What about a system which, instead of requiring an Ethernet connection and an internet connection, uses the cellular network for data transmission, for example SMS? This is a simple, two-way, text-based messaging system that should be relatively easy to take advantage of, except for the cost.

I suppose this is why people haven’t done this type of system more. I know this must be a thought process that Ambient went through a decade ago. And countless others, I’m sure. I guess that I need to accept that most everyone interested in the type of device I propose, has an internet connection, so this is free. And on the other side of the coin, there is as of right now, a commercial cellular model in the US where, though everyone has a phone, voice minutes are free but data bytes cost a fortune. Silly, and completely unnecessary, but true.

Cellular providers could take over the world if they got their minds around offering their networks for data communications. But, at this moment, I don’t think there is any way to support this. Until such time as I can send and receive data for free, or buy a data-only plan in addition to my voice plan, with an additional SIM, for my device network, forget it.

Is a gateway just a gateway, or can it be used as a standalone a device? If the latter, are all devices Ethernet-enabled? This would be a waste of hardware, right? If not, are the devices with Ethernet also devices on their own, allowing someone to have just one device? Would they ever have just one device?

And if they did, would this be better than a USB device? It would work even with the computer down, right?

I can see that I should do some fundamental architecture decisionmaking. I need to rank all the desired features, and see what is the best solution both short and long term.

Desired features, advantages and disadvantages

1. plug and play, no drivers: Ethernet, or USB as flash drive
2. USB: no power conn needed
3. Ethernet: no USB; power conn needed
4. radio: all devices have
5. local computer control (no internet needed): USB or Ethernet
6. Ethernet-internet control (no local computer needed): Ethernet

Proposed architecture

1. Ethernet
2. Simple power conn (switching supply for efficiency)
3. Radio

Could all devices have both ethernet and radio? This would mean, from a both Zigbee and an aiosphere network perspective, that they should be able to be both net master/gateway.

I know from looking at some of the common Zigbee modules that this is not as easy as simply having fixed roles. So, for my first system draft, I think I will take the easy route and have some devices, which cost more, which are gateways, and others which are simple radio devices.

Next decision: do I make the gateway be a device, itself? On the one hand, this is attractive, as in my system there are no ugly appliances, only interesting and useful objects. On the other, it would mean that I would either need to force all users to have one particular type of device as their gateway device, or offer a range of devices with gateway capability.

To have a single gateway, which does nothing else, would tend to help avoid a situation where a user would plug more than one gateway in (because they both have desired features). Which would avoid my having to have some complex negotiation between two gateways, to decide which would act as a gateway and which would act as a simple device, ignoring its gateway capability.

So how do I connect a Zigbee network to a PC and the internet? MaxStream sells Zigbee USB radio modems ($250), but they are way too expensive. New Micros makes a carrier board for MaxStream’s Xbee modules ($75), which more or less fits the bill if you don’t mind the bulk, or plan to stick with the MaxStream system.

A couple of key-style USB dongles out there from different vendors. Integration Associates makes a Zigbee compliant dongle ($75) that probably could be interfaced to a module network from many of the vendors I have looked at so far. They apparently wil provide drivers for Windows, Mac and Linux.

I am not sure how easy it is to seamlessly transition from the simple proprietary networking methods provided by the module vendors to a Zigbee standard network. To be safe, I will probably selecting one of the module vendors which offer simple networking APIs then building my own gateway using a USB-serial solution like FTDI’s FT232R, then roll a serial-TCP driver for each platform.

At this point the two strongest contenders are Jennic and Meshnetics. Both make modules which integrate a microcontroller which is used for both networking and the user application. Meshnetics uses an Atmel micro and an Atmel radio, and provides software libraries for networking which one compiles with their own code. They offer modules with external and chip antennas. Network control is done using AT-type commands.
Jennic uses their own 32-bit processor with built-in radio, ROM and lots of RAM. The micro’s ROM contains the networking code and a bootloader, and user code is stored in serial flash and loaded on startup. All micro and network control is done through API calls.

The Jennic system has the advantage of offering a nice volume path as they sell both $25 modules and $5 micro/radios. The Meshnetics offerings are unknown at this point, they are out of Russia but anticipate Digikey distribution in the next month or so. Still waiting on pricing.

Meshnetics also offers an ethernet gateway, which is an interesting idea. I would probably prefer to do a USB gateway, but the ethernet approach would certainly ease my FireFox implementation! We’ll see how much it costs. I guess it will be prohibitive.