next up previous contents
Next: Principles of SOAP Web Up: main Previous: Contents   Contents

Introduction

Businesses today consist of many different branches, such as shops, production, management and others which are often situated in various locations. Organizing, coordinating and controlling these different branches efficiently is one of the keys to a successful business. In this context, one important aspect is communication, so that different enterprise entities are able to exchange information. For instance, a shop may report consumer satisfaction to the management, which may then use this information to optimize the production.

Nowadays information is most often transported electronically, therefore computer networks are an appropriate solution for solving the communication aspect. However, for successfully establishing such networks, a suitable technological perspective on the enterprise has to be chosen. From the engineering point of view, enterprises are structured in multiple levels, as depicted in figure 1.

Figure 1: Different levels in an enterprise
\begin{figure}\centering
\includegraphics[scale=0.7]{graphics/intro-levels.eps}\end{figure}

Establishing the seamless integration of these different levels was the main goal of various efforts in industrial automation. The key to success is the availability of information between all levels, as also stated in [WER03]. For instance, the company level should be able to retrieve data from the process level, while the shop floor level may need information from the factory level.

One well known approach is computer integrated manufacturing (CIM), which introduces the automation pyramid, which consists of four to six levels in an enterprise. Every level is associated with a certain functionality and suitable networking technology. These different networks are interconnected to form a complete multilevel networking topology. An in-depth description of CIM can be found in [HAR73].

Due to the introduction of new technologies, especially in the networking domain, the engineering viewpoint has changed over time. Therefore the CIM pyramid can today be seen as a three-level hierarchy, as shown in figure 2.

Figure 2: Three-level hierarchical information structure in industrial automation
\begin{figure}\centering
\includegraphics[scale=0.7]{graphics/intro_mes_erp.eps}\end{figure}

At the top of the hierarchy is the Enterprise Resource Planning (ERP), which deals with strategic and long-term planning. Below is the Manufacturing Execution System (MES), which is concerned with short-time production planning. At the bottom of the hierarchy is the control level, which accomplishes all low level tasks of the manufacturing process, such as controlling sensors and actuators. This topic is further examined by [SAUT05].

For a comprehensive integration of these three hierarchical levels, information has to be exchanged between levels, which is the goal of vertical integration. To achieve this, two separate aspects have to be addressed, as denoted in [SAUT05]:

1. Communication:
Different levels in an enterprise often use different networking technologies. Interconnecting these different networks is a major step in vertical integration.

2. Application:
In an enterprise, the same information may be represented in different ways. The information in ERP, for example, focuses on the business aspect of an enterprise, while the control level has to deal with technical details, such as sensor/actuator data. For this reason, the underlying format in which such information is stored is optimized for the specific task in the enterprise level and will be syntactically and semantically different from the format in the control level. For vertical integration, these different formats have to be translated or generalized so that applications in different levels can communicate which each other.

On the ERP and MES levels, Internet networking technologies, such as the Internet Protocol (IP) are most commonly used today. Therefore the communication aspect does not constitute a problem. Today the application aspects on this level are already solved to some degree, although various issues are still in research status. Further information about this topic can be found in [WER03].

On the other hand, there is still a huge gap between MES and the control level. As the functionality and requirements for the underlying technologies vary between these levels, the networks used are different. As shown above, the MES level uses common Internet technologies for networking, which is, however, unsuitable for the control level due to lack of real-time abilities and other reasons.

In the last decades, fieldbuses - often also called ``Field Area Networks'' (FAN) - were established as the most common communication technology in the control level. Fieldbuses are industrial communication systems and are designed to interconnect a variety of field devices, such as sensors and actuators. They are designed to operate in rough environments, as often found in manufacturing areas. They are capable of handling data in real time, meaning that the transmission time of data can be predicted and constrained. Moreover, fieldbus equipment is designed to be very cost effective, enabling the technology to be ubiquitous in the manufacturing environment.

Although many different fieldbus technologies exist, a common way of looking at them is illustrated in figure 3: various devices, called fieldbus nodes are interconnected by a fieldbus network. Data can be represented by so-called ``data points'', which are assigned to a fieldbus node and contain a piece of information, such as sensor or actuator data. These data points are associated with underlying devices, such as actuators/sensors and can be read and written, hence establishing control of these devices.

Figure 3: Fieldbus with Nodes and Data Points
\begin{figure}\centering
\includegraphics[scale=0.7]{graphics/intro_fieldbus.eps}\end{figure}

There is a variety of different fieldbus systems available, each with its specific strengths and weaknesses. Different fieldbus systems are most often incompatible, therefore interconnecting systems based on such different technologies is often a tedious task.

One aspect that is important for the success of vertical integration is the connectivity between fieldbuses and IP networks1.

As described by [LOB06], interconnecting these two networks may be done in two ways:

Tunneling:
A device that encapsulates one networking protocol into the other is set up between two different networks, so that data can be transported through another network. The receiving side extracts the encapsulated data and may handle it as if it was connected to the sending network. This mechanism is called tunneling and is transparent, because the sender and receiver may communicate through this tunnel as if they were connected to the same network.

