According to Gartner, businesses alone are expected to use 3.1 billion IoT devices this year, 31% growth to 2016. Businesses alone will drive $964 billion in IoT hardware sales in 2017.
No doubt we will see a lot of interesting moves on the market. The beauty of IoT is that less and less local servers are required, everything can be clouded with minimal number of instances using multi-tenant software architecture.
Google is now also penetrating to the thriving market. Google launched Cloud IoT Core three weeks ago.
No matter what the actual application is there are some common things to consider in IoT devices and business such as:
- Protection against security threats
- Cloud connection
- Powering and power consumption
- The Cloud service
- Who owns and uses the big data
Google Cloud IoT Core hits all but third bullet on the list. depending on application requirements Google might make these trivial or not.
Protection Against Security Threats
According to F-Secure’s Harri Susi and other cyber security specialists the most essential thing is to have constant automated firmware update process running in the background for all IoT devices. It does not matter how small or simple your device is. If it’s running embedded software with Internet connection, there are always flaws just waiting to be found by the bad guys thus requiring to be fixed.
For many small vendors Google IoT Core might help to reduce time-to-market without compromising security too much. According to Google, there is a neat firmware push service built-in. It probably suites very well for many simple IoT devices. For complex system having mesh networking and devices updating other devices, it probably isn’t enough yet.
Having update push from could to all devices is of course just the first step. It only gives the possibility to provide security patches. But the most time consuming never-ending job is still there: monitoring your device security and building security patches. While IoT devices today are expected to last more than 10 years on the field this certainly means that lifecycle cost coming from support of every new device model is going to be significant.
Another important thing is to have secure and encrypted communication from your device to cloud. According to Google, devices running Android Things or supporting the Cloud IoT Core security requirements can deliver full-stack security, using certificate-based authentication and TLS 1.2.
And lastly, almost no-one can help you if your firmware is poorly written and proper testing has been neglected. A sad example was few weeks ago published news about Foscam camera security flaws.
Well, you must provide connection to your device somehow. The design is chosen by the requirements. Clearly we can have at least three different criteria:
- Limited vs. high data rate (e.g. temperature measurement vs. video streaming)
- Historical vs. real-time reporting (e.g. visitor count last week vs. security cameras)
- Unreliable vs. reliable connection (e.g. showing targeted ads vs. shoplifting alert to guard)
Data rate requirement you must solve by yourself. For very limited data rate, there are multiple interesting solutions arising, such as Sigfox. But you should then do some ridiculously compact update binaries, wonder if someone actually did it to commercial product through 12 byte payloads. Anyone?
Google IoT Core relies on standard MQTT protocol. MQTT is a lightweight messaging protocol for small sensors and mobile devices, optimized for high-latency or unreliable networks. Having standard protocol with high latency means that it could be an easy way to build new devices which are reporting historical data. But on the other hand, how does it respond to low latency applications?
Who Owns and Uses the Big Data
This is a topic that would need own article. But needless to say, this is the reason Google is going to IoT. Imagine all the information arising from the devices and connect that to all the existing map and search engine data in Google’s possession.
Predicting the Future
It is interesting to see chip vendors such as Intel mentioned as partners by Google in this project. Today we see such a bunch of simple and more complex IoT hardware solutions missing. Key is to connect the possibilities of cloud intelligence to smart hardware.
Today there is a lack for example of hardware level solution for redundant powered firmware updating devices running with coin cell batteries. Split partitioning and writing update packet in small parts while letting chip to go deep sleeping between packets could probably be hardware supported and reduce embedded engineering costs.
Different kinds of Cloud IoT platforms will arise and evolve during the next five years. Winner is the one who controls biggest number of devices and is able to dig precious information from the huge amount of data.