Remy Malan
Barrington Partners
Management Inc.
Menlo Park, Calif.
Wireless mesh networks have been getting a lot of attention recently. Interestingly, many of the applications developers have in mind for these networks are either industrial or in building management.
A few simple examples illustrate the allure: It may soon be possible to install room thermostats that are wireless. This saves time and cost by eliminating the need for running signal wires. Ditto for lighting and HVAC controls in commercial buildings. Facilities operators will manage entire buildings with a handheld controller.
In factories, wireless sensors will likely speed up the commissioning of production machinery. During setup, developers must typically tweak the position of sensors that feedback the position of moving actuators. Wireless devices may simplify this process by eliminating the rewiring and cable management that generally accompanies deployment.
Some have voiced concerns about EMI and other factors that could pose problems for wireless links in such environments. But results presented at the recent Industrial Wireless Automation Summit (www.IWASummit.com) show that manufacturers are already deploying wireless sensor nets that function reliably despite such difficulties.
A number of companies, including several startups, are actively bringing wireless mesh technologies to market. They frequently cite terms like 802.15.4 and ZigBee in their literature. These are standards to which wireless sensor networks will be built.
Thus a good place to start a discussion about such products is to look at exactly where the standards are and what they specify. Today, two standards groups play key roles in wireless mesh networks: IEEE 802.15.4 and the ZigBee Alliance. Currently there is an approved standard for IEEE 802.15.4, which is a wireless mesh networking standard. This standard was ratified in May 2003 and covers wireless mesh networks using two different frequency bands, 868/915 MHz and 2.4GHz, as well as data rates of 20, 40, and 250 kbps.
The standard is basic from an application perspective; essentially, it only covers the physical transport layer (PHY) and the media access layer (MAC), the lowest portions of a networking model. Nonetheless, it is important because these two layers are the basic building blocks on top of which developers construct more abstract and powerful networking layers and, ultimately, the network interfaces to applications.
IEEE 802.15.4 does not specify many aspects of a network that are important to standardize, such as routing and session management. This is one reason the ZigBee Alliance formed in 2002. It is useful to think of the ZigBee Alliance as having a role similar to the Wi-Fi Alliance. The Wi-Fi Alliance was instrumental in ensuring the interoperability and, as a result, rapid and widespread adoption of 802.11 wireless LAN systems. Currently, there is no ratified ZigBee standard but one is expected sometime this year.
A QUICK PRIMER
It may be helpful to review the basics of the 802.15.4 standard and how ZigBee builds upon it. The IEEE standard spells out a direct sequence spread spectrum (DSSS) transmission scheme using binary phase shift keying (BPSK) for 868/915 MHz and offset-quadrature phase shift keying (O-QPSK) for 2.4 GHz.
The PHY part of the spec covers detecting the RF link and assessing the quality of the received signal. It also spells out access methods to the communication channel. Also covered are packet size and addressing of network nodes. The MAC layer further defines how nodes transmit over the channel, especially when there are multiple nodes trying to communicate simultaneously.
On top of this structure, ZigBee defines layers for network, security, and application profiles. Its network layer handles network topologies of star, mesh, and cluster tree. Of these, mesh and cluster tree are probably of most interest for industrial needs. Mesh or peer-to-peer networks provide more than one path through the network for a wireless link. This makes them highly reliable in environments characterized by a lot of RF interference. Cluster-tree networks are hybrids of mesh and star topologies. They provide reliability while keeping power drain to a minimum in battery powered nodes.
ZigBee further defines physical devices based on IEEE 802.15.4 as reduced function or full function devices, dubbed RFD and FFD respectively. The definitions pertain largely to the intended functions of the ZigBee device. For example, an RFD can be implemented with minimum hardware resources and is designed to serve as a simple send/receive node in a larger network. RFDs also have facilities for a sleep mode that conserves power. Thus most battery powered wireless sensors will probably take the form of RFDs.
RFDs are only able to talk to FFDs, which are devices that have enough resources to act as network controllers and handle network routing. FFDs talk to other FFDs and to RFDs, and discover ZigBee devices in the area to establish communications.
ZigBee also spells out what are called logical device types which perform various functions in network management. Other provisions spell out application profiles — basically, definitions that ensure common devices such as light switches, HVAC controllers, and so forth all work the same way and will be interoperable.
One of the principal attractions for ZigBee networks is that they are selfforming and self-healing. That means messages can pass from one node to another via multiple paths. If one path becomes unavailable, nodes have enough intelligence to reroute traffic around it. Further, there are provisions for security such as 128-bit encryption. Quality of Service definitions provide a guaranteed time slot for devices that must get access to a network quickly. Applications in this class include security alarms and medical alert devices.
Finally, 802.15.4/ZigBee networks are optimized for low-duty cycle transmissions. New nodes typically get recognized and connect within 30 msec. The process of waking up a sleeping node and transmitting data takes about 15 msec, as does accessing a channel and transmitting.
