posted by Bryon Moyer
You may recall that, some months ago, we wrote about SIGFOX. As a quick review, SIGFOX is installing a completely new cellular system optimized for low-data-rate IoT sensors rather than for voice and Youtube. As a French company, they started the build-out in their European neighborhood, but they are supposedly starting their North American build-out now.
Of course, for this to be successful, you have to have devices that can talk to the network. And if the designers of those devices have to cobble together their own discrete circuits or – worse yet – design their own custom SoCs, well, that’s a pretty big barrier.
It would be so much easier if there were a merchant-market chip available that implemented the SIGFOX protocol – both PHY and stack.
And now there is: Atmel has announced the ATA8520 transmitter – the first SoC to pass SIGFOX’s qualification tests. It consists of the RF block, baseband processing, and an AVR microcontroller for executing the SIGFOX stack and other control functions. Data, as well as other control functions, is entered via an SPI port. The chip has an “OFF” mode that’s not quite off – more like a sleep mode – and reduces current to around 5 nA.
(Image courtesy Atmel)
The ATA8520 would typically be driven by a system microcontroller or processor; that processor would be freed of any of the details of SIGFOX transmission (traded off for SPI).
Now… all we need is a receiver to listen to all the news being broadcast by the transmitter…
You can read more in their announcement.
posted by Dick Selwood
The phenomenon that is Raspberry Pi has moved up a gear. On Monday (2nd February) the next generation of the credit card computer was announced. It has a new Broadcom BCM2836 application processor, based on the ARM Cortex-A7 quad-core CPU which they rate at 900MHz. Memory has doubled to 1GB. This is claimed to make the new Raspberry Pi 2 better than six times more powerful than the first generation Raspberry Pi Model B+.
There is full compatibility of both hardware and software with the previous model, and it has the same price of around $35. The board layout, multimedia subsystem and peripherals remain fully compatible with the Raspberry Pi Model B+, including the 40-pin GPIO (General Purpose Input Output) connector, four USB ports and an efficient switching power supply
It will run a range of operating systems, with Windows 10 promised. To get it working you need a keyboard and screen and a 5V 2 amp USB power supply.
Since its launch around 3 years ago, 4.5 million Raspberry Pis have been sold with 1.5 million in the hands of the original audience, school children. Many of the rest are with hobbyists and there are companies using them in commercial products.
Two distribution companies, element 14 and RS/Allied Electronics are both selling them around the world and both host community sites.
Later this month I will be looking at the way Raspberry Pi, and other boards with their associated communities are opening up computing in a whole new way.
posted by Bryon Moyer
Zigbee rolled out their latest revision late last year: 3.0. I sat with Zigbee Alliance CEO Tobin Richardson right as he had arrived back in country from Europe. Alas, I didn’t manage to find a way to turn any jet lag to my advantage. (Gotta work on that…) He summed up the main contribution that this new revision brings: unification of all the profiles. The intent is to further, how to say, standardize the standard.
The original profiles are still there – it’s all backwards compatible, but there is now what could be called a “super-profile.” It’s not a layer over the existing profiles; it’s simply a way to make it easier for developers to develop or reuse code for more devices and to promote interoperability.
That could be a good thing; I’ve been generally asking developers that I run into here and there about their reaction to Zigbee 3.0 (and Zigbee in general), and one of the concerns coming up was that, while a given manufacturer might make interoperable Zigbee devices, it was common for devices from different manufacturers – all of whom designed to the standard – not to interoperate. (To be clear, this was their experience, it’s not a scientific study, so this should be considered anecdotal. More on this in a minute…)
Beyond here, the Zigbee Alliance is going further to define what I call “business objects”: a set of standardized semantics for various different devices. This would unify the properties and methods across all like devices being addressed via Zigbee.
Almost coincident with this, Greenpeak announced a new RF chip that supports Zigbee 3.0.
I still struggle with the role that Zigbee will play in the Internet of Things (IoT). Zigbee is well established in industrial applications, or so I’m (again, anecdotally) told. The strength of its meshing capability has served it well here. The bigger question is what will happen in the home.
