Building Smart Cities with Intel

By Siddharth Parwatay | Published on 01 Jul 2016
Building Smart Cities with Intel

We speak to Kavitha Mohammad, Director, Industry Solutions Group Public Sector, APJ to get a pulse on how Intel plans to build the technological backbone required to support the smart cities of tomorrow.

With the IoT sector booming, we sit down to understand Intel's roadmap of implementing IoT at the city-level. It's worth noting that smart cities are a very different ballgame compared to individual sensor based IoT projects that are developed by the DIY or maker community. The architecture needs to incorporate security and other consolidation standards. In this interaction we find out what Intel's stack (implementation model) looks like.

Digit: When it comes to the Internet of Things (IoT) the theoretical aspects itself are very exciting to us and our readers. The whole idea of billions of devices talking to each other, so many sensors all across the world sharing exabytes of data, big data algorithms analysing it and making sense of all that information – it sounds like a brave new world worth looking forward to. The predictive nature of IoT is especially interesting. But from an implementation perspective we touch IoT from the ground up at an individual level in the form of small DIY automation projects (like say a humidity monitoring device that waters your plants while you’re away) or the latest happenings in the maker movement. What you’re into – implementation at that macro level such as smart cities – is an equally exciting but entirely different ballgame. Can you give us an overview of what the architecture of such an implementation is likely to be, or what the roadmap is?

Kavita: Let’s talk about that moisture level monitoring / irrigation project you mentioned. That becomes an integral component of our Smart Cities. When we talk about Smart Cities, it’s a bunch of IoT solutions that come together, because for something to be smart, it just needs a context. To define context let’s take the example of a bus going in a certain direction. If the bus has an idea of how many passengers it will likely pick up and how well utilized it will be by the end of the day, it gives a better sense of resource management to the administration. Otherwise the same bus would be running ten loops every day without any knowledge of where the density is. Transport is at times underutilised for the simple reason that the last man is not connected. It can happen that the density of the population is elsewhere and the busses are plying in a different direction. So the question to answer is how can I get information on the population and the heatmap and marry that information with the transportation logistics. 

So is the case with irrigation. Take Andhra Pradesh for instance. People would like to plan their irrigation needs at least 30 days in advance for something like a rice paddy, which requires a lot of water. In this context, utilisation data needs to married with things like the flood management system, i.e. when the flood gates will be opened and how much water needs to be sent to all these villages. All these are now happening in silos. We take solutions from you – the maker community – put them together and create an architecture around that. We also try to do some pattern analytics between the multiple data points that are coming in. 

At the very basic level there are a bunch of sensors sending in data. These sensors can be soil sensors, moisture sensors, water level sensors, even a camera is a sensor. Multiple sensors are feeding in information and they’re correlated. Say for example a water sensor shoots up, should a controler panic? No, it could be an anomaly. That’s where a camera could help us understand whether there was some incident such as say a little boy pouring water (just as an example) or an actual increase in the water level. Intel also has something known as Edge Analytics. What this means is you don’t have to send all this data back to a city centre or a centralised command centre or a data centre because this may be something that could actually be controlled right at the location. If the soil sensor is showing low moisture, and the camera is not showing anything abnormal, I would trust the readings from the water sensor. This is something that can be computed right at the Edge. I don’t need to really pass this data back to the data centre.
Digit: What’s the advantage in not having it centrally analysed?

Kavita: Bandwidth and security concerns. We’re creating islands of intelligence. If you can control and manage that island as much as possible, you’re actually doing yourselves a favour because you’re controlling data that’s moving back and forth. Your security concern is to some extent mitigated because it’s not going on to a public cloud, and the cost of the bandwidth is also reduced to some extent. That’s the essence of it. So for us the tenets of Internet of Things are cloud, big data, and IoT M2M (machine to machine). M2M is very important because this is where the sensors and the intelligence are. Machine to machine can happen if the two machines speak the same language. But sometimes a camera can’t talk to a sensor simply because they don’t speak the same language. That’s why we bring in something called a Gateway, which the maker community loves. A gateway solves issues that the IoT community has been struggling with. Manageability: how do you manage multiple devices? Interoperability: how do you make sure that the different languages spoken are understood? And the security management. These three are very important anywhere in the stack. If you look at the Intel architecture, we have intelligence in the sensors. There is also a programmable chip that’s coming into play. You don’t need to program from a different operating system or console. The chip itself is a FPGA programmable chip. The  Gateway handles multiple sensors and it has an operating system built onto it. The OS will help you program the chip and it has security and analytics built into it. So the edge analytics island I was talking about earlier is taken care of here. And at the next stage in the stack, many such gateways then come onto a communications network.
Digit: And these gateways are Intel gateways?

