1. We are here at QCon London 2015. I am sitting here with Yodit Stanton. Yodit, who are you?
I run a small start-up called OpenSensors.io. I am a software developer and we started OpenSensors about a year and a half ago because we found there wasn’t enough tooling really for ingesting lots and lots of IoT data or Internet of Things data. So, we aim to enable people to publish and subscribe to real time, open data around especially the public space. So, if you think about it, there are lots of sensors in cities going out but really the data is being kept in silence. So what we aim to do is create an exchange and other people create IoT projects in private and that is how we make money. Generally – we are just trying to create an exchange where people can get easy access to data.
2. So, all of this data is then available to everyone, the public data?
Yes. It is available with an open data license. We incubated in the Open Data Institute that was started by Sir Tim Berners-Lee a couple of years ago, so it is really about creating a level of reuse and with that will come kind of an ecosystem where cities and local governments are building kind of “smart cities”, but really lots of people should be able to create services for these cities.
3. What kind of data can I subscribe to? Is it like temperature, is it traffic or what kind of data?
We have a lot of data around air quality so we have especially – I think we are nearly getting worldwide coverage with air quality data, real time data, and that is the easiest data to sense for. We have done a few projects around that and there is also some air quality hardware manufacturers that are using OpenSensors to enable their customers to publish data. We have some flooding data, so there is one group called the Oxford Flood Network. They put little Arduinos under bridges and that are sensing river levels and that is quite interesting to local residents. So, it is quite varied actually, it is quite surprising what people publish. There is electricity usage for some buildings and so on. So, seeing what interesting things people do with it, it is always kind of [..].
Werner: It is interesting to see what is important for people.
Exactly.
Both. The way we work is you subscribe to what is called “topics” and they are fairly similar to RabbitMQ topics, but we kind of enable you to create your own topics. The way the topics work is that we say Oxford Flood Network are publishing on this topic stream and you can trust that topic stream and people subscribe to it and they get the real time data. But, you can also get the usual historical API for today or last week or whatever. And yes, it works quite well. I guess we are facing a fairly technical audience, I mean, whilst we are also trying to make data accessible, there is still a certain level of knowledge that is necessary, but hopefully we will overcome that over time.
It is definitely not new. It is just a combination of lots of technologies, I think. I think we finally, possibly, thanks to the data storage, obviously the cloud and everything else, we are able to run these big systems that can actually handle this much streaming data. Is it new? Not really. I mean I think that electronics-wise, we probably had the capacity since the 70’s or the 80’s. I mean it is certainly not new. Machine-to-machine is not new. I mean RFIDs are not new. It is being able to produce lots of this hardware and process the data – the capacity is, I suppose, now available and people – I can’t remember the name of the guy that came up with ubiquitous computing [Editor's note: Mark Weiser]. I mean this was the dream, where everything was connected and on the internet. Finally now, I would not say it is mature yet, but it is kind of looking like it is turning into an industry. So, what is IoT? It is still to be defined. I don’t think it will still be called IoT in a couple of years. It would probably be like the norm, like the web is the norm. The capacity to connect everything is quite interesting and there are a lot of interesting business models coming out of it like car manufacturers and pretty much anyone that is a traditional manufacturing. Industry is starting to rethink what they are going to sell and what their products will look like and why. I think there is a lot of question marks around how useful is a connected toothbrush, for example. Is anybody going to buy it other than some early adopters. We’ll see.
Yes. There is always issues like that. The joke is what happens when your toothbrush comes with a user agreement?
7. Am I just allowed to brush my teeth or can I use it to clean the sink?
Exactly, yes. So, interesting times, I think.
Werner: Definitely, yes. So it is interesting to see that it is the lower prices of sensors that allows this to be available to everyone and not just specialized industries.
Yes. I mean, specialized industries are starting to take it up. I know that, for example, engineering companies, when they dig underground, they are starting to put lots of sensors everywhere to a) monitor the amount of air quality and dust they are releasing and also for the health and safety – is the ground moving and all sorts. So, they are definitely starting to adopt it for trivial things where you can just throw the sensors away afterwards. Then, obviously, there is the cheaper, maybe consumers side with all the kick-starter projects and wearables and all that. I think that we are still at the experimental stages, which is exciting really, but there is definitely no clear path to where it will all go in the consumer’s side. I think it is much easier to imagine for logistics and so on there is useful things that people can save money with or automate things.
Werner's full question: With IoT, we are talking about lots and lots of devices in all kinds of hostile environments. So, do you see any patterns in how to keep those devices running, how to keep them connected? Are there any patterns that you can see developing or that, maybe the IoT movement has stolen from the past, from people who did that 30 years ago?
I think it is a really challenge, actually. There are trade-offs for everything and choices. So there are trade-offs for power. If you have something in the wild that will have one or two batteries, there are trade-offs you make of what the capacities are. It is quite different to if something is plugged in and there is a smart meter that is reading the energy – that is really no big deal. There are also trade-offs around connectivity. I think there is – even in cities, you are not really going to need WiFi and WiFi drains the battery. So, one of the more interesting outcomes is that people are really starting to explore things like white space technology which is the old TV white space and streaming data over that. Sigfox is really interesting and they are creating their own network. I think that especially for rural applications like farms there are still a lot of challenges and trade-offs. One of the interesting outcomes has been that Ofcom in the UK has approved white space about 2 weeks ago, in February and it would be really interesting to see the companies that develop that around that and it is one to watch, really.
