Designing multi-agent systems based on pairwise agent interactions

Systems comprised of multiple interacting mobile agents provide an alternat e network computing paradigm that integrates remote data access, message ex change and migration; which up until now have largely been considered indep endently. On the surface distributed systems design could be helped by a co mplete specification of the different interaction patterns, however the num ber of possible designs in any large scale system undergoes a combinatorial explosion. As a consequence this paper focuses on basic one-to-one agent i nteractions, or paradigms, which can be used as building blocks; allowing l arger system characteristics and performance to be understood in terms of t heir combination. This paper defines three basic agent paradigms and presen ts associated performance models. The paradigms are evaluated quantitativel y in terms of network traffic, overall processing time and size of memory u sed, in the context of a distributed DB system developed using the Bee-gent Agent Framework. Comparison of the results and models illustrates the perf ormance trade-off for each paradigm, which axe not represented in the model s, and some implementation issues of agent frameworks. The paper ends with a case study of how to select an appropriate paradigm.