Supervisor/s Julián Proenza Arenas | Luís Almeida
In the last years industry has shown a growing interest in using Ethernet as the protocol for developing distributed embedded systems, even in the automation, avionics and automotive fields. This coincides with a growing interest in the use of multimedia-based applications for entertainment (e.g. video streaming) as well as for assistance (e.g. Advanced Driver Assistance Systems). These applications must coexist with traditional distributed control applications, what generates a great diversity in the traffic traversing the communication network. More specifically, multimedia traffic is characterised by having a large size, low intensity (average occupation of the channel) and soft real-time requirements. Conversely, control traffic usually conveys small amounts of information with high intensity and hard real-time requirements. Furthermore, many modern applications can be launched at any moment of the system’s operation and thus, the network must support the on-line connection and disconnection of participants. Nevertheless Ethernet does not provide adequate timing services to support the transmission of hard or soft real-time traffic. Moreover, Ethernet does not allow to dynamically change the Quality of Service (QoS) of the network and thus it does not support the connection and disconnection of participants at run-time. Several Ethernet-based protocols were proposed to cope with these drawbacks, namely Flexible Time-Triggered (FTT) and more recently Audio Video Bridging (AVB), among others. In this work we study the relevance of the Admission Control (AC) mechanisms of these protocols for the dynamic management of their QoS. Moreover, we carry out a qualitative comparison of the AC mechanisms of the aforementioned protocols. We also present the implementation of the FTT’s AC mechanism over a preliminary simulation model of the HaRTES implementation of FTT that is currently being developed in the University of Banja Luka. Finally, we present a quantitative analysis of the performance of the HaRTES’ AC using said model. The analyses presented in this work constitute a first step towards a future complete comparison of the mentioned protocols.