HomePublicationsMSc Thesis

Arquitecturas de Comunicação Industriais para Suporte a Sistemas Computacionais Móveis
Ref: HURRAY-TR-071201       Publication Date: 14, Dec, 2007

Abstract:
In the analysis of real-time distributed systems one of the most important objectives is to foresee the timing behaviour of the system. In order to do this, two approaches are commonly used: worst case analysis and typical behaviour analysis, each allowing to extract distinct system characteristics, which are complementary to each other. The use of both approaches allows us to better understand the system timing behaviour.
On one hand, worst case analysis establishes a minimum expected system behaviour, on the other hand its underlying pessimism leads to results that are far from real system behaviour (the most probable to occur in reality). To obtain a more probable behaviour occurring in distributed system, the discrete-event simulation technique is commonly used.
The European RFieldbus project (IST-1999-13316) aimed the expansion of one of the architectures integrated in the European standard EN50170 (PROFIBUS, WorldFIP and P-Net) in order to include new important functionalities. Within the project, the architecture of RFieldbus communication network was developed, based on the PROFIBUS architecture, and a test application (MAF – Manufacturing Application Fieldtrial) based on typical discrete parts manufacturing system has been implemented in the Critical Systems Lab of the IPP-HURRAY Research Group. In this test application, system components are interconnected by an RFieldbus communication network, which comprises and demonstrates the architecture novel functionalities: wireless communication, efficient control traffic scheduling algorithm, multimedia traffic integration and wireless devices mobility.
RFieldbus, as an industrial communication network, is embedded in distributed systems available at the manufacturing floor, thus its behaviour affects the overall system performance. For this reason, the study of the communication network behaviour is important, since it will contribute to the understanding of the distributed system behaviour. A software application has been developed within the RFieldbus Project to compute the worst case behaviour of an RFieldbus network. The application has been used to compute this response for the MAF topology. Later, the MAF behaviour has been observed and both responses have been compared in order to understand the pessimism introduced by worst case analysis. A discrete-event simulator (LLRS - Lower Layer RFieldbus Simulator) was developed aiming the simulation of RFieldbus typical timing behaviour. The developed model, which was implemented in the simulator, incorporates device behaviour and parameters, network and message cycles that significantly contribute to the RFieldbus network behaviour. The simulator was validated with the observed data on the MAF application, and allows, in a simple and inexpensive way, to study RFieldbus network behaviour considering different elements, topologies, message cycles and parameters.
This dissertation presents the underlying theoreticians, the model, the simulator LLRS, the obtained results and their validation.
Keywords: Fieldbus, communication networks, Discrete-Event Simulation, Distributed Computer Controlled Systems, Real-time systems, RFieldbus, PROFIBUS (EN50170), OMNeT++.

Veríssimo Brandão Lima

Document:

View
PDF Download
PDF