Wireless-sensor networks find a fit in the unlicensed band

Low-power, short-range, low-data-rate wireless networks use the unlicensed RF band. In addition to their overriding need to operate on severely constrained energy sources, such as using one battery for their entire lifetimes or employing energy they scavenge from the environment, these networks must contend with interference from sources such as Bluetooth headsets and microwave ovens. These networks use network protocols that largely determine the networks' efficiency, robustness, and security. As such, network designers need to first determine which sub-band is best for their application: 900 MHz or 2.4 GHz. In addition to selecting the frequency band, they also need to decide on a hardware-transceiver scheme: a roll-your-own or an SOC (system-on-chip) approach. They must also decide which network protocol to use.

Both the 900-MHz and 2.4-GHz bands have advantages. Both reside in the ISM (industrial/scientific/medical) band, an unlicensed frequency band of 902 to 928 MHz, 2.4 to 2.483 GHz, and 5.725 to 5.875 GHz. Almost all of the transceiver and SOC products targeting wireless-sensor-network applications use the 900- to 928-MHz and 2.4- to 2.483-GHz bands. The 900-MHz band touts long broadcast range because of its relatively longer wavelength and its correspondingly longer battery life. However, lower frequency means the use of a larger antenna than a 2.4-GHz system requires. And, if you plan to sell your system into a global market, you will quickly encounter a lack of standardization in the 900-MHz range. For example, in Europe, you cannot use the 900- to 928-MHz band because it is part of the GSM (Global System for Mobile communications) network for cell-phone communication and, thus, is unavailable.

Wireless-sensor-network-system pioneers, such as Aerocomm, Dust Networks, and Crossbow Technologies, take different approaches to these constraints. For example, Aerocomm sells wireless-network systems into the global market, offering different radios and frequencies for different regions. Randy Macke, international marketing manager at the company, explains, "We typically use 2.4 GHz for higher-data-rate applications at short range. The flip side is when you go to a 900-MHz product family, there is better range performance but typically a slower data rate. These are the basic trade-offs." He says that the company aims at making its products global by using one proprietary protocol and interface on every radio it sells; hence, a customer can base a product on Aerocomm's 900-MHz product and then sell into another country that doesn't accept 900 MHz just by swapping in a pin-compatible radio.

David Boylan, marketing manager for Analog Devices, also recommends this approach. Boylan sees companies opting for the 900-MHz or the 2.4-GHz frequencies because of the technical attributes of the two bands and whether these companies need to adapt to a proprietary or a standard network. According to Boylan, vendors often choose the 2.4-GHz band because either the vendor or its customers like the security of standards. But this preference is not universal. Other companies prefer to create their own proprietary networks, he says. Reasons for this preference are the value-added and security aspects of proprietary networks. "If they have their own proprietary network, they can implement security features that wouldn't be available in a published standard," Boylan says. These vendors typically prefer to use frequencies of less than 1 GHz, such as 900 MHz (Table 1).

Click for larger image

He cites as an example a popular 900-MHz application of an automatic meter reader. "You usually have a central node, which might be the utility employee with his handheld unit pinging several units. The 900-MHz band is popular because range and battery power are key requirements," says Boylan. For networks relying on tiny sensor nodes, the 900-MHz antenna of approximately 8 cm is probably too large. In this case, even if your market is 900-MHz-friendly North America, you may need to consider 2.4 GHz because of its smaller antenna (see sidebar "Antenna technology feels the squeeze"). Airtime also affects power consumption. Wireless-sensor networks conserve power by spending 99.99% of their time asleep, and they wake up only to contact their network and briefly transmit their information, thus achieving their goal of consuming an average of less than 1 µA of current (Reference 1). Therefore, the faster they can transmit their data, the more quickly they can go off the air. This point favors the use of the 2.4-GHz spectrum.

