A fiber-optic-based protocol for manufacturing system networks
Asok Kumar Ray, Arun Ayyagari · 1989
This research work presents the development and analysis of a novel fiber-optic-based medium access control (MAC) protocol for integrated factory and office communications over a common transmission medium. The heterogeneous traffic in such a network can be broadly classified into two types for real-time and non-real-time data communications. The physical topology for the proposed MAC protocol has a unidirectional dual-bus architecture. The protocol execution is divided into two parts: one for the execution in the real-time mode and the other in the non-real-time mode. Within each mode the access to the medium is collision free, and the access right is established by an implicit token passing scheme. The protocol enables a selective preemption scheme whereby the non-real-time transmission mode can be terminated by a station with a real-time message. The preemption and controlled access scheme within the real-time mode ensures that the real-time messages meet the requirements of the data latency bound. The non-real-time message transmissions are permitted when the channel is not used by the real-time data. An acknowledgement scheme would ensure integrity of the non-real-time messages. The details of the protocol are represented by a finite-state-machine model and a timed Petri net model. Correctness of the protocol has been verified from this Petri net model by use of the global state generation technique. A statistical model is formulated for the performance evaluation of the proposed protocol. The statistical model derives the Laplace-Stieltjes transforms of the queueing delay distribution for both real-time and non-real-time messages. A discrete-event simulation model has been developed from the Petri net model to analyze the performance (e.g., network-induced delays for different classes of messages) of the protocol under various operational scenarios of the network. This simulation model has also been used to verifying the statistical model and vice versa.