Chicago, IL -- As is typical before a fall, the build-up to the demo was good: Wade Patterson, president and CEO of Synapse Wireless, used his opening remarks at Wednesday’s ESC Chicago keynote address to outline how a wireless embedded network OS has fundamentally changed the game with regard to ubiquitous wireless connectivity, thus enabling 27.6 billion devices to talk to each other, potentially via 140 characters or less via Twitter.
Then David Ewing, Synapse’s vice president of engineering, came up to start to the demo.
In fairness, what happened next had nothing to do with Synapse’s capabilities or anything it had direct control over. In fact, the Synapse portion went really well, despite the challenges embedded wireless networking and control faces. And those challenges are many and varied.
”You were told this was easy,” said Patterson. “Just load the stack, it just takes 64 Kbytes of flash and no time to design.. but 10 years into this networking of machines and wireless was still really hard.”
According to Wade, the amount of memory required is at least 256 Kbytes of flash instead of 64 Kbytes, and two processors might still be needed, one for the stack and one for the application. Then it may still take up to two years to develop the system and even then it may not perform as expected. “Something had to change,” he said.
For Patterson, that change had to start at the root of the problems: the stack. “The stack focuses on the network, all the application layer is your problem,” he said. “The stack is also daunting to use. You need a networking background, have to read a 500-page manual, you’re locked into a process and physical layer and you have to decide your topology early on, in terms of the right nodes in the right place.” On top of all that is the extensive debug process, he added.
The solution, according to Patterson, is a wireless embedded network operating system, though he didn’t say so specifically, that OS is the Synapse Network Application Protocol (SNAP), a combined stack with application that can run on 8-bit processors with 40 Kbytes of memory. Such an OS, he said, allows the developer to focus on the application and actual networking, versus getting it to work. It also allows easy code writing with no compiler and over-the-air application code updates to speed debug. “The application code is also cross platform and hooks up to the Internet easily,” said Wade.
Armed with a networked OS such as SNAP, Wade said engineers are ready to truly enable machine-to-machine connectivity, a market that according to a recent Harbor Research report could encompass over 27 billion devices. In fact, it may even make sense to connect to Twitter.
Harbor Research puts the combined smart systems scope at upwards of 27 billion devices.
Click on image to enlarge.
While Wade admitted that at first glance that might seem crazy, he pointed out that the needs are similar:
- You need to hear from your machines if they have a problem
- You need to hear from their [your customers’] machines: “they’re your ‘friends’”
- Machines to get info from other machines
- Need to talk like browsers talk to websites—without firewalls. Access-on-install
- Need to be location agnostic
”Seventy five percent of your customers’ machine-to-machine (M2M) applications are Internet enabled,” said Wade. And with that, he introduced David Ewing to perform the M2M demo.
At first, all went extremely well. Primarily because it was point to point, with no external networks required. Ewing showed how to quickly set up and wirelessly control a DMX RGB stage lamp using SNAPLink458_9 and remote procedure calls (RPCs) over the air from a laptop with a basic 802.15.4 2.4-GHz radio. He ran the script and the lamp duly went from red to green to blue.
Ewing then quickly set it up a direct M2M network so that a temperature probe could change the colors as it went from ambient, to coffee to water. Again, it worked flawlessly, showing the ease of use of the OS for setup and running a basic M2M control function.
However, it was when trying to link to Twitter via the theater’s wireless local area network (WLAN) that things went downhill fast. Aware of the potential for disaster, Ewing referred to Steve Jobs’ fiasco with his WLAN when debuting the latest iPhone last week. Sure enough, the WLAN connection didn’t respond right away. When the local WLAN did finally work, the biggest issue with a Twitter-based M2M feed showed itself: Twitter responded with “Twitter is over capacity”.
So, while Twitter may indeed work for M2M communications, and SNAP may be a quick and easy way to do it, those 140-character-or-less updates had better not be for mission-critical monitoring applications, such as medical or infrastructure monitoring.
That being said, Wade did point out that for such mission-critical applications, a database can be set up in ‘the cloud’ such that data coming in can be stored and accessible by customers from anywhere. So, while M2M over Twitter may be interesting, it’s not necessary, and in many cases may not even be desirable, especially if a lot of back-and-forth is expected.
According to Ewing, if Twitter gets more than 150 messages per hour from a single source, it will shut down that source.
While Twitter-based M2M is interesting and may have certain non-mission-critical applications, the larger picture was succinctly brought into sharp focus by Wade’s parting question: “So, as the incremental cost of adding wireless becomes free, how is it going to change your products – and your competitors’ products?"