I agree with you David...most will send small volumes of data
Although we cannot possibly envision all use cases
It is very easy and low cost to get hold of Electric Imp stuff.....the dev kits are like $10 or $20 and the cards are $25. Personally I envision that these might be far more educational than Raspberry Pi boards.
As I see it (and I'm no expert I hasten to add) the IOT on networks will be much like SMSs on Cellphones. They take up a tiny fraction of usage but because they are small (data wise) you can piggy back a lot of them onto the network without degrading service much. Most IOT devices will only send small packets of data infrequently - examples like swithcing devices on and off will only take a few bytes.
Peter - would I be correct? Thanks for these articles - the Electric Imp stuff looks very cool, I'd love to get some to play with.
@Peter: A finite supply of IP addresses is the reason for IPv6. But we haven't run out of IP addresses anywhere near as fast as some assumed because so many connected devices *are* behind routers, where the address visible to the outside world is the router. There may be thousands of devices behind that. but they are the router's problem, and handled by NAT.
There may well be 50 billion Things connected to the Internet of Things, but the vast majority won't be publicly visible, and won't *need* to be.
One thing on Electric Imp's site was the motion that anything using their hardware needed to authenticate to their servers. But that's not an IP problem, either. If it's behind a router, what their servers will see is the router's IP. What they'll identify will be the unique hardware MAC address of their device. The authentication looks to be "once on power up and initialization", so the volume of traffic might not be that great. Many things on the IoT are likely to be powered up *once* and left on 24/7.
Save for authenticating to Electric Imp, I see *no* reason for the vast majority of devices on the IoT to ever communicate beyond their local network. Tell me why any of the things you mentioned might have to, beyond signing in to Electric Imp?
My scarce resource is WAN bandwidth. My connection is a 10Mbit/sec cable modem. On a normal basis, my desktop and SO's laptop are connected, but nothing we do saturates the bandwidth, so while my ISP offers higher bandwidth, I've had no need for it. The Internet of Things is highly unlikely to change that.
Would I see degradation if I approached the limit for the number of devices my router could handle? Simply allocating more local addresses would affect nothing. It would depend on the volume of traffic those devices generated. IoT devices would all be local traffic, and as mentioned, I expect that to be minimal. Should I see degradation, I can add hardware.
I still don't see the problem you envision.
You mention 255 devices can be connected to your low-end wireless router. That may be a limit imposed by an addressing system within the router but long before you got there I think you would see a marked deterioration in quality of service.
Well I have read predictions that with a few short years there will be 50 billion things in an Internet of Things. That is about 10 per person on the planet. But clearly there will be many more per person in the more affluent parts of the world.
Tires will talk to cars. Cars will talk to the road, to other cars, to roadside indfrastucture. Switches will talk to lights, to all domestic applicances, to smart meters and to the smart grid.
Basically many of things we have today will become networked. But then there are things we have not yet imagined that could be enabled by such a ubiquitous IoT, bandwidth permitting.
So every window is a dimmable solar cell that talks to the IoT. The list is potentially endless.
And although it might make sense for your things to communicate only within your domestic set up that is unlikely to be the way it will be set up.
The analogy is email. You send an email to the person on the next floor, perhaps only feet away, but that message doesn't go direct; it goes via the corporate server and probably out to some service provider before coming all the way back.
On the other hand MOST of these things will have very low payload requirement.
And in the cases of both Electric Imp and Neul this IoT activity is mediated through their servers. So the data always goes to cloud and back. They do this so that they can control the data and presumably extract value from it at some point in the future.
"However, there is an argument that if IoT is going to be truly successful the nodes will become so numerous that they would overwhelm Wi-Fi bandwidth."
Excuse me? How so?
I'm not sure what Wi-Fi bandwidth will be run out of. There are two finite quantities: the number of IP addresses provided by a router in a wifi hotspot, and the combined bandwidth devices connected to it will require.
My low-end wireless router lets me have up to 255 devices connected through it. How many people are likely to have enough connected Things to bump into that limit? What Things might those be?
And what sort of total bandwidth will those Things require? I'd be willing to bet most of them would use next to none, as what would be involved are essentially control loops, with command/response pairs, like when my A/C which has been turned off while I'm away to save power is told to turn on an hour before I get home to have things cool for my arrival. *That* could be handled over a 2400 baud modem line.
The vast majority of those Things will be communicating entirely within my local network once up, and won't need any connection to the outside world after being turned on and initialized, so I'm not looking at increasing the bandwidth from my ISP at a higher price to handle them.
I may be missing something, Please provide a scenario in which what you suggest could be a problem?
This sounds a bit like comparing apples to pies. Looks like ElectricImp is offering a unique integrated system focused on the UI and UX with old radio technology, while Neul has unique radio technology with old UI and UX. Perhaps they should team up.
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...