23.03.2017 Views

wilamowski-b-m-irwin-j-d-industrial-communication-systems-2011

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

13-6 Industrial Communication Systems<br />

themselves and process the encapsulated traffic as a node, not as a router. This effectively enables purely<br />

IP-based fieldbus nodes, which can communicate directly to legacy (non IP-based) nodes in the system.<br />

The tunneling approach has been chosen for LonWorks or KNX/IP.<br />

Other fieldbus <strong>systems</strong> implement a dedicated IP data link layer for the fieldbus protocol natively.<br />

Nodes in these <strong>systems</strong> can be connected directly onto the Ethernet network without the use of a tunneling<br />

technique. Examples are BACnet/IP or Modbus over TCP. To access other, non IP-based CN nodes,<br />

routers are used.<br />

Notwithstanding the ease of IP everywhere, a separation of field-level and office-level networks is still<br />

recommended for performance and security reasons, even if—as in the case of <strong>industrial</strong> Ethernet—a<br />

direct integration of both networks is possible. Another issue in using IP-based transport <strong>systems</strong> for<br />

fieldbus protocols is that the original, native <strong>communication</strong> media had partly different properties.<br />

For example, delay times, delay variance, and packet ordering may impair the fieldbus protocol state<br />

machine itself. These effects are typically combated by using extra sequence numbers and synchronized<br />

time-stamping on the tunnel. Once the protocol state machines are maintained, the application on<br />

top is still influenced in its design. Application timers or feedback control loops must be aware of the<br />

changed timing relations.<br />

Once the problem of merging isolated network islands into a flat network hierarchy is solved, logical<br />

boundaries within the same networking technology still exist. These are typically imposed by different<br />

application domains to keep the traffic, addressing, and installation tasks separated. While <strong>communication</strong><br />

would be enabled over the flat network, the domain boundary bars traffic. The use of proxy devices<br />

solves this problem. Proxies are basically devices that pass the process data and expose it from one<br />

domain into another. An implementation of a proxy is placed as one instance in each domain. Strictly<br />

speaking, a proxy can be seen as a gateway between the same technologies. An example for using proxies<br />

is LonWorks in automation <strong>systems</strong>. In this system, address spaces are separated by address domains.<br />

The address domains are set up according to different application domains, such as HVAC or emergency<br />

lighting. A LonWorks proxy can expose data points of one domain to another domain. Other examples<br />

of proxies are HTTP proxies. These are frequently used to isolate local from public IP traffic for security<br />

reasons. For instance, the office-level network needs access to Web services in the field-level network.<br />

While it is common practice to keep those two IP networks separate, an HTTP proxy can establish the<br />

Web service link.<br />

Another area of focus is protocol convergence. Today’s automation <strong>systems</strong> represent highly heterogeneous<br />

<strong>systems</strong>. Different application service domains may use different CN technologies. The goal<br />

for vertical access or distribution of <strong>communication</strong> services is clearly a seamless <strong>communication</strong> also<br />

between the application domains and across all layers of the automation system. The classical approach<br />

to let system A communicate with system B is the use of gateways. This concept is being adopted widely.<br />

Gateways abstract the network service at the application level (e.g., data point values, trend data, schedules,<br />

and alarms) and represent it in different networking technologies. Commonly used gateway concepts<br />

today are based on monolithic gateways. These are devices that have the global view of one system<br />

and expose it to another system. This concept, however, presents an inherent problem: The complexity<br />

to integrate an arbitrary number of <strong>systems</strong> is of quadratic order. Another problem is that the gateway<br />

is a single point of failure (SPOF) in the system.<br />

In the move to a more service-oriented model, the <strong>communication</strong> services such as data points,<br />

scheduling, trending, or alarming can be abstracted and presented in a single interface, which is technology<br />

independent [DPAL]. Applications developed on such devices use the abstracted interface and<br />

are essentially made independent from the underlying networking system. It is important to note that<br />

this approach is only feasible for actual data <strong>communication</strong>, not the configuration of such devices,<br />

which still remains technology specific.<br />

As computing power is readily available, devices can add another network protocol stack to provide<br />

native access to certain services to the system of another technology. Figure 13.4 presents an example<br />

for this scenario. A controller implements not only the LonWorks technology but also the BACnet<br />

© <strong>2011</strong> by Taylor and Francis Group, LLC

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!