Image Processing Reference
In-Depth Information
In contrast to configuration, the term
management
dealswithhandlingofanalreadyestablished
system, and includes maintenance, diagnosis, monitoring, and debugging. As with configuration,
different systems can greatly differ in their support and capabilities for these areas.
Often configuration and management are difficult to separate since procedures such as plug and
play (see Section ..) involve configuration as well as management tasks.
22.2.2 Smart Devices
he term
smart
or
intelligent
device was first used in this context by Ko and Fung in [], meaning
a sensor or actuator device that is equipped with a network interface in order to support an easy
integration into a distributed control application.
In our context, an
intelligent
device supports its configuration and management by providing its
data via a well-defined network interface [] and/or offering a self-description of its features. The
description usually comes in a machine-readable form (e.g., as XML description) that resides either
locally at the embedded device (e.g., IEEE. []) or at a higher network level being referenced by
a series number (e.g., OMG [Object Management Group] smart transducer interface []).
22.2.3 Plug and Play versus Plug and Participate
Plug and play
describes a feature for the automatic integration of a newly connected device into a
system without user intervention. While this feature works well for personal computers within an
office environment, it is quite difficult to achieve this behavior for automation systems, since, with-
out user intervention, the system would not be able to guess, what sensor data should be used and
what actuator should be instrumented by a given device. herefore, in the automation domain, the
more correct term
plug and participate
should be used that describes the initial configuration and
integration of a new device that can be automated. For example, after connecting a new sensor to a
network, it could be automatically detected, given a local name, and assigned to a communication
slot. he task of a human system integrator is then reduced to decide on the further processing and
usage of the sensor data.
22.2.4 State
Zadeh states that the “notion of state of a system at any given time is the information needed to
determine the behavior of the system from that time on” [, p. ]. In real-time computer systems we
distinguish between the initialization state (i-state) and the history state (h-state) [].
The i-state encompasses the static data structure of the computer system, i.e., data that is usually
located in the static (read-only) memory of the system. he i-state does not change during the exe-
cution of a given application, e.g., calibration data of a fieldbus node. The h-state is the “dynamic
data structure
that undergoes change as the computation progresses” [, p. ]. Examples of an
h-state are the cached results of a sequence of measurements that are used to calculate the current
state of a process variable.
hesizeoftheh-stateatagivenlevelofabstractionmayvaryduringexecution.Agoodsystem
design will aim at having a
ground state
, i.e., when the size of the h-state becomes zero. In a distributed
system, this usually requires that no task is active and no messages are in transit.
[
...
]
22.3 Requirements on Configuration and Management
The requirements imposed on a configuration and management framework are driven by several
factors. We have identified the following points:
