I was solicited on Quora to answer some questions about IoT, and I decided to do it knowing well that I’d repost them here. Here’s my take on a handful of questions related to contemporary (2015) IoT trends and issues.
1. What kind of work are you in?
I am CTO of an IoT startup called Haystack Technologies. I invented a low-level wireless protocol called DASH7 that has about 100 users now, mostly in europe.
2. Have you developed or worked on any IoT projects?
HayTag, DASH7/OpenTag, several other hardware projects. All for low-power wireless IoT.
3. What products or solutions have you used in your IoT projects?
Electronics and POSIX, also Apple’s iOS tools. For data aggregation web services I know we’ve used Socket.IO and heroku in the past, but the backend is not my focus. Web data backends change pretty quickly, and there are tons of open source coders who seem to scramble to create APIs for all the latest technologies. I don’t think it matters that much which one you choose, because you’ll probably want to change it in 12 months anyway.
4. Most of the research on IoT only speaks of lack of interoperability and open standards and ease with which one internet enables things. Is this something you agree to?
Interoperability is there if you want it. So far, most business plans for IoT don’t seek interoperability, in fact they don’t want it. They want to own the devices and force users into their websites. There isn’t really a good monetization strategy yet for IoT apart from selling devices and data, although there are some players like SigFox who want to create a sort of IoT cellular service — maybe that’s next.
Pretty much everyone and their cousins are trying to set up “cloud” services for IoT data aggregation and analysis. This market is already overbooked, in my opinion. To succeed here, you need deep partnerships with device makers and distributors, and probably a really good story as an IoT certificate authority, for which there are currently none.
5. Do you think there is a need for open standards based solution in the IoT market?
Yes, but these already exist. DASH7 is a wireless stack. In your case, you probably care more about the app layer. CoAP is open, so is MQTT, and others. What remains is a lean public key auth standard.
6. Does limit of protocols (Wi-Fi, RF, Bluetooth etc.,) supported by an IoT solution impact your decision to buy one product vs the other?
Wireless IoT first separates into mains-powered vs. battery-powered. WiFi is ultimately a mains-powered solution. Bluetooth and some others (like DASH7) are battery-powered. Battery and Mains solutions rarely compete. In the battery-powered category, next, there are bursty/multicast vs. stream/point-to-point. Burst-oriented systems are like DASH7, Thread, maybe some others. Bluetooth is point-to-point streaming. The part of wireless IoT that hasn’t really taken root yet is the most game-changing part: the low-power wireless with bursty multicast. These technologies are what will make “fog data” happen, which will be disruptive to many contemporary big-data analytics operations, but also complementary to many others. There will be an arms-race to have lower-latency access to the endpoint itself, to provide the most “real time” IoT data.
7. Do any of the IoT solutions address the power needs for devices in remote areas?
For remote areas, you need long range. LoRa, SigFox, and DASH7 all support long ranges and low power. There are lots of proprietary options as well.
8. Do you think building or Internet enabling existing things (ex. flow sensor, garage door opener, soil mineral sensor…) is costly?
Doing this wirelessly tends to be cheaper than laying-down copper wire. If you want building sensors, wireless is the cheapest option. This market is better in Europe than in USA, for a long list of reasons, but mostly related to the way building contracts are done there. EnOcean has a lot of business here.
9. Do any of these IoT solutions address reliability, failover, self-healing networks capabilities?
The best way to have a robust IoT network is to have greater signal strength and range. LoRa and DASH7 use this strategy. Thread and ZigBee use automated wireless meshing, which is more complicated (and therefore less robust), but it seems to work well in environments that are relatively static (i.e. stuff isn’t moving around much). In more dynamic environments, the network needs to be reconfigured very often, which can be a problem for low power devices.
10. How are you tackling the reliability issues present day when building your IoT products?
Quality engineering. Seriously. If you’re wondering why the Bluetooth tracker you bought on kickstarter isn’t living up to expectations, it’s because it wasn’t quality engineered.
11. Do your IoT projects/Products utilize multiple protocols? Ex nodes are communicating via Wi-Fi, RF, Bluetooth?
A lot of devices use WiFi or Bluetooth to tether to home gateways or smartphones respectively, and then they have another system to communicate with the endpoints. This sort of device is actually called an “edge router.”
12. Are you following any standard data format (xml, json) to send data logged by your sensor or device to the back office server?
IoT works better with binary data. CoAP is a decent protocol standard. DASH7 also defines a handful of binary data formats. There’s also ISO 21451-7 which is a sensor data standard. Once this gets to the data center, 99% of the hard work has been done, and the software at the data center can convert the IoT data to whatever bloated data format you like.
13. How are you powering devices in remote areas? Are you relying on solar power or battery power?
Yes, solar and batteries. Solar is actually quite difficult to engineer and it rarely makes sense economically for consumer products. For industrial products, however, the added $2-5 of a solar charging system can save lots of time and money in device management (namely, the act of replacing or servicing dead endpoints).
14. If you are relying on battery power for your remote devices, are you taking into consideration battery life based on consumption?
For a well-engineered DASH7 device, an AA alkaline cell can power an endpoint for years. DASH7 is a little bit better than everything else in the 2-way wireless IoT technology sector, but quality low-power engineering tends to make a much bigger difference than the standard you’re using.
15. Do the security features offered by the present solutions enough for business critical IoT Product development, ex -Industrial Automation..?
Security is, at present, an area that needs more attention. Frankly, the only standard I can think of with sufficient low-level encryption and authentication is DASH7, which can encrypt beneath the MAC address. It uses a streaming cipher called EAX. Everything else seems to use CCM as a low level encryption, which has all kinds of problems. Namely, MAC addresses tend to be unencrypted, decryption can’t begin until the packet is complete, and quite a lot of data padding is required. These create vulnerabilities and waste huge amounts of energy for low-data-rate protocols.
16. What is your view on how the sensor data is being made available for use by different applications?
Sensor data is mostly delivered by proprietary formats, which is a problem. More companies should adopt ISO 21451-7 or insist on it.
17. What about over the air firmware upgrades, how easy it is for you present day to upgrade end node firmware?
It’s not really a big deal for any of the IP-like stacks that can manage a UDP transport, and thus support TFTP or similar protocols. DASH7 and Thread are two examples of such. The major downside is that OTA upgrades requires extra memory on the endpoint. For cost sensitive endpoints, this is not always desired. Before long, I expect OTA transfer of apps and applets will be a required feature for most IoT endpoints, though.
18. Do you think present day IoT solutions offer you extensible end node modules that let you plug in any device with much ease?
No. There is really no such thing as plug-and-play in IoT yet. It will happen, but it will take time. Maybe by 2020 when the market for IoT devices matures a bit. A lot needs to happen, and then standardization needs to follow.
19. What you think in your opinion next generation IoT solution should offer at bare minimum?
Ideally, the next generation of IoT devices can disrupt the cloud-silo business model. It will need better data interoperability and authentication against IoT certificate authorities. The goal here is to prevent a single entity from gating the access to the endpoints. “Cloud” as an industry is really about the emergence of internet servers as a commodity. The IoT “Fog” should follow the same logic, endpoints as a commodity, and to get there we need alternatives to today’s state-of-the-art, which is a heavily proxied (siloed) hub-and-spoke transaction model where the internet client never really interacts with the endpoints.