Image Processing Reference
In-Depth Information
during replacement. A detailed presentation of redundancy and fault-tolerant systems
can be found in [].
•
Shutdown part of the application
:Incaseoffactorycommunicationsystemsthatotenare
organized as a multilevel network and/or use a modular approach, it might be feasible to
shut down a local subnetwork (e.g., a local control loop, or a process cell as defined in the
ANSI/ISA-.- standard).
The replacement node must be configured with individual node data, such as calibration data (this
data usually differs between replaced and replacement node), and the
state
of a node. The state
information can include
•
Information that is accumulated at run-time (the history state of a system): This infor-
mation must be transferred from the replaced to the replacement node.
•
Timing information, so that the node can synchronize with the network: For example, in
networks that use a distributed static schedule (e.g., TTP/A), each node must be con-
figured with its part of the global schedule in order to get a network-wide consistent
communication configuration.
One alternative approach for avoiding this transferring of system state is to design a stateless system
in the first place. Bauer proposes a generic approach for creating stateless systems from systems with
state in []. Another possibility is to provide well-defined reintegration points, where this state is
minimized. Since fieldbus applications typically use a cyclical communication style, the start of a
cycle is a “natural” reintegration point.
22.10 Conclusion
Configuration and management play an important role for distributed embedded systems. he need
for configuration and management in the domain of industrial fieldbus systems has led to interesting
mechanisms which evolved during the last years. Most of these mechanisms can also be applied
inthedomainofembeddedsystems.
The configuration phase can be subdivided into a part that requires local interaction such as con-
nection of hardware and setting dip switches and a part that can be done remotely via the fieldbus
system. A good design requires that the local part is as simple as possible in order to simplify the inter-
actions done by local personnel. he other part should be supported by tools that assist the system
integrator in tedious and error-prone tasks such as adjusting parameters according to the datasheet
of a device. Examples for systems with such a configuration support are, among others, the IEEE ,
the FDT, and the EDDL that all employ machine-readable electronic datasheets in order to support
configuration tools.
Management encompasses functions like monitoring, diagnosis, calibration, and support for
maintenance. In contrast to the configuration phase, most management functions are used concur-
rently to the RS during operation. Some management functions, such as monitoring, may require
real-time behavior for themselves. In order to avoid a probe effect on the RS, the scheduling of the
fieldbus system must be designed to integrate the management traffic with the real-time traffic.
Acknowledgment
This work was supported by the Austrian FWF project TTCAR under contract No. P-N.
