Let me get this straight. You want to replace a long twisted-pair cable with a wireless link? Are you nuts?
That's what I said to my brother several years ago when he wanted my input on his project for a new control-system design in a Texas process control plant. He needed to install eight resistive temperature devices (RTDs) to measure the temperature on pipes to control their temperature with heat trace, a type of wire heating element.
The project was pretty straightforward, but the runs of cable from the RTDs to the controllers were very long, from a couple hundred feet up to a thousand feet. And installation of even simple twisted pair was very expensive.
In a harsh industrial environment, most cable runs have to be in conduit, and redundant runs were required in critical operations. On top of that, all wiring had to be installed by a licensed plant electrician. That cost was blowing the budget by several hundred thousand dollars. He correctly reasoned that you could actually design and install wireless links on those sensors and have money left over. So that's what he did.
That same scenario is playing out all over the world as engineers discover that it's less expensive to put wireless links in place of wiring in most new industrial systems. It's even simpler and cheaper now than a few years ago when the above project was implemented. Today, engineers have multiple wireless options to monitor sensors remotely or to perform remote control on a pump, lights, HVAC system, or whatever.
Knowing that the average cost of industrial wiring is about $200/foot, you have to wonder how many industrial networking projects or systems are delayed or never implemented because of the unbelievable cost of laying cable (think backhoe or conduit) coupled with the related complexity. Wireless has existed for many years in industry, but the new technologies, lower cost and power consumption, improved reliability, and security now makes it a truly viable option.
THE INDUSTRIAL ARENA Industrial networks are used in factories, process control plants, warehouses, oil and gas pipelines, building controls, hospitals, trucks and automobiles, public utilities, and city facilities like lighting and traffic control. The older network and protocols (e.g., DeviceNet, FieldBus, Profibus) are slowly being replaced by none other than the ubiquitous Ethernet. Because it's so widely adopted, the cost per node is very low. Moreover, equipment from multiple vendors lets you do just about anything.A key rationale for Ethernet in the factory is the ability to tie the plant into the office as provided for in many of the newer automated Computer Integrated Manufacturing (CIM) systems. Many companies have developed hardened Ethernet equipment to withstand the harsher industrial environment. And, the Ethernet/Industrial Protocol (IP) standard has emerged to deal with some problems of using an office local-area-network (LAN) technology in a real-time industrial setting. This protocol uses standard off-the-shelf Ethernet products but enhances them with software that ensures deterministic messaging. Thus, monitored time-critical data isn't lost, and control signals don't arrive too late, as might occur with standard Ethernet messaging.
Furthermore, many older protocols are being repackaged in TCP/IP and transmitted via Ethernet. Wired Ethernet industrial networks have seen phenomenal growth, and that pace continues unabated. With the wide availability of wireless Ethernet (Wi-Fi), suddenly wireless becomes a viable option in the plant. In addition, a slew of newer wireless technologies has emerged over the past few years, many that are fully adaptable to industry and some designed particularly for industrial networking.
WIRELESS ALTERNATIVES Wireless networks operate in four basic topologies: point-to-point (also known as peer-to-peer), point-to-multipoint, multipoint-to-point, and mesh (Fig. 1). Point-to-point is found in the simpler systems, where only a single sensor is used or only a simple control function is required. More complex systems employ the multipoint configuration. The most common is where multiple sensors must be read and multiplexed into a central control point. The point-to-multipoint approach is less common but serves a broadcast function to multiple nodes when needed. The mesh topology isn't yet widely used, but wider adoption is expected as the price drops for wireless technologies.
The mesh network uses many closely spaced nodes (Fig. 1d). Each node can communicate with its nearby neighbors that are within range. The nodes can exchange data between themselves or store and forward data meant for another more distant node that's out of range. One of the nodes also serves as a connection to a wired node.
The neat thing about the mesh is that it allows nodes to transmit over a longer range than possible with a normal line-of-sight (LOS) link. The mesh is also more reliable because if one node fails due to a power loss or other defect, communications are still maintained. Data is simply routed through another nearby node. If LOS obstacles occur, the link can be revised to get the data through via another path.
Aside from its reliability, the mesh offers the benefit of requiring only very low power because the distances between nodes is usually short. Low power means lower power consumption and longer battery life in nodes that are independent of standard power.