TRACKING THE PLAYERS
The ratification of IEEE 802.15.4 and the expected arrival of the ZigBee Alliance standard has motivated vendors to work on products in this area. A useful way to gain perspective on what is happening is to categorize vendors within a "technology stack" that would be necessary to build applications. Organizing this way makes it easier to see where vendors are focusing their attention.
The current membership of the ZigBee Alliance serves as a directory of companies with an interest in standardized wireless mesh networks. A technology stack starts with semiconductors and then moves up to hardware, software, sensors, and finally to integration and management platforms. It is also helpful to map vendors by their relative focus on mesh networks. The results of this exercise take the form of a landscape diagram (see Zig-Bee vendor landscape figure).
One of the striking things we note in the resulting landscape mapping is the sheer number of vendors actively working in semiconductors, hardware, or both. We count 47 companies, or almost 80% of the current ZigBee Alliance members, active in these areas.
The other noticeable trend is that numerous startups and smaller companies are positioning themselves to focus on specific semiconductor, hardware, and software product strategies. This is consistent with an early-stage market where a large number of market entrants build the technology stack from the ground up.
We also note that a number of broadline vendors are on the landscape. We expect these broad-line suppliers will become channels for the focused companies in the landscape. They will probably take on OEM, value-added reseller, and system integrator roles for the markets that they serve.
With all this vendor activity, what kinds of market developments should we expect as products based on IEEE 802.15.4/ZigBee Alliance standards become established? First, we anticipate vendors now developing low-cost, highvolume, standard products will see the lowest layers of the networking stack as an opportunity. For example, single-chip 802.15.4 radios could serve as anchors for commodity products.
By way of analogy, the 802.11 semiconductor market is becoming dominated by companies able to exploit highvolume economies of scale and that know how to continuously lower production costs over a product lifecycle. Over time we expect the process of commoditization will move into successively higher layers. The ultimate result will be a complete, high volume, networking stack available as an off-the-shelf item.
Second, as the lower layers become commodities, the price points of wireless sensor products will drop. Lower costs should open up new application opportunities as well as make it economical to retrofit existing products with 802.15.4/ZigBee.
For example, it is not hard to imagine dense arrays of wireless sensors for instrumentation, or applications with " disposable" sensors (i.e., those where the cost of the sensor/node is no longer a significant part of the BOM). Examples of candidates for retrofitting would include HVAC systems in buildings as well as controls for remotely operated machinery.
Another often-cited example is retrofitting a mesh network to an array of power meters for automatic meter reading (AMR). This would be particularly helpful where banks of meters sit in close proximity to each other.
Third, we expect vendors to offer a vertical integration of the complete application stack, including network technology. To illustrate the idea, it's likely that HVAC and meter-reading systems might both use the same IEEE 802.15.4/ZigBee components. But there will still be protocols and functions in the application stack that will specifically target HVAC or meter reading.
One might envision scenarios where it would desirable to have interoperable applications. Examples might include between HVAC and meter reading in the same physical plant, or between products from two different vendors. To handle such needs, there is still a role for integration interfaces between products at the application level.
On a related note, we expect many vendors will continue to offer products based both on industry specific standards and on proprietary enhancements. It would behoove vendors in a vertical market to adopt IEEE 802.15.4 and Zig-Bee standards for the sake of interoperability. But such policies by themselves would not be enough to create a "plugand-play" environment among products from different vendors.
Take AMR as an example. Power meters would have to follow IEEE 802.15.4 and ZigBee. But they would also need to support the ANSI standards for AMR such as ASNI C12.19 and C12.22. Otherwise they would not be able to work with the other components in an AMR system.
Fourth, we expect plenty of hype! IEEE 802.15.4 and the coming ZigBee standard are poised to rapidly become entrenched in the industrial landscape. There are a lot of claims being made today and there will no doubt be even more made in the future.
Inevitably, the impression created will be that IEEE 802.15.4/ZigBee will solve all networking and application connectivity issues. However, we also see industry specific and application-specific issues affecting products with wireless mesh technology.
All in all, the ratification of IEEE 802.15.4 for wireless mesh networks and the emerging ZigBee Alliance standard will greatly simplify wireless industrial applications. As component volumes increase, IEEE 802.15.4/ZigBee will become ever more economical. End users can expect benefits from new applications.
But as with any new standard, adherence to published specs will not be a given for some time to come. The onus will be on the users to determine how well products have been integrated and tested.
Barrington Partners put together this view of how vendors currently relate to ZigBee offerings.
MAKE CONTACT:
Barrington Partners Management Inc.,
(650) 218-0043,
www.barringtonpartners.com.
ZMD AG,
(858) 674-8433 (San Diego office),
www.zmda.com.