Realization of a distributed OS component for internal clock synchronization in a LAN environment

A key challenge in keeping distributed clocks well synchronized in a LAN environment via OS-or middleware-level mechanisms is to create an interference-free condition in a network channel(s) and a nonpreemptible on-alert condition in the responsible thread in each node during each resynchronization round while keeping the length of each round as short as possible. An approach for meeting this challenge which is based on the middleware architecture TMOSM, has been presented. A prototype implementation and its measurements indicated that with a typical network of PCs connected via a 10 Mbps Ethernet, the PC nodes can be kept synchronized within the deviation bound of 0.5 milliseconds. The organizing principle is believed to be applicable to many real-time OSs or middleware used to support distributed computing applications.