By Harry Forbes, ARC Advisory Group
Two very different events this week showed a common thread of the Internet Protocol (IP) creeping into wireless sensing. First at the CTIA show, amid the glitz and distractions of Las Vegas, Arch Rock Corporation released their PhyNet routers for sensor networks. Half a world away, the ISA Wireless Compliance Institute and 3 suppliers demonstrated a prototype ISA 100.11a application at the 2008 Industrial Wireless Conference in Chongqing, China. What did these events have in common? Both are bringing wireless sensors closer to the Internet.
Arch Rock’s announcement of a new wireless sensor network (WSN) router is a departure. What Phynet does is break the conventional WSN gateway into 2 pieces: a router and a server. The router is a relatively simple hardware appliance that must be located near the sensor network. The server is purely software, and can be located anywhere on the Internet, which is a nice benefit.
Arch Rock is huge supporter of RFC 4944 (which one might call “the protocol formerly known as 6LowPAN”). Though it lacks a catchy name, RFC 4944 is an implementation of the Internet Protocol that can be used by the most popular low power sensor radios. These radios use IEEE 802.15.4, which was specifically designed not to work with IP, but RFC 4944 lets these radios use IP anyway. The advantage of IP is it gives a sensor network full interoperability with and exactly the same model as the global Internet – that model being “smart” edge devices and “dumb pipes”.
Using IP, the set of services a wireless sensor can provide becomes unlimited, at least unlimited by the protocols. If you want to develop a field sensor that can send e-mail or host a web page, or run its own OPC UA server, then have at it. Nothing (except a dead battery) will stop you. You can create a secure IP connection from the wireless sensor all the way to anywhere else on the global Internet. Another real value is extensibility. Over time, as you add new sensors with additional features, you won’t have to modify any gateway functions, because there is no sensor network gateway, just a lowly IP router. The Phynet router keeps on “dumbly” routing IP packets the same as always, clueless to any new services they support.
Another advantage of using RFC 4944 is the ability to create high availability and scalable sensor network configurations using the well-proven techniques that are used on the Internet. The simplest technique is just to add multiple routers and routes. The sensors use these multiple routes to bypass a failed router and reach the server. It is far more difficult to engineer redundancy in “heavy” WSN gateways.
So now let’s travel (in our imagination) 7,100 miles from the Las Vegas strip to Chongqing, China. We are at an industrial wireless conference co-located with the ISA 100 standards meeting, and there is not a slot machine in sight. What does this meeting have in common with Las Vegas? Very little in common with Las Vegas, but quite a bit in common with RFC 4944.
The first ISA 100 specification to be balloted will be ISA 100.11a (as you can see, 6LowPAN is not the only ungainly name in wireless standards today). ISA 100.11a is designed to address low-power wireless field devices, such as wireless field transmitters. As drafted, ISA 100.11a will use RFC 4944 for these sensor networks, giving it a lot in common, technically speaking, with the Arch Rock folks partying in Vegas. Three suppliers collaborated to produce the prototype demo in Chongqing: General Electric, Honeywell, and Nivis (an Atlanta-based systems integrator familiar with sensor networks). So in the same week, on 2 different continents, people were watching companies demonstrate IP on sensor networks.
If suppliers are now showing IP sensor networks, what about WirelessHART? WirelessHART uses the classical WSN design, with a real gateway. There’s no IP anywhere on the sensor network. In WirelessHART, IP begins at the gateway. So why are some folks (including some very smart folks) placing their wireless sensor bets on the WirelessHART square?
The answer is that they are betting on long term reliability, meaning low power consumption and very long battery life. Remember that wireless sensors must be low power devices. Wireless field transmitters need to operate for very long periods on their batteries. Nobody wants to service an installed field transmitter at the transmitter’s location. That’s precisely the reason (wired) HART was developed back in the 1980s. If a wireless transmitter can’t provide a long unattended service life, manufacturers don’t want it. The important question is can they get the same service life using WirelessHART and RFC 4944?
That is where you find a lot of disagreement. The WirelessHART folks use a protocol called TSMP, which was developed by a nice, very smart Berkeley professor with the objective of conserving battery power as much as possible. What the WirelessHART folks don’t point out is that they chose TSMP at a time when IP on sensor nets was considered a fool’s errand.
Internet Protocol was not developed for low-power or duty-cycled devices. But, the developers of RFC 4944 (one of whom is also a nice, very smart Berkeley professor) knew that RFC 4944 had to work well on low power devices. IP supporters claim that the two are now close in terms of energy consumption, and getting closer.
Adding yet another twist, there are more nice, very smart folks in Honeywell’s corporate R&D lab (sadly, we can’t all be Berkeley professors) who insist that they can and will win the long life/low power wireless sensor game. That is why Honeywell is heavily promoting the idea of adding other radio types to ISA 100.11a Release 2. The ISA 100 committee voted to use a single radio in this draft, but has committed to consider others for Release 2.
The important thing to remember here is that products with a superior match to customer demands usually succeed in the market, and process manufacturers do not want their technical personnel visiting field transmitters. Watch what length of unattended service life various commercial products offer. Proprietary products have been around a few years now. WirelessHART products are reaching the market now, while the ISA 100 and RFC 4944 products remain in earlier stages. Regardless, ARC believes that the measure to watch is how these products match up in terms of their unattended life and reliability. This will greatly on how well each sensor technology preserves its battery.