Image Processing Reference
In-Depth Information
the system uses a dedicated gateway node which routes the configuration and manage-
ment access to the sensor/actuator nodes. hus, the timing of the embedded network is
not disturbed by configuration traffic and the embedded nodes do not need to implement
a USB or CORBA interface and can thus be kept slim. In order to reduce the complex-
ity of the involved conversion and transformation steps, the interface to and from the
embedded node must be comprehensible, structurally simple, and easy to access.
22.4 Interface Separation
If different user groups access a system for different purposes, they should only be provided with an
interface to the information relevant for their respective purpose [].
Interfaces for different purposes may differ by the accessible information and in the temporal
behavior of the access across the interface.
Kopetz, Holzmann, and Elmenreich have identified three interfaces to a transducer node []:
he configuration and planning (CP) interface allows the integration and setup of newly connected
nodes. It is used to generate the “glue” in the network that enables the components of the network to
interact in the intended way. Usually, the CP interface is not time critical.
The diagnostic and management (DM) interface is used for parameterization and calibration
of devices and to collect diagnostic information to support maintenance activities. For example,
a remote maintenance console can request diagnostic information from a certain sensor. The DM
interface is usually not time critical.
The real-time service (RS) interface is used to communicate the application data, e.g., sensor mea-
surements or set values for an actuator. his interface usually has to fulfill timing constraints such as
a bounded latency and a small communication jitter. he RS interface has to be configured by means
of the CP (e.g., communication schedules) or DM interface (e.g., calibration data or level monitors).
The TTP/A fieldbus system [] uses a time-triggered scheduling that provides a deterministic
communication scheme for the RS interface. A specified part of the bandwidth is reserved for arbi-
trary CP and DM activities. Therefore it is possible to perform CP tasks while the system is in
operation without a probe effect on the RS [].
22.4.1 Interface File System Approach
The concept of an interface file system (IFS) was introduced in []. The IFS provides a unique address-
ing scheme to all relevant data of the nodes in a distributed system. hus, the IFS maps real-time data,
all kinds of configuration data, self-describing information, and internal state reports for diagnosis
purposes.
The IFS is organized hierarchically as follows: The cluster name addresses a particular fieldbus
network. Within the cluster, a specific node is addressed by the node name .heIFSofanodeis
structured into iles and records . Each record is a unit of  bytes of data.
The IFS is a generic approach that has been implemented with the TTP/A protocol as a case
study for the OMG smart transducer interface. he IFS approach well supports the integration and
management of heterogeneous fieldbus networks.heIFSprovidesthefollowingbeneits:
It establishes a well-defined interface between network communication and local appli-
cation. he local application uses API (application programming interface) functions to
read and write data from/into the IFS. he communication interface accesses the IFS to
exchange data across the network.
IFS provides transparency on network communication, since a task does not have to
discriminate between data that is locally provided and data communicated via the
network.
Search WWH ::




Custom Search