•  
• EDN Europe MAGAZINE
•  
• Free SUBSCRIPTION
• Subscription Renewal
• View Media Kit

• EDN
• EDN Australia
• EDN Asia
• EDN China
• EDN Japan

ZigBee and other wireless-sensor networks: time to connect?

THE PROMISE OF UBIQUITOUS WIRELESS CONNECTIVITY FOR LOW-POWER SENSOR-AND-CONTROL PRODUCTS SEEMS TO HAVE BEEN WITH US FOR A CONSIDERABLE TIME WITHOUT MAKING THE FINAL TRANSITION TO HIGH-VOLUME SALES. REVISED PROTOCOLS AND STANDARDS, AND A NEW GENERATION OF SILICON, MAY BE ABOUT TO MAKE THAT STEP A REALITY.

 

You may recall that, when the ZigBee standard first emerged in public— and that time is now far enough away for us to measure it in years—its proponents assured us that quickly, and certainly by 2009, we would not only be seeing the technology deployed in professional wirelesssensor applications, but would be able to enter our local hardware store and select off-the-shelf wireless products for the home from multiple vendors, confident that they would happily inter-operate according to the standard. We are—as you may have noticed—not entirely there yet.

Multiple implementations and standards—or would-be standards—have not helped, but, in truth, the standards issue has not been the limiting factor in this story. Very few of the proposed wireless-connection schemes have fallen by the wayside, and there is not much evidence that this fate is about to befall any of them. There are, generically, wireless-connection schemes that reside in the licence-free bands, the most significant being the spectrum just above 2.4 GHz, together with a number of narrow allocations below 1 GHz, exact bands varying according to geography. A subset of these schemes conforms to the provisions of IEEE 802.15.4. Within those, in turn, reside standards such as ZigBee—which fills out a complete protocol and includes the provision for selforganising mesh networking. This much is well-known, but it is worth repeating because you will sometimes hear “ZigBee”, in particular, used loosely as a generic term for low-power wireless networking in general—more on this in the sidebar “Whatever happened to….”.

Having identified that generalisation, I will fall into it by asking, “What is the status of ZigBee—and, by association, the complete sector—today? And what happened to the expected proliferation of applications?

A FAMILIAR TALE

The story seems to be almost routine in the wireless business: a proposal for a standard emerges, with promises of rapid deployment. Before any vendor achieves significant production, enthusiasts conceive large numbers of possible applications scenarios, diluting the development effort. Users try out the first systems that come to market and discover that everything is not as simple as they had been led to believe. Disillusion reigns for a time; the standards body then creates a revision that works as they first intended: vendors do likewise. All being well, sales then recover from the collapse they suffered after the first enthusiasm faded and begin rapid growth towards high-volume production. Bluetooth followed almost precisely this trajectory, and there are signs that ZigBee is now, finally, about to enter its rapid-growth phase. ZigBee has long used the tag-line “Wireless control that simply works.” This might be the appropriate moment for the sceptic to acknowledge: “OK—now it does.”

ZigBee is not alone; both within and outside the 802.15.4 class of products, other offerings bases on alliances, standards and would-be standards exist (for example, see the expanded online version of this article at www.edn-europe.com/article.asp?articleid=2956 for more information about the offerings from EnOcean and Z-Wave). These multivendor solution reside in the market alongside purely proprietary offerings—one of the first decisions you will need to make when specifying a wireless-sensor/ -control system is: Do you need a standardsbased, interoperable system, or a proprietary one? Vendors’ estimates vary, but most report that the majority of 802.15.4 systems shipped today are still proprietary.

Nor is ZigBee itself static: most recently, for example, it has absorbed RF4CE (RF for consumer equipment), a simplified protocol for applications such as remote controls.

SMART METERING ADDS ENERGY

A key driving force that is emerging to push ZigBee applications to higher volumes is the focus that the organisation has placed on smart metering—especially in the United States. ZigBee Chairman Bob Heile observes that, especially in today’s economic climate, the prospect of saving money on utility bills will be a key deciding factor for consumers to adopt ZigBee-based products. Added to this will be the “push” factor from utility companies deploying meters into homes. In this view, the electricity meter becomes a real-timedata and -control gateway between the electricity supplier and the home. Smart meters will gather information on power usage from sensors within the home, using wireless connectivity and giving the consumer information on their minute-tominute power costs. Clusters of meters, also linked wirelessly, will provide billing data upstream to the utility; and, in the reverse direction, the utility will be able to manage its peak demand by exercising a degree of control over consumers’ appliances and installations, adjusting the on/off cycle of a freezer or turning down a room thermostat for a short time. In return for providing that access to the utility, the consumer will benefit from a lower-price energy tariff.

Aside from this shift in emphasis, the overall architecture of ZigBee remains essentially unchanged: it is a wireless data network in which the outlying sensor nodes are battery-powered, and the infrastructure-connected nodes are linepowered. The battery-powered nodes operate, in the most common use case, at very low duty cycles, spending most of their time “asleep”, with the result that very long battery life—several years—is possible with small cells. The network is self-organising and self-repairing in a variety of topologies, the most widely publicised being a mesh configuration. The details are readily available on the Alliance’s Web site and on the Web sites of all the member—vendor—companies that the Alliance’s Web page lists.

STABLE STACK SETS THE SCENE

In the early days, the questions that would-be users—and editors—asked of ZigBee providers concentrated on basic operating parameters: Can your RF transceiver meet the range requirements, indoors and out? How many nodes can you have in a single mesh? Will the system really successfully re-route data around a failed or blocked node?

Today, and not least because the standard has been through major and minor revisions over the period from 2004 until today—with the ZigBee Pro version (2007) now being robust—anecdotal evidence indicates that as long as you configure systems within data-sheet limits, you can assume that a system will meet those basic specifications. You should be free to focus on your main development effort on the application.

Against that background, several silicon suppliers are engaged in a round of updating their product offerings. As one example, Jennic has introduced a singlechip device with increased memory space to accomodate the larger protocol stack that comes with ZigBee Pro. The JN5148 has 128 kbytes each of ROM and RAM—which is enough, according to marketing vice president Tony Lucido, to host both the ZigBee stack and a user application. The 5148 uses 18 mA when receiving and 15 mA when transmitting at +3dBm. This lets you use a small coin cell as a power source; Lucido points out that peak operating current primarily sets the minimum size of battery that you can use in a sensor node: battery life is a function of the quiescent current and, to a much greater extent, of the duty cycle of the battery-powered node. Jennic addresses the issue of scalable processor performance by using its own RISC CPU on the single chip alongside the 802.15.4 transceiver; you can set clock speeds from 4 to 32 MHz and selectively activate five power domains. RF link budget is 98 dB. The Jennic device incorporates a time-of-flight ranging function for more precise and reliable node location than is possible with RSSI (receivedsignal- strength) techniques. You might triangulate from fixed nodes to locate a mobile one, or simply measure distance periodically to ensure that an asset has not moved. Additional features include 128-bit AES encryption.

ORIGINS OF RF4CE

In addition to its involvement with other low-power wireless systems, Freescale has been a major player in the story of the standard that has become RF4CE. In 2008, the company offered its SynkroRF entertainment-control-network technology as an open specification to a wider array of CE (consumer-electronics) manufacturers; this became the RF4CE Alliance. ZigBee and RF4CE joined forces earlier in 2009 to create a standardised specification for RF remote controls. The CE manufacturers are seeking to overcome the limitations of infra-red controls, permitting more sophisticated control functions, enhanced battery life, nonline- of-sight operation and bi-directional functions. Battery life improves compared to an infra-red product as a direct result of bi-directional operation: an RF remote can send a single packet, receive an acknowledgement and shut down. RF4CE is, like ZigBee, 802.15.4-compliant, but limited to communication over a single point-point link. The simpler protocol stack occupies much less memory than does that of ZigBee; Freescale cites under 40 kbytes, as opposed to 60 to 120 kbytes for a full ZigBee implementation. This equates to reduced silicon area and cost, which is of paramount importance to the CE sector. RF4Ce and ZigBee will not interoperate directly at the air-interface level—at least not in early implementations. However, you might create application-level interoperation: you could bring data from one signal standard up through the protocol layers, process it in the application layer and dispatch it to the other standard—“a simple software translator” is how Freescale describes what you would need in order to get from RF4CE to the ZigBee home-automation profile. Once you have interworking between home automation and entertainment, the possibility exists that the utility meter may not be the gateway of choice for your utility company to interact with your home—the set-top box, or the broadband modem, might equally be the portal. Remote-control products using Freescale’s Synkro are already in the market; the company says that the first products using RF4CE—the differences between the two standards being relatively slight—are due for launch later in 2009.