Dust Networks develops both the hardware and the software for companies needing wireless-sensor networks. Dust makes 802.15.4-compliant products in the 2.4-GHz band and in the proprietary-network, 900-MHz band. Many customers use the 900-MHz network devices for building automation. For this application, says Rob Conant, Dust's vice president of marketing and business development, the 900-MHz band presents a technical advantage because of its longer range and better penetration in buildings, which allows lower power consumption (Reference 2). As for the marketing side of the equation, Conant believes that the 2.4-GHz band has an edge. "The global ISM band is the only way to go for companies that want to bring out global products," he says. "Although 2.4-GHz products suffer a penalty in power consumption to get the same range [as 900 MHz], 2.4 GHz is a standard, and customers know it's going to be around for years."

Karl Torvmark, strategic product-marketing manager for Chipcon, agrees. Chipcon's 2.4-GHz Zigbee devices reach 100m or more in good conditions, he says. For a longer communications range, you can increase the output power beyond 1 mW, but doing so directly affects the power consumption. If you go below 1 GHz with the same output power, you reach several hundred meters, but then an antenna issue arises. "In the hundreds of megahertz, you like to have a whip antenna, usually mounted externally ... if that's compatible with your overall product design," he says. Although you can use pc-board antennas operating at less than 1 GHz, they can be impractically big," he states.

Torvmark sees customers leaning toward custom protocols selecting 900 MHz, just as ADI's Boylan does. Chipcon's customers want to customize their networks, which provides advantages and trade-offs in power consumption, he says. For example, Chipcon's 900-MHz-band products, the CC1100 and the higher performance CC1020, don't offer the total network stack that the 2.4-GHz Zigbee products have.

Click for larger image

Clearly, a range of opinion exists about the best way to handle network protocols for 802.15.4 networks (see sidebar "A brief guide to 802.15.4 and network protocols"). The benefits of being a standard-compliant technology combine with the approach's global reach and its use of the 2.4-GHz band to make Zigbee a likely winner. IC vendors Freescale, Ember, and Chipcon have all introduced Zigbee-compliant SOC or SIP (system-in-package) products.

Freescale was an early proponent of 802.15.4 and Zigbee, releasing its 802.15.4 transceiver on a chip the day after the IEEE ratified the specification. However, Jon Adams, director of radio technology and strategy for Freescale, claims that system designers don't want to know how to design a network and applications protocol; they just need the basic digital-radio function. So, the company introduced a software-based configuration tool that lets them develop applications using the 802.15.4 stack. Once Zigbee came along last year, he says, the company recognized that designers also need a function for using that Zigbee stack, so that they could quickly build a Zigbee process without reading 700 pages of the spec. "All of our efforts have been toward creating a one-stop shop," he says. Ember and Chipcon have embraced the same approach.

Ember started out developing wireless-sensor-network systems, such as its proprietary EmberNet, before Zigbee existed. The company was also an early member of the Zigbee Alliance. Vice President of Engineering Skip Ashton explains the importance of a multicompany synergy in developing such a complex mesh: "With Zigbee, in addition to the network, you need an ecosystem of the things around your network," he says. Although one company can provide a network stack, a lot of companies are working on other tools and services in a standards-based environment. Other advantages of Zigbee include interoperability testing, multiple-vendor testing, and the availability of multiple sources. Ember's EM250 SOC combines a 2.4-GHz RF transceiver with a 16-bit XAP core, which Cambridge Consultants developed and Cambridge Silicon Radio uses in its Bluetooth chips. The EM250 also includes two timers, a sigma-delta ADC, a USART, an I²C port, and an integrated encryption core for the Zigbee security algorithms. The company also offers the EM260, which lacks the processor core, allowing designers to choose their own processor and peripheral controller. Ashton agrees with Zigbee's critics that its developers have yet to address all its deficiencies, but he's confident that they will over time. "There are always migration and backward-compatibility issues," he says. As the Zigbee standard evolves, users will be able to perform upgrades in the field through software updates.