Mesh networks are beginning to appear in industry for monitoring large networks of sensors. Such networks are sometimes referred to as micro-meshes. The military has already started adopting mesh networks in battlefield systems, and homeland security is an ideal application for this growing technology. As the cost, size, and power consumption of the nodes (sometimes referred to as motes for remotes) improve, many organizations are studying and researching mesh networks in an attempt to define how hundreds or even thousands of sensors can be used.
When it comes to designing and implementing a wireless industrial network, there are more options than you may think. The choice of a technology depends greatly on the application and whether the wireless connection is embedded or added on.
For some applications, the widely available 802.11 Wi-Fi systems are generally easy to adapt to industrial applications. The 802.11b standard has been around for years, and its maximum 11-Mbit/s data rate is almost overkill for most industrial applications. You can easily operate the Wi-Fi transceiver at one of its alternative rates of 5.5 Mbits/s, 2 Mbits/s, or 1 Mbit/s and get extra range and reliability while still providing more than an adequate data-transfer rate. The primary design challenge with using Wi-Fi is interfacing to the industrial equipment, which often involves original design.
What about the faster versions of this standard, like 802.11a and 802.11g? Both offer up to 54-Mbit/s data rates. Few if any industrial applications require this exalted rate. As for the 802.11a standard, it uses the 5-GHz band instead of the 2.4-GHz band. The range is more limited at that frequency, and range is a more important factor than speed in most industrial situations. As for the newer 802.11g, it too has overkill speed, but it does operate in the same 2.4-GHz band used by 802.11b. The really good news is that it's fully backward-compatible with 802.11b. Most Wi-Fi equipment manufacturers and the chip companies have all but totally abandoned the pure 802.11b products. So if you plan to go to Wi-Fi in an industrial setting, you'll be using 802.11g whether you need it or not.
Another option is embedded Wi-Fi. If the application is unique and you have the expertise, chips can be bought and built into the system. But a faster and easier way is to use the module approach.
One such example is DPAC Technologies' Airborne Wireless LAN module (Fig. 2). It includes the fully 802.11b-compatible radio, a baseband processor to handle the MAC functions, and an applications processor. The applications processor runs an RTOS and includes a TCP/IP stack, a command interface, and a wide range of I/O support. It handles a variety of serial I/O, including standard UART, I2C, and SPI. Up to eight parallel I/O ports are available, as are eight channels of analog. A neat feature is the built-in Web server, which allows you to remotely monitor and control any device with a browser via the Internet. With this module, you only need an antenna, a 3.3-V supply, and your I/O.
Andrew Samson, DPAC's product marketing director, indicated that the module is being widely adopted in trucking and automotive applications for diagnostics and fleet management, medical patient monitoring, and farming for automated irrigation systems. Mike Grobler, director of wireless applications for DPAC, says that 802.11b systems provide the greatest range of almost any industrial wireless technology, mainly because it transmits at the highest power permitted by ISM radios (up to 20 dBm). In ordinary situations, the range of Wi-Fi is about 100 feet at 11 Mbits/s. But longer range is possible at the lower rates. And if you use a gain antenna such as a simple Yagi, LOS range can be several miles. Also, the diversity switched antennas help mitigate multipath problems.
The industrial wireless technology with the greatest potential must be ZigBee. This short-range wireless technology was a premeditated design specifically for industrial applications. Over 80 companies have come together in the ZigBee Alliance to promote and support this technology. ZigBee is also known by its IEEE standard designation 802.15.4. The standard defines the PHY and MAC layers, while the ZigBee Alliance sets the standards for the networking, security, and applications layers.
ZigBee comes in three versions. The 868-MHz European version has a data rate of 20 kbits/s. The 915-MHz version with 10 channels has a 40-kbit/s rate. The real speed demon, the 2.4-GHz version, has 16 channels and a 250-kbit/s data rate. All use direct-sequence spread spectrum (DSSS).
Also, ZigBee is a personal-area-network (PAN) technology that can automatically establish links with nearby nodes. It can be configured in a star, tree, or mesh topology, which makes the technology ideal for mini-mesh networks.
