Custom Search

Wednesday, April 4, 2012

Can I Develop a Standard? Yes, You Can.

Here is a very informative post “The Cost of Wireless Standards” by Nick Hunn. Reviewing the development history of Wi-Fi, Bluetooth and Zigbee, as well as recent industrial data, he offers some stunning numbers in dollars and years to easily scare away people who ever have dreams to start developing a standard, here are some of my thoughts that may re-energize such dreams.


Near the end of the article, it concludes because of the huge investment of dollars and times, “... that most of the recent 'standards' which have appeared are essentially proprietary solutions, led by a single chip vendor.”

On the contrary, few of the recent “standards” are developed by a chip vendor. With the technology advancement, a different model and development path seems deliver better results. The process seems like this:

1. Chip vendors provide the RF radio + programmable MCU architecture in either separated chips or integrated SoC. Chip vendors usually offer a simple and basic protocol for free for users to experiment and start. Normally such protocol suits simple applications quite well.
 
2. Some talented and ambitious developers start protocol development for their own hungry application needs, maybe forced by the incapable free protocol. This is a relatively low cost (but challenging) embedded software development. There are even open source projects of such.

3. The outstanding protocol from such developments reaches out and gets adopted by other companies in the industry and becomes the de facto “standard”, and an IP licensing business is naturally formed around the protocol.

“Led by a single chip vendor” is an illusion of such process that, economically speaking, the protocol developer, at the beginning, can only develop on one platform from one silicon vendor that offers the best parameters and features relevant to the application. This vendor is naturally the go-after when it is time to license the protocol.

The development of ANT, Nike+ and Polar WIND are all following this model using the Nordic Semiconductor’s radio. Apparently, at the time of early this century, Nordic offers the best radio in 2.4GHz in terms of low power operation, so these protocol developers trying to solve the sports application puzzle select this radio. (There were other options from vendors like Cypress) Now ANT is the only one that have eventually licensed to multiple vendors, thus become the standard. The 802.15.4 radio tells the same story. Reports say more 802.15.4 radio are sold and used for protocols other than Zigbee. WirelessHart is apparently one of such that wins the industrial automation niche.

In this model, the chip vendors provide the platform in the format of radio IC and/or SoC, and are definitely in the no-losing position in terms of chip sales. The competition lies on silicon performance, logistics and engineering supports. The embedded software guys write protocols. Sounds familiar? Such model has resulted in hundreds of thousands applications that have been developed on iPhone and Android. Though 99% of apps remain unnoticed, few Angry Birds become the must installed apps by consumers’ choice.

In the context of what the technology allows us today, first developing a protocol is not that hard a mission on a given radio or physical layer with a focused objective. Second a protocol eventually becoming a standard requires somewhat luck, somewhat the wow factor like “it simply works” and somewhat good timing. Polar keeps WIND in house, so WIND never has the chance to become standard. Nike+ is a purchased protocol, so the protocol developer made a fortune. But Nike+ also keeps it close. ANT was developed by the then start-up Dynastream, who started licensing ANT before being acquired by Garmin; and Garmin looks like keep funding this business. Now ANT has become the de facto standard with multiple supporting silicon vendors, a growing ANT+ Alliance, and several cell phones. WirelessHart was originally developed by a UC Berkeley lab, and then marketed by the spin-off Dust Networks. It was later adopted by the Hart organization and became its wireless protocol. These are the prominent protocols that we know. There are plenty of others that have never got the chance to be noticed. But the invisible hands of marketing selection eventually let the best be the standard.

The fundamental disruptive point here is that protocol development is a software project that can be done by a separated group outside of semiconductor vendors, who is much closer to and knows the in and out of the actual problem to be solved.  A focused group does things more efficiently and better than a committee of people having multiple interests.
 
On the chip vendor side, it is important to focus on improving the chip overall performance, to be configurable on certain parameters (software radio), but not to lock down to one protocol. This will lower the risk and allow a chip platform a better chance to succeed.


4 comments: