|Using radio communication in automation networks
Reliability and stability by WLAN Meshing
Dipl.-Ing., MBA Olaf Schilperoort, Product Manager at the Hirschmann Automation and Control GmbH, Neckartenzlingen, Tel.: +49 (0)7127-14-1640, E-mail: Olaf.Schilperoort@hirschmann.de
Following its conquest of the office environment, WLAN (Wireless Local Area Network) is now also being used increasingly in industry. Because its reliability and stability here could be considerably increased by new or extended protocols like WLAN Meshing. The applications include driverless transport systems, for example, which can be linked up to the data network without complicated wiring using this transmission technology. A production plant can also be converted faster, more flexibly and at a lower cost with WLAN. Some individual actuators or measuring instruments are already equipped with Ethernet or TCP/IP connections even.
However, the demands on LAN in industry differ greatly from those in the office environment. Instead of a wide bandwidth – for fast Internet or database access for example – the reliability and stability of the data transmission are in the foreground. Because current Industrial Ethernet protocols and applications are very sensitive to interruption of the connections. For example, a wirelessly controlled system cannot be allowed to repeatedly come to a standstill because the radio connection is disturbed and therefore the trouble-free operation cannot be guaranteed.
However, the industrial environment frequently harbors unfavorable propagation conditions for electromagnetic waves. Because metals parts which can disturb or interrupt the transmission line can be found almost everywhere. Or to be more exact: These parts reflect the waves which leads to packet losses or even destruction of the signal. A high-shelf warehouse may be filled with paper rolls, for example, when the WLAN is installed. If containers filled with liquids are stored later, the radio conditions may change to such an extent that sufficient WLAN coverage is no longer guaranteed so that the receiver of the radio signal is difficult to reach. An even greater problem arises when an interference source suddenly moves into the transmission line and interrupts it.
Redundant WLAN networks
However, an efficient antenna technology is now available with which reflections and interferences can largely be prevented. WLAN networks can also be set up redundantly for stabilization. Many of today's industrial access points, for example, have two independent radio interfaces which can operate in the 2.4 and 5 GHz range. Careful planning of the network with the aim of as good as possible a coverage of the area by WLAN can reduced the number of potential problems in advance. In addition, the administrators of WLAN networks are provided with a special management and monitoring software to detect and eliminate problems at an early stage. These are also the reasons why WLAN is becoming more and more accepted in industry today. And the technological development continues to progress rapidly
How the stability of wireless networks can be improved has always been an important consideration. Because pure star or ring structures never offered full protection against interference. The demand to make WLAN networks more flexible and at the same time more stable came from military channels: in the mid 1990s the technology of the meshed network for communication on the battlefield was introduced as a research project of the DARPA (Defense Advanced Research Projects Agency, United States Department of Defense). The aim was to have several connections to every network user. The nodes in the individual meshes also had to be able to communicate with different users simultaneously or transmit data. The efficiency of the network increases with the number of nodes. At the same time, the addition of several existing connections between two points in the network enables the bandwidth to be increased. If this property is extended to wireless networks, new nodes can also be integrated flexibly and quickly in the network. And nodes which have failed can be bypassed quickly. The individual devices therefore no longer require their own cable connection to the network as is the case at the moment with the conventional WLAN networks.
A meshed network is therefore represented as an – at best random – arrangement of access points and clients which are able to set up and manage a network without additional wiring. All users may be mobile. And both indoor and outdoor components are supported. A dynamic routing recognizes a faulty connection immediately and the components are switched to another connection. This so-called "self-healing" function in conjunction with a protocol which searches for the "best path" makes great demands on the software and hardware of the components, however, which could be increased depending on the area of application of the wireless network.
Demands on meshed networks
The meshing technology which is already an integral part of wireless sensor networks provides the answer to these demands. However, no constant, widebanded data transmission is demanded in such networks. Meshing is used primarily to increase the reliability of the data transmission of measuring instruments which have to work for years under difficult ambient conditions for example. The energy consumption should be as low as possible. Packet runtimes or real time capability are therefore not a problem. It suffices when the measured data reach their receiver within a very reasonably allotted time.
On the other hand, WLAN as the "Ethernet of the air" has the task of guaranteeing a wideband data transmission constantly and free of interruption. The switching times for the recalculation of a connection may therefore not be longer than is already the case today for Fast Roaming – the transfer of a client between two access points. Applications such as voice transmissions, for example, tolerate interruption times of a maximum 50ms. After all, the packet runtime before and after switching to another connection may not change significantly.
However, a particular problem of using WLAN is that the security requirements are already very high because every connection needs to be coded. This is done by the IEEE 802.11 standard with an AES-CCM coding. Although this method offers security on the level of the best VPN tunnel, a key exchange must also take place for every new connection. Even more complicated is the coding when using an 802.1x authentication, because this requires an inquiry with the 802.1x every time. But such a server may not be available within the meshed network.
