Carleton University
Technical Report TR-00-05
July 2000
Transfer of Mobile Agents Using Multicast: Why and How to Do It on Wireless Mobile Networks
Abstract
Mobile Agents (MAs) have the ability to migrate from node to node on a network in order to get closer to the resources they use. The migration capability is often used as an argument for MAs because of possible reductions of latency and network load. We study three MA migration strategies. In the three cases, an MA is dispatched from an origin node, visits N nodes, and extracts data on every node. In the first strategy, the MA visits the N nodes sequentially, then returns to the original node. In the second, the N nodes are visited in sequence, bu the MA periodically downloads it sobject state to the original node. In the third, the MA is sent in parallel to the N nodes using multicast-capable MA transfer protocol. We develoip a mathematical model of bandwidth usage on a local area network for each strategy. We exhibit conditions under which there is a decrease in bandwidth usage from the first strategy to the third.