I’ve noted before that Zigbee is, by regulation in the US, in smart meters. Since then, Comcast has also brought Zigbee into its set-top boxes. Mr. Richardson’s vision is that this set-top box (and others like it) will form the hub for the entire home. Phones will communicate with the hub via WiFi; the hub will translate to Zigbee for communication amongst the devices.
Others suggest that Bluetooth’s presence in phones gives it an edge (as I’ve argued before) because it allows communication directly with a Thing without going through a gateway. Mr. Richardson speaks to a large Zigbee installed base – even in the home. I don’t know if that base is being actively used (or whether it’s effectively dark silicon to date), but Bluetooth aficionados counter that its installed base – by virtue of being in phones – is far higher.
As I mentioned, Zigbee has traditionally been able to turn to meshing as a key differentiator. That clearly has value over the longer distances in a factory or warehouse; it’s less clear that it’s an enabler in standard homes, where distances aren’t great. (OK, maybe it’s useful in a zero-lot-line McMansion for getting from the basement theater to the third-floor music conservatory…)
Then again, that advantage was, at least in principle, was neutralized last year with CSR’s Bluetooth meshing capability. And just last night, Nordic announced a mesh lighting offering. So even if meshing is useful in the home, Zigbee is no longer sole provider.
Mr. Richardson doesn’t see this as a war; he sees it as three standards having three roles:
- WiFi for internet media content
- Bluetooth for wearables and headsets and such
- Zigbee for low-power, low-data-rate wireless sensors and controls
But that focuses on Bluetooth’s traditional point-to-point architecture, which changes with meshing. Greenpeak wrote a whitepaper suggesting that, while the meshing may be gone as a differentiator, that Bluetooth should still pretty much stick to its traditional point-to-point role and cede home control to Zigbee – and specifically not start a standards war. They even suggest the possibility of Zigbee/BLE combo chips. (Although, when I pressed this issue, they clarified that their logic points to this as a possibility – they’re not specifically saying that such chips are in the works.)
There’s yet another former differentiating aspect that’s been – or is being – neutralized. WiFi used to have the advantage of being the only IP-based standard of the three. And IP didn’t generally look promising for low-power nodes.
But the advent of 6LoWPAN has made IPv6 accessible to constrained nodes. And Zigbee now supports IP packets; Bluetooth is said also to be readying support for IP. One of the benefits here is that the IPv6 IP address can then be used everywhere, from small wireless nodes to large home devices to anything else. No separate addressing will be needed to identify devices on a non-IP internal home network, and no translation will be needed to convert that into an IP address when it goes out to the cloud. Even for WiFi, no more network address translation (NAT) is needed to convert from an internal IP address to a public one.
Zigbee’s coup in getting Comcast’s buyoff does give them a good position there, but from a strategic standpoint, device-makers aren’t all keen on Comcast inserting themselves into the service. The concern is that Comcast will try to further monetize access to the cloud without really offering anything in return (other than bandwidth, which is already paid for). It’s kind of a gatekeeper thing, and as I’ve noted before, what’s good for the gatekeeper isn’t always necessarily good for the consumer.
So IP support is disappearing as a differentiating factor. And meshing is gone. And set-top box incumbency has yet to prove valuable. Leaving… what as a differentiator? The basics: cost, power, performance, and range. Typical ground war stuff. Which is a function not only of standards, but of implementation.
The other key factor is preference: both designer and consumer. It becomes a marketing war. Mr. Richardson noted that they may do some brand-awareness building in the consumer space, since Zigbee is more familiar amongst geeks and hobbyists (my words, not his) than retail salespeople and lawyers and, well, just regular folks.
By the way, we debate these three high-profile standards here, and yet there are other ones in play as well. I’m surprised how often Z-Wave comes up. It’s a proprietary protocol, but apparently interoperability is quite good (again, anecdotal). But it’s an open question whether proprietary standards will have staying power in the face of open standards.
And new IoT standards keep showing up, each claiming to solve some IoT problem better. We’re going to look at those soon. Things are getting messier before they get cleaner, although some of those standards ride above the levels that we’re discussing here, so they don’t necessarily compete with WiFi, Bluetooth, and Zigbee.
Feel free to add your own thoughts on these three standards in the comments box below. Perspective, even if anecdotal, can always be enlightening.