Analysis of Two Mobile Agent Data Migration Patterns

Mobile agents has found increasing used in the development of applications for distributed systems such as the Internet based on the premise that it often reduces network load and improves performance as compared to other design paradigms as client-server. However, studies have shown that mobile agent often performs poorly as compared to client-server. The poor performance of mobile agent is attributable to its too simple existing migration pattern. In this study, the migration characteristics of a single agent system are analyzed when the migration pattern for the data collected on each host is different. Mathematical models were developed for each stage of the migration process and programmed in Java, new migration pattern is developed and compared with the existing pattern for situations when data is migrated or not migrated. The results show that the transmission time is better for the new data migration pattern when an agent migrates with data.