M3RT: an Internet end-to-end performance measurement approach for real-time applications with mobile agents

In this paper the M³RT (Micro Mean Message Roundtrip Time) IEPM (Internet end-to-end performance measurement) tool is proposed for supporting agent/object mobility in mobile-agent-based real-time applications. If an agent finds that the service roundtrip time (RTT) is too long, then it may try to cut it down by migrating to another idle node. Since migration is an expensive activity, the agent should decide correctly before making the move. One way of achieving sound decision making is to measure the mean RTT of the channel and to ensure that the RTT up-trend is not a transient phenomenon. The M³ RT works on past history accumulated since the commencement of the channel operation. It never has computation overflow because of it integral nature. In the M³RT development process the corresponding time/stochastic Petri net model was built for exhaustive verification. The aim is to ensure that convergence stability exists under all conditions. The input waveforms used in the simulations mimic the dynamic reality of a sizeable network such as the Internet. Preliminary and limited validation tests in a controlled environment indicate that the M³RT works equally well as its macro predecessor, namely, the stable M²RT (Mean Message Roundtrip Time) package