Skip to Main Content
In a crisis response scenario, the availability of information to first responders can be greatly enhanced through the use of mobile computing systems. However, such systems are limited by available storage space and battery life. Substantial degradation of their utility over the course of a prolonged operation is likely as batteries wear down and storage space fills up. While frequent offloading or backup of data can ensure persistence, an indiscriminate approach can accelerate the consumption of resources and further shorten the system's availability. This paper presents a method for transferring data on demand between cooperating mobile systems based on system state and both current and projected geographical location, taking into account the observed mobility of neighboring nodes. Peer selection is accomplished by first rating potential candidate machines to receive the data transfer, then further evaluating each with a utility function that conservatively estimates a predicted window of opportunity over which the transfer can occur.