Sherif Hanna of Cypress Semiconductor published an article on wireless design line. This is by far the best and unbiased analysis, in my eyes, of the home and industrial control sector of wireless sensor network. It analyzes the causes of the hype around wireless sensor network, in his term, embedded control (WiEC). For many of his points, I can not agree more.
"The promises varied in their boldness and scope. But some key terms were used by all WiEC technology promoters, almost without exception, to describe their offerings: low power, low cost, high reliability, high security, ease of design-in, and ease of use."
"There are several reasons for the failure to deliver on all those promises (though some were indeed fulfilled). The first is that some target performance goals are fundamentally opposed to one another, and provide formidable engineering challenges. Consider low cost versus high reliability."
"Slimming down the network stack would likely mean that intelligent features like complete node-to-node routing and network self-healing have to go."
After the statement "The new thinking addresses the fundamental flaw in the hype: A single technology cannot solve all problems equally well. But, it may solve a subset of problems extremely well.". Sherif gives his review on different technologies and their fine tuned marketing niches, such as ZigBee in AMI (Automated Metering Infrastructure) and building automation, WirelssHart in process control, Z-wave in home automation.
He points out that Zigbee has "less talk now about consumer applications (where low cost needs make ZigBee less competitive than other technologies) and process automation (where higher reliability and stronger security are needed)." Those two markets are left to Z-wave and WirelessHart.
Both Zigbee and WirelessHart on based on the same PHY. "It turns out that 15.4 is a good transceiver and MAC technology to use for several applications, as long as the network protocol on running on top is fine-tuned to each application."
Sherif then discloses Cypress' high level marketing strategy.
" The good news for designers is that many silicon vendors offer IEEE 802.15.4-compliant solutions, which will eventually mean lower costs and higher performance as the competition heats up. The good news for the silicon vendors is that whether or not a particular protocol like ZigBee or WirelessHART wins, the underlying transceiver is the same, and thus they have a chance to recuperate their investments. This has been one bright spot in the highly fragmented wireless market.
For consumer market (consumer installed home used) to reach millions of units in shipping volumes, "this would only happen if each of the nodes cost the consumer under $5, and they were able to install it in their home unassisted."
I am glad to see my assumption of $5 per node cost (see my blog "average 40 nodes per home") get a resonance here. And the unassisted installation is why ZigBee can not win home automation market.
Enjoy Sherif's article,
Embedded.com - The embedded wireless promise: Where do we stand?
Promoters of the competing WiEC technologies have promised many things to design engineers over the past few years, and the question should be asked: where do we stand?
First, a definition of what qualifies as a WiEC technology should be given. At the transceiver level, they are low power radios that typically have a range between 10m and 50m, data rates under 4Mbps, and operate in any of several Industrial, Scientific, and Medical (ISM) frequency bands.
Network protocols used to control communication between the wireless nodes range from simple point-to-point topologies for machine-to-machine (M2M) communication, through star topologies for basic wireless sensor networks (WSN), to advanced self-healing mesh networks, where all nodes are able to communicate with each other.
Any design engineer who has spent sufficient time in the embedded space is likely familiar with the promises that the promoters of WiEC technologies have made.
The promises varied in their boldness and scope. But some key terms were used by all WiEC technology promoters, almost without exception, to describe their offerings: low power, low cost, high reliability, high security, ease of design-in, and ease of use.
Along with the technical promises came bold forecasts of rapid returns on investment to all involved—the vendors would sell hundreds of millions of units within a few short years, customers would realize increased efficiencies and lower costs, and the world would be covered with small, low power transceivers that connect everything to everything else.
After the hype
Unfortunately, a few years into the hype, no single wireless technology has delivered on all those promises simultaneously. As so often happens in engineering, compromises have to be made.
How can one technology be equally suited to both controlling the lights in a home and controlling a safety valve in a factory? The promoters of some WiEC technologies did make that very promise to a hopefully optimistic market.
However, after the promises made in PowerPoint presentations were vetted in real hardware, the market became increasingly skeptical. Perhaps the promoters over-promised. Perhaps the silicon and stack vendors under-delivered. Perhaps engineers should have not believed that any single technology could solve all problems.
