[Held in conjunction with the 31st IEEE Rel-time Systems Symposium (RTSS '10).]

Cyber Physical Systems (CPS) are software-based systems that control and interact with physical processes. Such systems play a key role in an ever increasing number of industries. For example, over 80% of the innovations in automotive systems are for cyber-enabled system capabilities. Modern aircraft are carefully co-designed and tightly integrated CPS machines. This trend is also demonstrated by modern medical systems. The movement towards increasing dependency on software is driven by the fact that software enables the delivery of a large number of customized capabilities in a
product using a relatively small number of physical and computing platforms.

However, the co-design of physical platforms, computer systems, and embedded software systems and their tight integration also creates a high degree of complexity of
interactions within and across these systems that far exceeds the capability of existing system composition technologies. These technologies are of paramount
importance to industries that integrate independently-developed parts into their final products (e.g. automotive and avionics). Furthermore, even though analytical models of such systems are currently used to predict different system-level properties they are often developed on a property-by-property basis by different teams leading to inconsistent assumptions and conclusions. The lack of effective system-wide analytical results prevents the discovery of design flaws that stem from the interaction
complexity of the system parts; i.e. prevented until the system is physically integrated. Due to such flaws, the system integration time often exceeds 50% of the total development time for non-safety critical applications. In safety critical systems, such as avionics, the system integration and certification time often exceeds 70% of total development time and costs.

Looking ahead, the success of next generation CPS systems demands system-wide architecture design patterns and supporting technologies that can integrate legacy
components, COTS components and co-designed new components in such a way that properties such as real time, safety, fault tolerance and security can be analyzed
and predicted before the systems are physically built. Moreover, it is necessary to have a system-wide composition model that integrates the different analyses into a single consistent semantic framework to avoid conflicting results.

This workshop focuses on analytical system composition technologies that will eventually include:

(1) Composition technologies to automatically propagate the impact of modifications in one modeling domain into others.
(2) Assumption resolution between modeling abstractions and constructs of different analysis domains.
(3) System-level schedulability optimization technologies that integrate constraints imposed by other analytic domains (E.g: security, mechanical stress, heat dissipation, etc.)
(4) A quantitative and early analysis of the system architecture performance in an end-to-end fashion, deriving perhaps even the worst/best/average case behaviour for the
entire platform and new hardware abstractions. In fact, existing task/system models reason at levels that are well abstracted away from real details (and variations inherent in) hardware components (e.g. multicores, memory architectures, I/O, network-on-chip, etc.). They also are unable to cope with workloads that are beyond the
capabilities of traditional computational resources (e.g. video streams, weather data, GPS, etc.)
(5) Fault tolerance technologies and reliability analysis techniques that integrate the different natures of physical, hardware and software faults in a common, consistent framework.
(6) Safety analysis such as model checking for mixed criticality CPS applications, for example, flight management systems and/or safe medical devices plug and play (MDPnP).
(7) Security protocol development and verification techniques for CPS applications.
(8) Models for describing/quantifying the environments where
such systems must operate.

The goal of this workshop is to explore architecture design patterns, tools and the theoretical analytical foundations for creating common system-wide composition models where key properties can be studied and guarantees provided before the start of actual development. Of particular interest are the case studies on the challenges of expressing the properties of the final product in terms of component properties and
the architecture that governs their interactions. Both solutions and/or open problems are welcome.