The advantage of this concept lies in its simplicity: the communication aspect can be solved quite easily as encapsulating one protocol into another is not very complicated. However, certain network functionality, such as predefined delays and real-time abilities, may be lost due to the transport over the other network.

The application aspect, on the other hand, is not addressed at all, as the receiving side is able to handle the original data. This way either the fieldbus has to implement Internet specific protocols or the application on the ERP/MES side has to handle the fieldbus specific protocol. This is also the disadvantage of this concept, as the end point has to implement the networking protocol of the sender. In vertical integration, implementing lower level protocols in higher levels is a bad solution as the application aspect is still unsolved and is merely shifted to another level.

Therefore tunneling solutions are not very widespread and are most often be used where applications need in-depth access to the foreign network. For instance, specific functionality such as fieldbus configuration from upper enterprise levels may be solved by tunneling.

Gateways:
Gateways are nowadays the most common technique of interconnecting fieldbus and IP-based networks. Figure 4 shows where such an Internet/fieldbus gateway will be placed.

Figure 4: Internet/Fieldbus Gateway
\begin{figure}\centering
\includegraphics[scale=0.7]{graphics/fb_gw1.eps}\end{figure}

As mentioned above, interconnecting networks consist of communication and application aspects. The communication aspect is relatively easy to manage: the gateway should look like an IP device from the Internet side, while it should be able to integrate into the fieldbus as any other FAN node. Therefore the gateway will have to implement both networking protocols. Gateways are often implemented on embedded hardware, on which both networking protocols can be implemented without much effort.

The application aspect is more complex. Fieldbus specific data has to be presented in a way so that upper-level applications can understand and handle it. The situation is complicated by the fact that various different fieldbus technologies exist with varying functionality, providing data in different ways. Upper-level applications should most often not deal with fieldbus specifics, therefore a way has to be found to represent fieldbus data in a general, abstract way.

A common solution is to specify an Internet interface that defines the semantics and syntax of fieldbus access. Such an interface tries to generalize functionality and data representation of fieldbus systems so that the interface can be used for a variety of fieldbus technologies. Moreover, this interface has to specify a protocol which can be used to transport fieldbus data over IP networks. As there are already various Internet protocols available, it is appealing to use an existing one that suits the needs of such a fieldbus/Internet gateway interface.

The traditional approach in distributing information across the Internet is to store data on servers, which can then be retrieved by clients, which format and present this data to the user. One well known example is the Hypertext Markup Language (HTML) and its associated protocol, the Hypertext Transfer Protocol (HTTP), which is certainly the most common technique of transferring information across the Internet. Although HTML and associated technologies are well suited to displaying information to users, it is inappropriate for communication between software applications.

Internet connectivity will soon be available for all sorts of devices such as household appliances, multimedia systems, cars and also equipment and devices in the industry, making the Internet ubiquitous in the near future. Internet applications will be placed on these devices that have to exchange information which they have to process and understand. For instance, a heating system may periodically retrieve weather information from a centralized weather station over the Internet to optimize its operation. Another example would be a smart refrigerator that may automatically order food from a store when it gets empty.

To establish successful machine to machine communication, suitable communication formats and protocols are needed. The key to success seem to be the following points:

Interoperability:
As denoted above, different applications should be able to interoperate. Standardization is an important aspect in establishing interoperability, therefore formats and protocols will have to be thoroughly specified, so that different developers can build interoperable applications.

Discoverability:
To establish successful communication between applications, developers have to understand the meaning of the data being sent. Therefore communication formats should have self-describing abilities so that developers can easily implement applications on top of these communication formats.

Easy to implement:
Ease of use and rapid application development is a key feature of success. The communication format should be designed in a way so that software frameworks can be used to automatically generate functionality, which can then be easily integrated into the custom application.

Historically, there have been various technologies and protocols that were designed to fulfill these needs. However, none of them proved as a universal solution.

Today, the common opinion is that a new technology, called ``Web Services'', will be the solution to the requirements above. Web services are designed to provide a high level of interoperability while covering a variety of communication aspects. They use the so-called ``Extended Markup Language'' (XML) as their key technology, which is nowadays seen as the most promising communication format, enabling developers to create self-describing documents. Moreover, Web services have simplicity in mind and are therefore allowing easy access of underlying services.

Figure 5 illustrates the difference between traditional Internet communication, where a client application retrieves and displays information, and the Web service approach which consists of various ubiquitous, interoperating applications.

Figure 5: Classical Internet Communication Versus Web Services
\begin{figure}\centering
\includegraphics[scale=0.7]{graphics/intro-webservice.eps}\end{figure}

Regarding the Internet interface of an Internet/fieldbus gateway, it can be observed that the requirements are very similar to the needs of interoperating Internet applications. Interoperability, discoverability and simplicity are key aspects of a successful communication between fieldbus and IP networks, too.

Therefore Web services are nowadays seen as a suitable technology for an Internet interface on Internet/fieldbus gateways. In fact, there are already various standards and solutions available which are based on this technology.

This thesis examines in which ways Web services may be applied to Internet/fieldbus gateways, how well they interoperate and perform and if they are really a solution to limitations of traditional technologies.


next up previous contents
Next: Principles of SOAP Web Up: main Previous: Contents   Contents
Hermann Himmelbauer 2006-09-27