There are several reasons for the failure to deliver on all those promises (though some were indeed fulfilled). The first is that some target performance goals are fundamentally opposed to one another, and provide formidable engineering challenges. Consider low cost versus high reliability.
Engineering a low-cost solution requires a holistic approach to reducing expenses. First, silicon size must be reduced, requiring compromises in transceiver architecture (open-loop versus closed-loop modulation, as an example, with the latter providing higher reliability at increased die size).
Next, network stack size has to be trimmed down, to minimize the amount of code space needed in the processor running the RF transceiver. Slimming down the network stack would likely mean that intelligent features like complete node-to-node routing and network self-healing have to go. Need those features? You must pay the expense in silicon area.
There is nothing fundamentally wrong with this kind of compromise; after all, "low cost" and "high reliability" are relative phrases. How low cost is "low cost?" That depends on who the customer is. Let's say that a particular wireless technology would cost $100 per node. That is prohibitively expensive for a light switch targeted at consumers in home environments. But $100 may be quite acceptable, perhaps even down-right attractive, for a customer designing an industrial process automation solution for factories.
Comparing low cost and high reliability provides only one example of the many concessions that have to be made when designing a wireless system, but it effectively illustrates the challenge. It is not clear why anyone thought that there must be a grand unifying wireless technology.
The unfortunate initial response to this reality, however, was the proliferation of a multitude of WiEC technologies, each promising to deliver what none of the others could.
The list is a veritable collection of trademarks, brands, and clever marketing: ZigBee, ANT, Z-Wave, INSTEON, Wavenis, ISA SP-100, WirelessHART (to name a few), and a host of other proprietary RF technologies from companies like Cypress, Nordic, TI, and Freescale (and many, many more). Each believes that they can build a better widget. It is worth noting that not all those technologies target the same markets, but many of them overlap.
The consequence is a market that is highly fragmented. Customers have a multitude of WiEC technologies to choose from, but no clear winner has emerged. Some customers are in a holding pattern, unwilling to invest in technologies that may eventually disappear because of lack of adoption.
The problem is that this has become a self-fulfilling prophecy—lack of adoption is feeding uncertainty, which in turn leads to lack of adoption. This has also meant that profits have been largely absent for a large number of suppliers. What little money customers are willing to commit to cutting the cord on their applications is being spread too thinly amongst too many suppliers.
Good news
The good news is that this kind of pressure is exactly what is needed to narrow the field down, and allow the more competitive technologies to float to the top. The principle of natural selection is in full-swing.
Already, the market is observing a growing maturity from these technologies, at least when it comes to the promises being made. The hype is slowly dying down, and is being replaced with rational, measured thinking that is based on sound engineering.
The new thinking addresses the fundamental flaw in the hype: A single technology cannot solve all problems equally well. But, it may solve a subset of problems extremely well.
A light switch and a valve in a factory may both benefit from a wireless connection but perhaps not the same type of wireless connection. By better understanding target markets and end applications, suppliers and promoters are fine-tuning their focus and their offerings.
Take ZigBee for example, the mesh technology that promised to be in everything from ultra low-cost consumer applications to mission-critical systems in factories. More recently, the ZigBee Alliance has been focusing on a few select applications that play to the strengths of ZigBee, including Automated Metering Infrastructure (AMI) and commercial building automation.
There is less talk now about consumer applications (where low cost needs make ZigBee less competitive than other technologies) and process automation (where higher reliability and stronger security are needed). The ZigBee specification continues to evolve (most recently with addition of the ZigBee PRO Feature Set) to address the results of real-world issues faced by customers who have attempted to implement ZigBee systems over the past two years. Compatibility between the evolving versions of the specification remains a challenge for both suppliers and customers.
The inability of ZigBee to address the industrial process automation market has resulted in some spin-off technologies. Designers have realized that the RF transceiver technology underlying ZigBee, defined by IEEE 802.15.4, may be the right transceiver for the task if paired with a networking protocol better targeted at challenges specific to process automation.
This has yielded, among others, WirelessHART, promoted by the HART Communication Foundation (HCF). The wired version of HART technology has an installed base of over 20 million units, and the wireless version aims to capitalize on that large existing deployment.
