In industrial systems, there are a number of devices with low processing capacities called field device. Currently, these devices (sensors and actuators) are connected by dedicated wires, often of a considerable length. The idea of replacing bundles of cables with one field-bus, linking up a set of intelligent field devices seems to be a valid solution to the problems presented by point to point connections.

Usually in fieldbus systems, a variety of different nodes, sensors and local node applications. Differences stem from varying hardware architectures, vendors, and sensor types. In order to guarantee their inter-operability we have to introduce abstraction mechanisms that hide these inherent differences from the other nodes in the network[1].

In fieldbus, the cyclic time-critical traffic generated by the exchange of measurement and control data, normally used by the system, will be accompanied by another kind of traffic (acyclic traffic). Scheduling in fieldbus is characterized by the presence of a processing unit acting as a Link Active Scheduler (LAS), whose task is to manage the bandwidth and distributing it among all the producing devices.

In this paper, recent evolutions of the initial protocol definition concerning transmission of synchronous and asynchronous messages are presented. The performance of the Link Active Scheduler (LAS), is also discussed, which assigns the stations capable of sharing the fieldbus communication capacity, called Link Masters (LMs), the right to transmit by using token mechanism. In the present work, a general semi-Markov model is developed for a field bus system with priority levels associated with the requests. The paper presents a model for handling the on-line scheduling of acyclic processes. Based on the priority scheme the model simulation results provide an optimal allocation method for acyclic traffic.