|Wireless Sensor Soup||| Print ||
|Written by Akiba|
|Monday, 03 March 2008|
With so many wireless sensor network protocols vying for world domination, I thought I would write a brief survey of the ones that are currently standardized or in the standardization process. Come with me and join in on the fun...
Okay, that's a little unfair since this blog is mostly about Zigbee. The reason is that a lot of my current open source work centers around this protocol (yes, I'll post more details later). Well, Zigbee is the 900 pound gorilla right now in the wireless sensor network space, however the cage is really small.
This protocol has been around since 2003/2004 and rides on the back of the 802.15.4 spec. Actually, a lot of the people that came up with the original v1 Zigbee spec were also on the IEEE 802.15.4-2003 committee as well, which is why you see references to the exact protocol in the Zigbee specification. I probably don't need to give much of an introduction to this protocol, since they've built up a big enough repository of intro material from all of their open house powerpoints . Suffice it to say that they came up with their own routing, transport, and application protocols targeted at the low power, wireless market. There's a lot of information at the Zigbee Alliance website , so I'd recommend a quick perusal of it.
6LowPAN is actually a specification that is under consideration by the IETF. Their main goal is to implement wireless sensor networks using TCP/IP (v6) instead of coming up with their own networking and application layers. According to the 6LowPAN group, this will allow the reuse of proven protocols to implement the sensor networking functionality instead of "reinventing the wheel". They will also have the added benefit (hopefully) of not requiring a gateway to interface to a TCP/IP network.
The challenge that they'll be facing is that they need to massage the TCP and IP headers so that they'll fit into the limited space of the 802.15.4 frame that they'll be riding on. On top of that, TCP/IP was designed for data intensive communications so they'll need to figure out how to handle the fact that the nodes will be sleeping most of the time into the TCP/IP protocol which assumes that the nodes are always live.
Actually, whether proper or not, I have to give credit for the idea of TCP/IP over sensor networks to Adam Dunkel and his guys at the Swedish Institute of Computer Science. They were doing TCP/IP over sensor networks for a long time and in fact, already have their stack going via the Contiki/uIP suite. You'll never find a smaller OS or IP stack. 2 bytes per thread and a 6kB stack!!! You should check it out!
This is a standards body which was started by Zensys and is trying to promote standardization around Zensys' proprietary Z-Wave protocol. Unfortunately, the spec is not publicly available, and you have to pay $2500 to become a member in order to get access to it. I can't even afford to pay the money for Zigbee membership... and I'm developing software for it.
It'd be hard for me to call this standardization since its a proprietary protocol using proprietary chips. If you want more info on it, you can read the flame war starting here and continuing here . In the meantime, as biased as it sounds, I'm only interested in open protocols (ie: not controlled by one company). Okay okay...you got me. Zigbee is mostly open, except for the membership thing. Same for USB, though. Yada yada...
ULP Bluetooth (Wibree)
Not wanting to miss out on all of the wireless action, or perhaps because they're finding wireless headsets getting a little bit boring, Bluetooth decided to get in on the wireless sensor networking action as well. ULP Bluetooth (Ultra Low Power for those that are acronymically impaired) was the brainchild of Nokia, and was given the fascinatingly strange name of Wibree . I found it hard to believe that anybody could outdo Zigbee in the naming category. The technology has since been adopted by the Bluetooth SIG as part of the Bluetooth specification and has changed its name to the very unexciting ULP Bluetooth .
The protocol uses the same radio as Bluetooth, however the software will be different. There are two modes, standalone and dual mode. Standalone contains the protocol stack for a ULP Bluetooth only implementation, whereas dual mode would allow a standard Bluetooth device to communicate with a ULP Bluetooth network. According to Nokia, the main objective is to be able to allow Bluetooth enabled phones to communicate with ULP Bluetooth enabled wireless sensor networks. Yay! Now I can call my toaster to turn it on.
And now we get to the industrial wireless protocols. There are basically two protocols that are competing to be top dog in the industrial-only space and one of them is SP100 ... I mean .... ISA100.11A. Yes, its the same ISA behind the ISA bus that you used to have on your old 386 PC.
ISA100 will be using the IEEE 802.15.4 protocol for the PHY and transport layers and 6LowPAN for the networking layer. They will create their own routing and application layers. Some interesting features of the ISA100 protocol is that they will have frequency agility which means that the whole network will be moving around on different frequencies in synchronization to avoid interference and they will have a tunneling mode where you can have different protocols running on top of ISA100, ie: Profibus, Fieldbus, Hart.
This is the second of the big industrial wireless protocol standardization efforts currently taking place. wHART is not a full industrial sensor protocol. It's more of an extension of the venerated HART protocol which is still popular in many industrial networks. However it does have many industrial strength features: mesh networking, frequency agility, 802.15.4 MAC/PHY. Wait a second, you say!! That sounds like ISA100. Yes it does, and they are currently investigating ways in which they can collaborate to have a unified protocol.
Well, that about does it for me and my brief survey of the standardized wireless sensor network protocols. I'm gonna get me some shuteye.
|< Prev||Next >|