At Texas Instruments’ division in Oslo, Norway, the product line features the CC2520, a transceiver chip for 802.15.4 systems, and the 2430, a single-chip device that includes an 8051 microcontroller core. TI’s marketing and systems engineer Peder Rand confirms that most new activity in the sector focuses on Smart Energy, but he also sees prospects for the entry of RF4CE into the mix. If and when these two efforts join forces, “it could change home automation…and lead to a take-off [of the market sector].” TI is focusing its software efforts on home automation, smart energy and RF4CE.

6loPAN EDGES IN

In recent weeks, Atmel (www.atmel. com) acquired all of MeshNetics’ ZigBee intellectual property; it had previously licensed the MeshNetics protocol stack. The acquisition includes the BitCloud ZigBee PRO software and the ZigBit wireless modules. Atmel’s offering covers 700 MHz, 800 MHz, 900 MHz and 2.4 GHz frequencies with a claimed link budget of over 120dB. The BitCloud ZigBee PRO software stack is a secondgeneration embedded ZigBee-certified stack for Atmel’s wireless platforms. The company says that it believes that a twochip solution is a good fit for this market as the application space is so diverse: “We can offer over 100 microcontrollers with the RF elements… the acquisition brings module products that we have added to our range.” Atmel believes that RF4CE is potentially a major driver for the sector; a company spokesman also observes the overlap that exists with yet another standard, the IETF’s (Internet Engineering Task Force) 6loPAN, which aspires to being an “all-IP (internet protocol)” environment. “It’s most attractive to those building big networks, who want IP end-to-end,” Atmel concludes. He notes that the electricity utilities driving the smart-metering initiatives may want to see IPv6 signalling on the grid, but ZigBee in the home—a hint of some possible further standards integration in the future?

According to a spokesman for Ember, 2009 will be a transition year, with the smart-metering programmes getting under way. He points out that there’s a threshold effect: the utilities can’t benefit from demand management—and therefore will not promote the in-home systems to consumers—until they have a critical mass of meters installed. In contrast to other suppliers in the sector, Ember is a “pure-play” ZigBee company and expects to benefit as home-automation-system makers design more products for the smart-meter environment. “They want the metering data, and the meter is not going to be proprietary,” the spokesman specifies. Acknowledging the fact that all of the silicon suppliers in the sector are in the process of introducing “next-generation” chips or chipsets, Ember hints at an imminent announcement in its own range. The company is also considering the possibility of the “sleeping-network” structure—ZigBee has conventionally had an always-on network. This structure can be achieved at the application level, Ember says; in effect, you trade power for latency, waiting for pre-determined synchronisation windows to open before communicating. “Now that the ZigBee stack is stable, reliable and works well, it’s possible to look at these other cases,” the spokesman says.

Along with several other suppliers, Ember also notes the growing level of interestin Zigbee in the healthcare and medicalinstrumentation sector.

RIVAL STANDARDS

It is not only in the emerging RF remote market that Zigbee finds itself facing competition. In the building-automation arena—which Zigbee might regard as its “home patch”—there are alternative approaches from factions such as the EnOcean Alliance, and from Z-Wave.

Approximately one year ago a grouping of companies in—broadly—the buildingautomation and -control sector joined forces to form the EnOcean Alliance, employing the products of EnOcean GmbH. The particular focus of this programme is very-low-energy systems, and their application to the management of “green”—energy-efficient—buildings. Switches and sensors in the EnOcean environment operate at energy levels that you can meet with energy-harvesting techniques, such as small solar cells, and with electromechanical contrivances that employ permanent magnets and coils to fleetingly power up a sensor node. EnOcean uses the unlicensed bands at 315 or 868 MHz to take advantage of the more favourable propagation characteristics—compared to 2.4 GHz—for very-low-level signals within buildings.

EnOcean constructs its Dolphin architecture around the EO30001, a single-chip solution with on-board RF, an 8051 core, and 32 kbytes of flash memory that operates in the tens-of-nanoampere region.

The Z-Wave Alliance also promotes a mesh-networking wireless scheme and claims several hundred already in the market place. It, too, is specifically targeting the smart-metering market and recently strengthened its presence in Europe by adding Danish metering company Kamstrup to its grouping.