New architecture of industrial systems with real-time ATM middleware

We propose a programming model, a real-time client-server model, for distributed real-time systems for monitoring and controlling plant. With the model we can make a system scalable in a flat architecture with an ATM network. The system can be scalable because it has no bottlenecks in plant data transmission as traditional systems has with their cyclic transmissions. On-demand transmission of plant data realizes this scalability. The architecture can be flat with the ATM network integrating various kinds of traffics transmitted separately in traditional systems. We also propose using our middleware which provides application program interfaces (APIs) based on the model and guarantees application-to-application QoS. In addition, to achieve on-demand transmission of plant data, we propose a concept of selective real-time channels. We show an example of an application with our middleware, a plant monitoring system with an operator station and a programmable logic controller.