مدل سازی سامانه های بی درنگ توزیع شده با MAST2

Switched networks have an increasingly important role in real-time communications. The IEEE Ethernet standards have defined prioritized traffic (802.1p) and other QoS mechanisms (802.1q). The Avionics Full-Duplex Switched Ethernet (AFDX) standard defines a hard real-time network based on switched Ethernet. Clock synchronization is also an important service in some real-time distributed systems because it allows a global notion of time for event timing and timing requirements. In the process of defining the new MAST 2 model, clock synchronization modeling capabilities have been added, and the network elements have been enhanced to include switches and routers. This paper introduces the schedulability model that will enable an automatic schedulability analysis of a distributed application using switched networks and clock synchronization mechanisms.

MAST (Modeling and Analysis Suite for Real-Time Applications) [7] and [14] defines a model to describe the timing behavior of real-time systems designed to be analyzable via schedulability analysis techniques. MAST also provides an open-source set of tools to perform schedulability analysis or other timing analysis, with the goal of assessing whether the system will be able to meet its timing requirements, and, via sensitivity analysis, how far or close is the system from meeting its timing requirements. The model defined in MAST is very similar to the model defined in the Schedulability Analysis Modeling chapter (SAM) of the MARTE profile (The UML Profile for Modeling and Analysis of Real-Time and Embedded Systems) [16]. A new enhanced model is currently being defined as a project called MAST 2, trying to incorporate new modeling elements that can be found in real systems. It is expected that the ideas introduced in MAST 2 will contribute to the future evolution of the MARTE standard. Some of the new elements being defined in MAST 2 are network switches and routers. Switched networks are being used increasingly to build real-time systems, as new network switches incorporate the real-time mechanisms being defined in standards such as IEEE 802.1p with prioritized traffic [9], 802.1q with various QoS mechanisms [23], or the Avionics Full-Duplex Switched Ethernet (AFDX) [1] that defines a hard real-time network based on switched Ethernet. This paper introduces the model elements required to add network switches and routers into the MAST model. These elements will allow an automatic schedulability analysis of applications using switched networks. Another addition to MAST 2 is the capability to model and analyze time-triggered systems, such as those developed according to the ARINC 653 standard [3]. In these systems the scheduler uses a table to drive the generation of the events that trigger the execution of operations in the processing nodes, and the transmission of messages across the networks. In many cases these time-triggered systems require the capability of having a global notion of time, and thus require clock synchronization services. In addition, another kind of scheduling that requires clock synchronization is global EDF [22], which has been shown to obtain better schedulability results than the non-synchronized local EDF [18]. Kopetz has a very interesting introduction to clock synchronization in [11] that describes basic concepts about global time. It provides definitions of the digital physical clock, the granularity, the reference clock, the clock drift, and the precision or the accuracy of the global time base. MAST 2 defines a means to model the clock synchronization service and to analyze its effects. The paper is organized as follows. Section 2 contains a brief summary of related work. In Section 3 we present a general overview of the MAST 2 model, and we focus on the network modeling elements in Section 4. The model of clock synchronization services is described in Section 5. The new elements introduced to model network switches are presented in Section 6, and similarly in Section 7 for network routers. Section 8 introduces the new modeling elements for AFDX networks and switches together with a simple example using these elements. Finally, Section 9 gives our conclusions.

In this paper we have shown how to model distributed real-time systems with MAST 2. We have also proposed new modeling elements to support network switches and routers and AFDX real-time networks in the timing models used to assess the schedulability of real-time distributed applications. We have also added support for modeling clock synchronization, which allows us to use timetable driven or global EDF scheduling policies in distributed systems. The new modeling elements, together with their associated analysis techniques are being implemented, and will be proposed for a future version of the MARTE UML profile for real-time embedded systems. With the MAST 2 model it is possible to describe the timing behavior of complex distributed systems, with the aim of performing an automatic schedulability analysis. It is also possible to perform simulations of the modeled system to obtain average-case performance results.