9. So that would be lower energy or just easier to do mesh networks and other things like that?
Yes, radio networks, all sorts of different bandwidth. I am not sure. I think Sigfox are rolling out a network in the US and they have some in France and the UK. Spark, which are a hardware manufacturer, have also now announced that they are going to become a mobile carrier so they will actually issue SIM cards with their hardware. So, it is interesting to see the hardware manufacturers actually becoming telcos in their own right whereas the traditional telcos are possibly not keeping up. So, there is a lot of disruption to be had there as an industry?
That is a good question. I think hardware firmware is an issue - if you have one farm, you can probably put together Arduinos and some sensors and maybe roll out 50 to 100 sensors realistically, within a team. I think it gets more complicated if you want 5,000 sensors. Then you are kind of going into manufacturing and you possibly would not chose to use or maybe you would chose to put together your own board based on ARM chips and so on. That cycle of designing and prototyping and then shipping it to a factory and then making sure that the QA works and all that kind of stuff is quite challenging. There is still a bit of a barrier for, I suppose, small projects, trivial projects to manufacturing. But things are starting to get better. There is fab labs and all sorts of things kind of popping up in the UK and there are manufacturing plants that are opening in Mexico, to be closer to the Valley so that start-ups can actually iterate on hardware. But, being honest, it is still a bit of a bespoke work.
You probably can but it is not for the early starter, shall we say. There is a lot of know-how and firmware is still immature. I think there are some interesting projects to put JavaScript on the chip, we are experimenting with putting Go on the ARM Cortex chip. So, I think a lot of that tooling is from whenever C was actually mainstream and there are very few people that can actually program this stuff. But I have no doubt that over the next year that the abstraction layer would be much higher and much more accessible.
For the record, I am not writing Go; there's an implementation on the chip but [someone's] compiling Go down to machine language. I can send you the details though.
Werner: It is interesting that we can actually move away from the standard, old C to a little bit higher languages.
Yes. It would be. There are a few interesting projects compiling JavaScript on the chip and I think there is possibly a KickStarter coming out to compile Lisp on the chip, which would be really cool. So, yes. I think there is a movement to just bring up all of this stuff and then it will be a lot more accessible.
Yes. Security is a difficult one because we, as an industry, we have kind of possibly forgotten how to secure things when we have very little memory and not many resources. The usual way to secure hardware does not work. There are a lot of ways, I suppose, to hack something: so there is the tampering of the hardware itself and making those tamperproof. But there is also – obviously they talk on the wire, over a radio and then down to the internet. So, there is there are many ways to compromise something.
Werner: There are many ways in.
Yes. So it is really thinking through how we design these systems. Now, making the hardware tamperproof is probably one of the easier ways, especially if it has been manufactured. I am not really sure if everyone has figured out the communication layer yet and it is really kind of asking yourself what someone can know if they get on top of the network. So, like I was saying before, for a car, it is making sure that your main OS is secured from the connected OS and really having sensible designs. There is a world of research to be done around and unfortunately, sometimes, in the excitement, maybe security kind of gets left behind – because it is the hard bit, isn’t it?
Werner: Yes, it is a vegetable course.
Yes, yes.
So some of the secure hardware that friends of mine are designing is to have the normal chip and then a chip over it and that monitors whether it has been physically tampered with. Sometimes you have things like a someone wiring a bridge for structural defects, and you want to be able to shut down the bridge and make sure that the trains are not going past. But really you do not want somebody to be able to kind of shut down the whole town.
Werner: There are many different sides to this.
Indeed, yes. And some things just don’t really matter. I mean, if it is just temperature sensors, it is just not worth investing the time.
Werner: Since it is largely public data anyway. You can look out the window.
Exactly. Well, yes.
15. To wrap up – your company, your projects, where can we find it?
It is opensensors.io. There are a few thousand people now publishing and subscribing to data and we are doing projects with cities. There is a twelve city project that is all about parking and putting parking sensors in the ground, but we are getting into some predictive analytics. But really, the point is to allow as many people as possible to ingest the data and create good parking applications and good services that actually most of the times local authorities can’t do or do not have the capacity to do.
16. You said predictive analytics – so, you can say where parking spaces pop up?
Yes. And things like “Should I even leave the house? Is there going to be parking in 2 hours time to where I want to go?” There is a lot of effort to cut down on congestion. I think the figure is something like – around London, 20% of the time people spend looking for parking spaces. So, can we actually make that a little easier? And then, obviously that is great for the person driving the car but you can also cut down on pollution and general traffic congestion and so on. So, there is a lot of nice effects that could be had with quite simple automation, really. I mean, when you think about it, we have been doing this stuff in computer science for a while now. I think the idea of programming the physical world is quite compelling to most computer scientists.
Werner: Oh, yes. And if I do not want to publish my data, I can use your services, I guess.
Yes. It is just a switch that you flip between “private” and “public”. There are start ups creating all sorts of wonderful products from it, which is really interesting. A lot of really interesting things are coming out in the built environment and in kind of optimizing how we use buildings – Can we shut down buildings, parts of buildings when it is not busy? Can we move people around? I think it is really early, but – my co-founder Malcolm always says “In a couple of years, we are going to look back and some of this stuff that really is going to explode, it is going to be so obvious, but we can’t really see what is going to come in the future”. So we are just kind of enabling lots of projects to come up and see what happens really.
Werner: Well that is a good line to end on. Thank you, Yodit.
Thank you.