Image Processing Reference
In-Depth Information
networked embedded systems where communication systems were optimized in some respect and
do not follow the mainstream networking technologies.
The question of how to interconnect different networks within the automation hierarchy is pri-
marily an architectural choice and does not depend on the two (necessarily different) protocols used
on both sides of the interconnection point. Against the backdrop of the ubiquity of the Internet tech-
nologies, however, the discussion and the development efforts today focus on the interconnection of
fieldbus systems and IP-based networks in whatever form. From an architectural point of view, there
are two main possibilities to achieve a fieldbus/Internet interconnection, both of which are being
used in practice:
Tunneling or encapsulation of one protocol in the other, either in a strictly vertical sense
or to achieve a connection between remote fieldbus segments via a higher-level backbone
network
Providing protocol, service, and data translation via a gateway
hetopologicalviewofbothapproachesisidentical.Inbothbases,thereisacentralnodelinking
the two networks. Usually, the term “gateway” is used for such an interconnection node. In the fol-
lowing, we will prefer the more neutral term “access point” (not to be confused with the access points
used in wireless networks) to avoid misconceptions regarding the functionality of the device, which
differs significantly between gateway and tunneling approach.
20.6.1 Protocol Tunneling
Tunneling in connection with communication networks essentially means that data frames of one
protocol are wrapped into the payload data of another protocol without any modification or transla-
tion. hus, two possibilities exist: the fieldbus protocol can be encapsulated in the Internet Protocol
on the higher level, or a protocol of from the IP suite can be passed over the fieldbus. Tunneling is
well-known in the office world as a means to set up virtual private networks for secure communica-
tion of, e.g., remote offices. Although tunnels in this area normally operated on the data link layer
(like the Layer  Tunneling Protocol LTP), in fact any PDU of any protocol layer can be tunneled
over any other; there is no compelling reason to restrict tunneling to the data link layer.
20.6.1.1 IP Tunneling over Fieldbus Protocols
In the automation area, the currently more common tunneling approach is to encapsulate IP packets
in fieldbus messages. his is a direct consequence of the widespread use of Web-based interfaces for
engineering tools and opens a channel for the upper layer protocols required for, e.g., direct Web
access to the field devices []. While in most cases not foreseen in the beginning, this possibility has
recently been included in several fieldbus systems [,]. At first glance, IP tunneling provides an
easy way to achieve integration, as Figure . shows. he ield devices run an IP-based service such
as a Web server providing data for a corresponding application in the Internet, which can directly
access the device or process data. While the typically limited computing resources at the field device
used to be a counter-argument for this solution for a long time, the availability of embedded con-
trollers with lightweight (maybe even on-chip) TCP/IP stack implementations [] and Web servers
forsmartdevicessolvestheproblem[-].Asthese devices usually also have integrated Ethernet
controllers, the question arises why additional fieldbus interfaces should be added instead of using
Ethernet directly on the field level.
Another essential design consideration for the access point is the way the traffic is being handled.
This has two facets: addressing and scheduling. he addressing problem becomes obvious if we con-
sider that the “tunnel” is not just a single communication path connecting two nodes, but has several
exits. Therefore, the access point must translate between the IP address used to identify the field
 
Search WWH ::




Custom Search