartmouth’s D’Agents mobile agent system, information clustering algorithms, (what else?).
In 1999 the scenario portrayed was an urban peacekeeping mission where soldiers in a platoon were ordered to surround and observe a building for looking for signs of suspected terrorists, report back observations to BN HQ, and apprehend the suspects after confirming their identity with BN HQ. The platoon segment of the testbed primarily demonstrated the networking components, as they sent back reports about their position, the building, and the insurgents, and as they made simple queries about the building and insurgents that they are observed. These simple queries were made against the black-gray-white (BGW) and SALUTE databases.
The Battalion HQ portion of the testbed demonstrated the information-retrieval components, as the HQ analysts made complex queries to determine whether the insurgent leaders were actually in the building. These queries involved the text databases, the agent construction tool MACE, and so on. We also demonstrated the networking components with BN HQ, specifically we shut down BN HQ, moved it to a new location, and turned all the computers back on, to demonstrate that network connectivity among the HQ machines was restored "instantaneously" due to APRL’s ability to discover the network topology.
The agent infrastructure played an important role in both the HQ and platoon components. First, the active-messaging system is implemented on top of the mobile-agent system. Each message is wrapped inside a mobile agent, which carries the message through the network. A second application of mobile agents in the ActComm scenario is that mobile agents that move from the soldiers’ machines into the main network perform all multi-step queries1. The agents interact with the needed databases without using the unreliable, low-bandwidth link that connects the soldiers to the main network. The agents complete the queries faster, and do not waste bandwidth by sending intermediate results back to the soldiers. Of course, with sufficient a priori knowledge of a multi-step query, database or proxy developers could implement a single high-level operation that performs the query. These developers, however, can never have a priori knowledge of all queries. Mobile agents allow the queries to be performed efficiently even when developers have not provided query-specific support.
Finally, after a soldier sends an observation to headquarters, headquarters might send back a set of pictures. The soldier confirms whether the person she saw is in one of the pictures. In the testbed, headquarters sends not only the pictures, but also the code that displays the pictures and allows the soldier to browse them. The code and pictures are sent as a single mobile agent. The picture-browsing code could have been installed on the soldier’s machine before the mission begins. The mobile agent, however, eliminates the need for the pre-installation step, something that is important if the mission is planned rapidly, and the soldier has never been involved in a “picture” mission before.
Additional details on the second demonstration can be found at:
http://actcomm.dartmouth.edu/demo2/scenario.html
1Mobile agents also perform the queries entered by the mission planners at headquarters, since headquarters itself often has an unreliable connection to the main network.