All these demands which present a challenge for conventional access points are, however, counteracted by advantages which make the use of meshing technology in industrial networks very interesting. Because the prime goal, namely the increased availability of the WLAN network, can already be achieved by using fewer nodes. As a first step, it is enough to offer every client just one single alternative path. At the same time, the meshed network scales with the number of nodes by the efficiency being increased directly by the added access points so that the reliability of the data transmission also increases. This can already be achieved by using static, WDS-capable access points with a spanning tree protocol. – a Wireless Distribution System (WDS) represents a combination of point-to-(multi-)-point and client connections. However, this structure is only statically possible. The switching times of spanning tree are also too high and addition of the bandwidths is not achieved.
The great advantage of the meshed network is in its ability to organize itself so to speak. Because the user has to do no more than move a handful of access points close to each other and these automatically set up a network. This may mean a considerable reduction in costs especially when setting up widely distributed WLAN network because not every device can be configured individually on site. And the avoidance of cable connections comes into play as a third important aspect. Because, unlike WLAN switching systems which may increase the wiring effort, meshed networks represent the "true" wireless networks – apart from the connection to a power supply.
New IEEE standard for meshing
Although solutions and applications with meshed wireless networks have already existed for about 15 years, an IEEE committee is only now working to integrate this functionality in one of the 802.11 protocol extensions. Initial suggestions and drafts of the standard known as 802.11s are already available. Here, an attempt is being made to unite the existing solutions of several WLAN manufacturers. Because, as with every standard, the aim here is to guarantee the interoperability of products of different manufacturers within a network. And this, in turn, will only become possible with the acceptance of WLAN Meshing in the industrial environment. Because the trouble-free interaction of access points and clients with different functions is decisive to be able to set up an efficient WLAN network. An appropriate standard will probably be passed in 2008.
However, it is already clear that the access points then used will have to meet certain requirements which they usually do not have at the moment. Since the software will not be able to do without routing functions in future, the hardware platform must include a greater storage capacity and more efficient CPUs accordingly. In addition, at least two WLAN interfaces are required, three would be even better. Two interfaces serve to maintain the wideband connection within the network between two access points and a third interface to link a client. The operating system used on the access point must support both the interfaces and the routing protocol. Mere bridging between Ethernet and WLAN, as is the basic function of an access point, is not enough. Devices which already have their own operating system and at least two WLAN interfaces as well as at least one 100Mbit/s LAN interface therefore offer advantages. Because the bandwidth available in the WLAN meshing must also be transmittable at the network transitions to the wired LAN.
The question posed by a meshed network is also how you can detect what is happening in the wireless network. Especially when it is constantly being reorganized by addition or removal of clients and access points. How can you recognize whether a device is still available? How can the available performance be determined? How can bottlenecks or security problems be detected? How can the configuration and the respective version of the firmware be kept up to date?
Professional WLAN management
The classic answer to all these questions is the use of a management system designed specially for the requirements of a WLAN. Such efficient systems already exist. Monitoring of such a dynamic network as a meshed WLAN poses special challenges however. Therefore, the individual nodes which are implemented in the form of access points must report the existence of other devices in the network automatically and communicate their status. Only then is the administrator able to recognize the status of the network quickly and easily at all times.
Another challenge are devices in classic WLAN networks which transmit on the same channels, for example, and disturb the network or try to 'rob' clients, i.e. get them to associate with foreign access points and spy out the transmitted data or log into the network illegally. This so-called rogue access point problem can only ultimately be solved in meshed networks, which can be structured much more openly on account of their functional principle, on the one hand with strict authentication methods and on the other hand with extended rogue AP detection systems. Since external servers should not be accessed for authentication, the mesh access points must have at least parts of these functions; this includes, for example, RADIUS server and 802.1x/EAP mechanisms.
To avoid losing track and increasing the efficiency of hardware and software unnecessarily, a useful solution is to limit the size of the meshed networks to a maximum number of nodes. Typical are networks with up to 32 access points. These are connected with each other by conventional point-to-point lines. There are gateways with interfaces to far-range, copper or glass fiber networks at the edges of the network. Direct connection to sensor networks is also a useful solution for industry.
The demands on the individual components of the network increase with the aim of improving the availability and stability of WLAN networks for industrial applications. Whilst the classic "fat access points" were still recently being dismissed as dinosaurs from a long forgotten age, the fact is that even more distributed functions are being demanded in the individual access points.
Meshing is already being used successfully today in wireless networks. A new IEEE standard is also being prepared for WLAN which will both increase the acceptance of meshed networks and speed up the implementation of such networks. So the corner stone for setting up networks quickly and easily as a stable and reliable platform for industrial applications has already been laid. Two figures show that it is worth it. On the one hand, meshed networks will increase the availability to up to 95% even under difficult ambient conditions. And on the other hand, the hardware costs will fall by 25% to 90% due to the omission of wiring and networks components such as switches.