Key benefits are ultra-low power consumption and simplicity, which ultimately translates into low cost. Industrial designers can now economically build wireless into the most mundane products, such as light fixtures and wall switches. The low power allows battery operation for up to several years in some cases.
John Adams, director of radio technology and strategy for Freescale Semiconductor (formerly Motorola), indicates that while no commercial products are on the market today, they're soon to come. He also says that the ZigBee Alliance is expected to finalize its standard in the fourth quarter of this year. The alliance will establish a testing and certification program similar to Wi-Fi to ensure standards compliance and full interoperability. Among the first commercial products expected late this year or early in 2005 are residential and commercial lighting controls for energy savings, security and fire alarms, and industrial sensors.
A typical ZigBee product is Freescale Semiconductor's MC13191/92 transceivers. These chips employ the 16-channel, 2.4-GHz band with a data rate of 250 kbits/s. They're designed for use with an 8-bit embedded controller like Freescale's MC9SO8GT family (Fig. 3). Also available is the low-cost MC13191/92 Developer's Starter Kit, which contains two sensor-application reference designs and the CodeWarrior Development Studio of software. ZigBee chips and support products additionally hail from Atmel, Betronic Design BV, Chipcon, CompXs, Microchip Technology, Rincon Research, and ZMD.
An interesting option to ZigBee is Cypress Semiconductor's WirelessUSB chips. Like ZigBee, they operate in the 2.4-GHz band using DSSS but with a data rate of 62.5 kbits/s. For simple point-to-point and multipoint-to-point applications, WirelessUSB is less expensive and simpler still than ZigBee. It doesn't have the automatic networking capability, but it's not a requirement in most applications. Carl Brasek, Cypress' product manager for USB, says that the long-range (LR) version of the CYWYSB86935 can give a range up to 50 meters, and its ultra-low power consumption makes battery power practical.
The lowest-cost and simplest form of wireless can be implemented with the chips and modules that operate in the UHF 300- to 928-MHz ISM bands. Common frequencies are 315, 433, and 915 MHz. Chip companies like Maxim, Micrel, and RF Micro Devices produce receivers, transmitters, and transceivers for these frequencies using ASK/OOK (amplitude-shift keyed/on-off keyed) as the modulation. However, some will incorporate FSK (frequency-shift keyed). Range is limited to roughly 50 meters in most cases. You usually find these chips in remote-keyless-entry (RKE) systems on cars and trucks, wireless tire-pressure monitors, and garage-door openers, but their ultra-low cost and simplicity can benefit a myriad of industrial applications.
A typical chip set comes from Maxim Integrated Products. The MAX1472 transmitter operates in the 300- to 450-MHz range and uses a phase-locked loop (PLL) to multiply the crystal input to the desired frequency. Data rate is up to 100 kbits/s using ASK/OOK. Output is 10 dBm into a 50-(omega) antenna. The companion superheterodyne receiver is the MAX1473, which has an amazing sensitivity of 115 dB. The latest product, the MAX1471 receiver, can receive ASK and FSK data simultaneously in automotive systems that use ASK for remote entry and FSK for tire-pressure monitoring.
You can also purchase ready-made modules using similar chips. Some common vendors are Abacom Technologies, Limos International, Linx Technologies, MaxStream, Radiotronix, and Xemics. Longer-range radio modems are available as well. For instance, there's Aerocomm's new ConnexLink transceiver (Fig. 4). It operates in the 900-MHz band with a full 1-W output. It can achieve a range up to 20 miles with LOS conditions. Its inputs are compatible with RS-232/422/485 and can achieve a data rate up to 115.2 kbits/s.
The main downside of these inexpensive products is that you have to invent your own protocol for addressing, data transmission, error detection, and the like. This is no big deal in simple applications, but if you want the least amount of development work, stick with an existing protocol product.
As for the other short-range wireless technologies like infrared (IR), magnetic, near field communication (NFC), and ultra wideband (UWB), they all have one or more limitations besides the very short range that makes them less desirable for industrial uses. The big exception is radio-frequency identification (RFID), the short-range wireless bar-code replacement that's taking the industrial market by storm (see "RFID: Another Wireless Technology Perfect For Industrial Apps," p. 54).
