This is part two of the Internet of Things track. It was fun to follow this track leading up to my talk on OTT this afternoon.
First was Henning Schulzrinne of Columbia University (and former CTO of the FCC) speaking on the Internet in the Small. Just getting a handle on what we are talking about is still difficult. Constituencies need to talk to each other but the languages (both machine and human) have not converged.
Overview: Most of the things are not exciting (parking meters), the network part is not that interesting (don’t really need a new web), it’s a software thing – multi-devices services, APIs, IFTTT, SECE (interesting things are where we get devices talking to each other), what are our obligations as entities playing in the IoT space?
All of the tags (M2M, smart city) all mean the same thing. Dates back to late 90’s when the RFID tag came out. Now these sensors are networked and can control physical or electrical objects. It’s not just about fancy thermostats and $200 doorbells, doesn’t always use cell network, isn’t always energy/cost constrained, doesn’t always use small microchips, is not always large businesses (many small to mid companies)
There is a natural evolution from mechanical to electrical to connected. What are the key enablers – Mature internet protocols, widely available cellular, unlicensed spectrums, Analytics, cloud hosting, cheap chips. All of these are required to have widespread IoT. Until a few years ago this was not the case. Have moved from Supply chain helpers, to more verticals markets, to ubiquitous positioning, now moving into a more integrated world with apps/things integrated.
A particulate emphasis will be on environmental sensing – multipurpose sensors. IoT = cheap microprocessor and cheap wireless device.
Where does IoT make sense
– automating manual data (health, car, etc)
– Remote maintenance (vending machines, appliance)
– incorporate additional information (thermostats, light switches, traffic lights)
– Software Defined Mechanics (locks, light switches)
– But where does it solve more than first world problems? Such as commercial maintenance savings, in home customizable assistive technology.
Three dominant cases of a wireless IoT models:
1. thing to thing (like vehicular)
2. thing to proxy (gateways, hubs, hubs in vehicles)
3. thing to internet (direct connection to 4G networks)
Merging islands into ecosystems where things are able to discover and talk to each other. Very very limited now. Interesting applications are shared readers (semi private or open sensors and actuators). The needs of the devices differ dramatically – some need to talk once a day some need to send many times a second. Will have cellular and local connected devices.
There is a proliferation of IoT connectivity technologies. What are some of hey he challenges? How do I attach and authenticate a device to a network (what are the credentials)? What do we do for the license side (multiple connections), from $50 to less than $1 a month. How does it charge? Can it be deferrable like Whispernet. For devices that are shared how can you authenticate.
Should you have phone numbers for each? Can’t really support everything. Will now have dedicated application specific sensors and protocols. Pretty clear that we won’t have enough spectrum to dedicate. Will have to share spectrum.
One of the principle things we have not solved is the discovery and access control. Some type of world read generically. Ability to delegate access. Currently all one off solutions. Enrollment is an issue – have to bring in 1,000 sensors when a building comes online.
Before this gets out of hand, we need to implement some best practices. How can we make sure these millionsl of devices behave so they don’t cause problems. Security is an issue – need to figure out how to have kill switch, be updated securely, can’t trust random access points, can’t have special spectrum.
SECE has been trying to build a smart object personal cloud. Glueing together many different systems. Exploring many models – got pretty technical here but looks like a lot of cool ideas.
Internet of things is a natural progression – the cool things that will make a difference will be boring and not written up. Network is relatively easy – security and software are much harder. Need to reach beyond connectivity and software islands.
Next was Jerome Schang of NXP speaking on securing the internet of things. He spoke via Skype video via San Francisco as he could not get in due to flight issues. NXP has a lot of customers that reach out to them after security breeches. After the morning, the question that is left is how do you separate out out malicious actors. NXP is a leader in security (banking, etc). They are trying to leverage in IoT.
What are we trying to protect:
– Avoid Service Interruption
– Revenue Protection
– Privacy and liability protection
– Tamper protection is the answer
Cyber Security solutions in two categories – Hardware and software. Examples come from the pass six months looking at customers that have been hacked and how they protect them. IoT is being driven more and more by things getting connected. New attacks will be from some of these new things.
He will split IoT with and without IP addresses. Biggest challenges are from objects that were not built with security in mind. cyber Physical systems, Embedded systems a and physical items use the internet of services. The smart electricity grid has created a different kind of access control from a supplier and customer perspective. IoT is already being hacked – used a baby monitor hack. Vertical variances need adaptive security. He showed many cases of hacks from low level to very famous ones.
Security needs to be designed into the product and be adaptable. Hardware security provides end to end so that no spoofing. The security process never gets exposed to the application layer. These problems extend to all IoT objects. You may think it is not a big deal that these smaller appliances get hacked, but once they are connected to each other there are scenarios which can cause issues.
How do you enable hardware security in the small items with very low price points? They have designed to interoperable with cloud providers. NXP enables compliance to many key protocols via a secure micro-controller.
Finally was Suresh Babu who spoke on enterprise and the internet of things. He started with an example of smart cranes using anti-collision systems with many different sensors built by Honeywell. The crane is almost like a spacecraft. Will cue the operator if it sees a problem and if does not see action being taken it will take action on its own. Showed a really cool video where you could drill down to see a crane in full 3D. They have over 1,000 cranes being monitored. The data from the cranes is also used for maintenance and compliance.
He discussed how business models can change when you can monitor so carefully (more outcome as a service). He believes that a new glue is needed (between autonomous and semi-autonomous things, sensors, etc.). What does Connected Assets create: You can monitor the assets, predict assets, tackle new forms of customer interaction (SLA monitoring), Higher asset uptime, marshal parts, create new business models.