Custom Search

Wednesday, October 28, 2009

The Long Tail of WSN and the Smart Semiconductor Vendors (1)

WSN has a long tail. Let's count: based on 802.15.4, we have Zigbee, WirelessHard, ISA100.11a, MiWi, Cyfi, SimpliciTI, Crossbow, PopNet, Synapse, Greenpeak, JenNet, MeshScape.... Under the cloak of Zigbee are Zigbee-2006, Zigbee-Pro mesh, Zigbee-PRO IP, Zigbee-RF4CE, ... Categorized into 2.4GHz GFSK radio are Bluetooth Low Energy, ANT, Nordic desktop, Nike+ (BodyLan)... Sub 1GHz are Z-Wave, EnOcean, Dash-7, Wavenis, Sensium ...

"Long Tail" describes a new (is 4 years old still counted "new"?) discovered phenomenon of what we are living with abundant choices, and provides the theory to the business model prevailed in the cyberspace and perfectly utilized by companies like Google, Amazon, Apple, eBay...

Most vendors or customers in WSN industry do not feel comfortable seeing such a fragmented or segmented industry. Sure, one of the hardest thing in the world is to make choice, esp. when the choice made by a person is not for the person but for a company. A lot of people are hoping an "open" standard will unite the segmented WSN industry, which will make their choice a non issue. One standard makes everybody on the same page. If the standard is bad, it is bad for everybody; of course, the same goes if it is good. On the other end, some vendors, not only by nature tout how superior their technology is, but also are eager to claim other technology's death, cursing others evil radios, blah blah blah. If you are one of them or you want to avoid to be one of them, I suggest you read the book "The Long Tail", then re-think, then learn to live in this longtailed world.

The long tail theory is about abundance, on the contrary to scarcity that most economy theory care about. It is first stunning by claiming that 80% of profits are coming from 20% of customers or products is outdated in many markets traditionally contained or distorted by the physical limits of the supply chain, such as the warehouse size, the retail shelf space and etc. As an example, Chris Anderson, the author of the famous book "The Long Tail"   gives the number of new movies and songs created every year, the number of movies and songs are able to hit the screen or radio, the number of movies and songs turn to tape, DVD or CD. But Internet gives those movies and songs able to be nowhere a place to stay. These apparently "killed" works, after being aggregated, categorized, tagged and are made searchable, amazingly have consumers willing to pay for. Further amazingly, the totally amount of profit earned from these otherwise dead movies and songs are no longer ignorable, 20%, 30% and some 50% fo the total revenue. Viewing the numbers on a chart, the used to be short list of profitable products is getting longer and longer, thus the name of "long tail".

No doubt, the technology advancement plays an important role to enable the long tail. The low cost of making first allows a surplus of products; Internet and PC creates the ever customer accessible retail points (right on your computer); Amazon, iTune, Adsense, eBay... create the business platform that aggregates and categorizes these products. But fundamentally, it is the unlimited differences among the customers or consumers needs, flavors, preferences. What all the technologies have done is just to free the right of choice, for both the makers and the buyers. 

Take a look of the WSN industry. The long tail is already a reality, a reality of no technology fitting all, a reality of the close to zero hardware cost to commercialize a protocol. The long tail is formed for some reasons. For the same reasons, there is no way to make it short.

Pick range as the starting requirement. The straightforward choice is to pick a low frequency band over 2.4GHz, but then frequency band is not globally available. Ok, choosing 2.4GHz and using a complex modulation scheme and mesh to extend range, but then you hit the wall to see an increasing power consumption, heavy loaded protocol and complaints of hard to design. Moreover, 2.4GHz is not usable under water and badly absorbed by human bodies. When the supporters of reliability have a louder voice, you tend to add things like CDMA, frequency hopping, redundancy and other sort of techniques, only to see an exploding code size, complexity and cost. You may create a protocol able to meet industrial applications, but basically be out of consideration by the consumer market.  To trade-off two, three factors are easy, how about 8, 9... 20, and inter-related ones? Somebody develops scoring and weighing system and even more complex models. The result is obvious a compromised one that does not fit 100% of any use case. 

But on the other side, the semiconductor industry has made programmable microcontroler with much reduced cost of flash memories, the different types of RF radio become commodity, integrated MCU and radio into one (SoC system on chip), integrated multiple MCU, (duo core and multi-core) and has been able to make multiple communication protocols co-exiting in one chip.

What does this mean? That means A can develop a protocol towards this direction and load it to a semiconductor for this market, and B can develop a protocol towards that direction and load it to a semiconductor to that market. Both protocol A and B are using the same radio, or even the same base semiconductor. That also means, as a semiconductor vendor, C's goal becomes to develop a semiconductor architecture that can be protocol agnostic to fit both the needs of A and B. C can ship products loaded with either protocol A or B with almost zero additional cost to fit different customer needs.

This is a non secrete for Zigbee semiconductor vendors, or to be accurate, IEEE802.15.4 radio IC vendors. Less than 1/3 shipped radio were loaded with Zigbee stack; the rest of the loaded protocols are sitting on the long tail.

To a customer, though it is difficult, I believe having options is a better situation than of no choice.

No comments:

Post a Comment