The majority of current supervision and control applications use distributed embedded systems in which several intelligent nodes communicate over a digital network to accomplish a global goal.

A set of these applications, e.g. process control, vehicle control, machine-tools and robots, may exhibit strict timeliness, predictability and precedence constraints. In these cases, special-purpose real-time communication networks, known as fieldbuses, must be used to achieve the desired properties.

Recently, several Ethernet-based protocols have been developed for those systems (e.g. ETHERNET-Powerlink , PROFINET, EtherCAT and Ethernet/IP), which take advantage of some of Ethernet’s appealing attributes (e.g. large bandwidth, cheap silicon and high availability), while removing or decreasing the non-deterministic issues arising from its MAC protocol and/or from the current switched architecture.

However, the timeliness guarantees provided by those protocols are essentially static, based on pre-run-time analysis. On-line admission control is not generally available and neither is on-line adaptation of the communication requirements according to effective needs or to a quality-of-service (QoS) adaptation policy.

This contrasts with the current trend toward more dynamic and adaptive embedded systems that are more resource efficient (e.g. by switching on and off subsystems that operate sporadically, executing mode changes according to environment stimuli, adjusting operational parameters to the current environment dynamics, and distributing the system resources available at each instant to maximize the QoS delivered to the current system users).