The members of HCF, companies like Emerson Process Management and Endress+Hauser, have fine-tuned the technology to solve problems they are intimately familiar with.
At least in theory, WirelessHART has all the right components to succeed in the market —well-understood problem, fine-tuned technology, and strong backing from industry insiders. It remains to be seen how the first products, which are slated to be released over the coming year, are going to perform in real-world conditions.
That is not to say that every supplier of process automation equipment is backing WirelessHART. Honeywell Process Solutions is pushing its own proprietary OneWireless technology instead, with a promise to adopt the measures outlined in the ISA's SP-100 specification once that is finalized.
An interesting trend to note is the coalescing of both suppliers and customers around the IEEE 802.15.4 specification. It turns out that 15.4 is a good transceiver and MAC technology to use for several applications, as long as the network protocol on running on top is fine-tuned to each application.
Some protocols are being standardized among alliances (ZigBee and WirelessHART), whereas others are building their own proprietary protocols on top of a standard transceiver technology.
The good news for designers is that many silicon vendors offer IEEE 802.15.4-compliant solutions, which will eventually mean lower costs and higher performance as the competition heats up. The good news for the silicon vendors is that whether or not a particular protocol like ZigBee or WirelessHART wins, the underlying transceiver is the same, and thus they have a chance to recuperate their investments. This has been one bright spot in the highly fragmented wireless market.
Moving away from industrial wireless applications to consumer applications, one WiEC technology that has had a sense of purpose from its inception is Z-Wave. Z-Wave is a proprietary technology being promoted by its creator, Zensys Inc, which also leads the Z-Wave Alliance of adopters.
Despite being a single-supplier technology for several years (the specification has recently been opened up to other silicon vendors), Z-Wave has created a moderately-sized ecosystem of adopters because it has specifically targeted home automation, and nothing else.
The Z-Wave Alliance pitted the technology as a lower cost, simpler alternative to ZigBee that still provided mesh networking capability. Focusing on the home automation market enabled Zensys to hone-in on problems more important to the consumer—low cost, ease of use, and compatibility between equipment from multiple vendors (lamp from one vendor, light switch from another).
It is not the most sophisticated wireless technology around—and its exceedingly simple architecture (very low data rates, for example) may in fact prevent it from being implemented in applications more complex than turning a light on or off. Similar to Z-Wave is another technology called INSTEON, which targets the same market, and has a comparably-sized ecosystem of vendors.
Wary consumers
Even with relatively simple WiEC technologies like Z-Wave and INSTEON, wide-spread adoption by consumers has been largely absent. The key reasons from an end-user perspective are cost and ease-of-use.
A regular light switch costs under $2. A Z-Wave enabled wireless light switch costs around $40—and of course it must have a wireless light source to communicate with, which is another $40+.
The 40x ($2 vs. $80) multiplier in price is tremendous for consumers, and the benefit obtained is not compelling enough to justify the added expense. Furthermore, installing a wireless system is still quite challenging, requiring optimal placement and configuration.
This is why ZigBee, Z-Wave and INSTEON remain in the domain of the custom-installer market, where well-off consumers spend thousands of dollars to have specialized companies come and outfit their homes with these wireless solutions. Consequently, these technologies still ship in low volumes, and healthy profits are nowhere to be found.
I once asked a marketing manager at a major lighting control company what it would take for such solutions to reach millions of units in shipping volumes. She said that this would only happen if each of the nodes cost the consumer under $5, and they were able to install it in their home unassisted.
It is clear that this goal is still far away; however, it is not beyond reach. So even though the promise of millions of small WiEC transceivers connecting the world remains unfulfilled today, it is only a matter of time before ubiquitous connectivity becomes a reality.
Businesses need it, consumers expect it, and smart engineers the world over will see to it that they deliver on their promise. It will take a better understanding of the markets, target applications, and the underlying motivations of customers to finally achieve that goal.
About the author
Sherif Hanna Strategic Marketing Manager for Wireless Solutions at Cypress Semiconductor Corp.
No comments:
Post a Comment