Ethernet is nowadays the most widespread communication standard for local networks in the domestic and office environment. Its main advantages are: high bandwidth, low price of its components and compatibility with other communication standards. For that reason it is considered interesting to use Ethernet in industrial systems.

Industrial systems have additional requirements not present in domestic or office environments. Specifically, they have real-time and dependability (reliability, availability and/or security) requirements. In addition, it is not uncommon that this kind of systems are deployed in dynamic environments, that is, environments where the operational conditions can change unexpectedly. Unfortunately, Ethernet by itself does not provide the necessary services to fulfil all these requirements.

To overcome this limitation, the Dynamic Fault Tolerance for Flexible Time Triggered (DFT4FTT) project aims at providing a complete infrastructure to support applications with real-time, reliability and adaptivity requirements. Specifically, the DFT4FTT architecture is based on the Flexible Time-Triggered (FTT) communication paradigm. FTT makes it possible to exchange periodic and aperiodic traffic with different criticality levels in a real-time manner. Moreover, it allows tomodify the real-time attributes of the traffic dynamically. The DFT4FTT architecture modifies FTT to achieve high reliability by means of fault-tolerance mechanisms. This is done by replicating the network and the nodes.

The main problem when implementing the DFT4FTT architecture is that FTT was not designed having fault tolerance in mind. Moreover, fault tolerance mechanisms are typically not orthogonal to the operation of the system. Consequently, it is very costly to extend the FTT software to include these mechanisms. In this regard, it was decided to implement FTT from a new design which removes unnecessary and non-reliable functionalities, and makes room for the new fault tolerance mechanisms.

This project represents the first step towards a new implementation of FTT for highly-reliable systems. Specifically, this project consisted in the implementation and validation of a basic FTT network which can then be easily extended to implement the necessary fault tolerance mechanisms.

Slides