Ad hoc Routing: We will implement, test, and evaluate several ad hoc routing algorithms including APRL (Harvard), GPSR (Harvard), STARA (UICU), AODV (UCSB). Experiments will be carried out on the Dartmouth campus over the spring and summer. Experimentation will include measurements of required overhead, reactivity to position, and packet routing, etc. time for each algorithm. The results of study will be presented at the final review. POC: Bob Gray
Network Sensing: RPI has developed the network sensing mechanism that can be incorporated into the demo in several ways. One way is to display the network link “health” to the user so that they can make decisions regarding their movement. A soldier moving out of communication range might alter their movement to stay maintain connectivity. Another option is to let agents make movement decisions or deploy a relay system based on the strength of the network connections provided by the network sensing software (see below). One scenario where this kind of decision making is plausible is when agents need to execute scheduled tasks, such as delivering logistics reports every 12 hours. If the agent responsible for delivering the report sees that is it close to time for delivery, and that the network conditions are worsening, it can choose to move before the connection is lost. POC: Lisa Shay
Task Scheduling: The goal of this research is optimize where and how agent tasks should be executed so that overall system throughput can be maximized. In the demonstration scenario this can be shown by having multiple places in the network where queries can be executed and letting the agents ask the task scheduler to determine the sequence and location of task execution. This work will be demonstrated in the wired part of the demonstration network because this is where the majority of queries take place. POC: Rong
Market-based Resource Control: The development of the market-based resource control algorithms is intended to provide prioritization of agent task execution based on the available resources, the contention for the resources, and the relative "wealth" of the agents requesting service. In the demonstration scenario, agents created by personnel would be given wealth in proportion to the information priority. That is, an authorized soldier would assign an agent a routine, priority, immediate, flash and flash override. As a result, agents created with a flash assignment would get precedence over agents assigned routine. This work will be demonstrated in the wired part of the demonstration network because this is where the majority of queries take place. POC: Jon Bredin
Information Clustering: The clustering algorithms developed by Dartmouth have been demonstrated as a side technology in previous demonstrations. We would like to emphasize this technology by having the algorithms be more tightly coupled with the demo scenario in some way. We will create a GUI to be integrated into the rest of the analysts tools to enable better use of the clustering technology . POC: Daniela Rus
Robotic Communications Relays: The notion here is to use deployed autonomous entities as relay stations for wireless communications. A soldier moving out of range will activate the relay to maintain connectivity to the network. We will use an existing robot from Dartmouth to act as the host for the communication link, and this work will be performed in conjunction with the ad hoc routing experiments. POC: Daniela Rus
D’Agents:The latest version of D’Agents will be incorporated into the demo. POC: Bob Gray