One final consideration is the M2M movement, which means "machine-to-machine," "mobile-to-machine," or "machine-to-mobile." This wireless trend supports direct machine-to-machine communications as well as communications between people and machines. By using wireless methods and the Internet, you can remotely monitor and control devices using a cell phone, PDA, or laptop.
Nokia has come up with an entire M2M system using the GSM cell-phone system. Devices to be controlled connect to a GSM terminal module via RS-232 and can then be accessed by cell phone. The number of potential industrial applications simply boggles the mind.
DESIGN ISSUES As usual, cost and ROI issues dominate any decision on the adoption of wireless networking. The current state of the industry is simply that almost anything now wired can be connected wirelessly for comparable cost. But other issues come into play too. The most important is noise/interference, range, power consumption, and security.When it comes to noise and interference, transients from turning power to motors and welders off and on creates horrible interference for wireless. Interference from other wireless systems may also be a problem. Another particularly worrisome problem is multipath.
While wireless signals penetrate walls and other obstructions to a degree, the attenuation is significant in limiting range. Furthermore, reflections from metallic obstacles create multipath fading and attenuation. And, the mobility of machines and materials in a factory environment creates an ever-changing multipath environment. Diversity receivers and antennas can help in tough situations. To anticipate problems, it's a good idea to get a handle on the environment by doing a site survey.
Range is another key factor. Most available wireless technologies are designed for the short range, usually less than several hundred feet. Range can be extended by using more transmit power, more sensitive receivers, and directional gain antennas high and in the clear. Many miles are possible.
Pay particular attention to the antennas. You have full control in this area. Select the best type and carefully place it in the optimum position. The right antenna will give you the end result you're expecting.
Power consumption is a major consideration in many applications. Fixed wireless nodes in a factory can use existing ac power, so it's not usually an issue. But for wireless sensors that use battery power, it becomes a huge factor.
In the past, battery-powered devices were virtually banned by industry simply because of the enormous cost of battery replacement or maintenance. Even today, engineers are reluctant to use such devices. But great progress has been made in battery technology as well as the availability of very low-power wireless chips like ZigBee. With the right combination of battery and chip, battery life can extend for many years, making maintenance less of an issue.
Security is said to be the greatest issue in wireless these days, especially in the wireless-LAN (WLAN) area. If you use a WLAN in the enterprise or hot spot with 802.11, you can be subjected to monitoring and other hacking from outside. For most business communications, security is paramount. But it's not such a big deal in most industrial applications. Generally speaking, over 90% of industrial applications are simple sensor data reading or on/off controls. So there's little benefit to hackers with such information.
For those few applications that must be secure, like medical, a variety of encryption methods are available. Enabling WEP in Wi-Fi or using one of the other available encryption methods should solve the problem. ZigBee implements access control lists or 128-bit AES for security. Elliptic Curve Cryptology (ECC) has been suggested for ZigBee because it uses smaller key sizes and scales better than other forms of public-key encryption.
While the bulk of industrial will be simple monitoring and control, Hesh Kagan of Invensys, a founder of the Wireless Industrial Networking Association (WINA), says that the really big killer apps have yet to be implemented. For example, with inexpensive "peel and stick" wireless sensors, we can eventually look forward to improved capital equipment asset management via condition monitoring. Multiple sensors will be used to monitor the many details of an expensive machine so that users can predict when maintenance is needed. Ultimately, it's expected to save significant cost that's currently earmarked for preventative maintenance.
Process optimization is another possibility via multiple sensors to measure conditions in a process not now being used. The more parameters monitored, the better the process can be analyzed and optimized to minimize energy, materials, and other resources for major savings. We're now at that point where it can be done economically.
NEED MORE INFORMATION?Atmel Corp.www.atmel.comBetronic Design BVwww.betronic.comChipcon Awww.chipcon.comCompXs Inc.www.compxs.comCypress Semiconductor Corp.www.cypress.comFreescale Semiconductor Inc.www.freescale.comMaxim Integrated Productswww.maxim-ic.comMicrochip Technology Inc.www.microchip.comNokiawww.nokia.comTexas Instruments Inc.www.ti.com/tirisWireless Industry Networking Alliance www.wina.orgZigBee Alliancewww.zigbee.comZMD AGwww.zmd.biz