Kavita: These are Intel gateways yes. The OEMs are different, but the underlying chip is Intel.
Digit: And the OS?
It’s a Wind River OS. I’m not sure if you’ve heard of it but this is the OS that’s on being used on Mars.
Digit: That’s not an x86 OS?
No, it’s a defence OS. Intel acquired this company five years ago. The mars vehicles all use this OS. It gives us this unique advantage at the Edge. Whatever gets filtered from multiple islands of intelligence – be it transportation, or public security, or even flood management – they all come onto an IP backbone which is a communications network (the CISOs, Huawei and Ericssons of the world) and it then goes back into the data centre. Once it goes into the data centre what we try to do is bring in multiple applications that can understand domain specific data. Say a camera is sending me a feed, that feed can be interpreted differently by a flood management system, the same feed will be interpreted differently by the police, and also by emergency response because they know exactly how they’re supposed to react. If flood levels are going up, and the camera sends in that feed, a flood manager will act on it. But a transportation guy doesn't need to do anything unless he has a fleet running across. That’s why here at the application level we bring in domain specific intelligence.
Digit: So here at this stage there is Human intervention?
 This is where independent software vendors can bring their applications in. This stage is getting closer to the end user. Here is where we also try to do data aggregation, data cleansing, data fusion and then also insert something called Trusted Analytics Platform. TAP is simple: you do a lot of big data analytics and then that gets triggered out to different agencies on their iPads or smartphones. For example, sending out alerts or giving them dashboards – it all gets done here. Intel is slowly picking up the pieces across the stack to make it sticky. We took care of Security through the acquisition of McAafee, Wind River is the OS which is meant for IoT, manageability and all that included. Here we have another strategic investment, which is Cloudera, which helps with the data analytics. The servers and storage have Intel Chip Xeon built into them. Then we have partnerships with the ISVs. The likes of SAP, Microsoft – the guys who are developing applications. What we’re also trying to do is, if there is a camera, with India in mind, you'd want it to be welded to the bus stop. You'd also want the camera to be solar powered, because you don’t want to keep changing batteries all the time. That's when we go to our ODMs in Taiwan and Japan such as Foxconn and Advantech, who now are slowly moving away from the PC business as the PC is in decline, as we all know. What do they do with the excess capacity present in their manufacturing units? They build things like sensors which are very specific for different markets, and they want to do it at very competitive prices. We work with our ecosystem of partners and we’ve been acquiring some companies to ensure that we have that end-to-end stack in place. That way we can become sticky for anyone who wants to play or build on it.
Digit: At this stage where the cloud analytics happen, the agencies that have access to this are I’m assuming govt. agencies?

Kavita: This could actually be a third party as well. The entire stack could actually sit with someone like Reliance or Airtel and they’ll tell the agency, "You take surveillance as a service from me". All I need to do is send in a dashboard or direct feed everyday, whenever there are alerts you will be alerted. The other way is, the govt. itself will say, "I’ll take this entire stack, I’ll pay for it, I’ll build it in my command centre in my city". This way I have total control over what I want to do. Once I have this data, this data is made open to entrepreneurs like you so that they can create a lot more insight out of this data.

Digit: So when it’s supplied to say, a garbage disposal for example, in a smart city then you would want municipal corporations to sort of have that data and manageability. 

Kavita: Two methods here. There are garbage sensors where the camera is just looking at the garbage, then there is a citizen that has seen the garbage and taken a picture of it and sent it in through a social media platform.

Digit: That’s really happening? Anywhere in the world? 

Kavita: Jakarta smart city, which is in Indonesia, is a country with a DNA very similar to ours. They have an app called Clue, which everyone has downloaded. Everywhere they see any garbage or leakage they basically take a picture and put it on Twitter or Facebook. There are multiple social media handles they can use. That data comes in and they have multiple views coming in. Similarly they are able to manage around a hundred and eighty thousand lights because it's all crowd sourced. Crowd sourcing is one thing, hardware being on ground is another.

Digit: And all these pictures are machine analysed?

Kavita: Yes. Video analytics, really-high end video analytics. Sometimes certain machines which deal with critical or sensitive things – such as the kind that identify terrorists for example – are avoided as it could get very political. But a few things can totally be done. It is present in countries like Singapore, but comes with a price, which is very expensive. 

Whether you want to do heavy lifting on the coding or whether you want to do heavy lifting on the intelligence of the chip is a call you have to make, because if the hardware is unable to support then you’ll have to do heavy lifting on the software. If the hardware itself comes with that intelligence like when a camera has video analytics built into it. If the camera is doing most of the stuff you want it to do then you don’t have to do heavy lifting at the application level. That’s where the FPGA that I was talking about comes into play; it’s a programmable chip. So all the stuff that I'd do, otherwise at a software level, can be done here. Does that give you a picture of what we’re trying to do?

Digit: Yea of course it does, quite a bit. 

Kavita: Intelligence everywhere essentially. Wherever intelligence is required we try to make it intelligent so that it becomes easy for the ecosystem partners to pick up the pieces and build the stack.

Digit: Do we have any model new/smart cities across the world, that we can use as perfect examples of IoT in place?

Kavita: Barcelona is one such city where you can see five square kilometres in action. Then there is Songdo and of course Jakarta smart city, but Jakarta at this point of time is not hardware driven, it’s crowd sourced. But they slowly want to get into having sensors and instrumentation going into the city so that they have real time inputs. They're thinking big on big data, like processing unstructured data, processing structured data but not in real time, and processing structured real time data, there are many challenges. Even right here at home, Bandra Kurla Complex – with Intel's help – is a good example. We have environmental sensing, smart parking, street lighting, and WiFi. BKC is a ripe ground that way, as it is more conducive to technology.

Digit: So the crowd source model should work in India as well more than the device or sensor specific models?

Kavita: It should work provided the citizens are conscious enough and responsible enough to keep giving that information because otherwise it’s not foolproof. You’re not getting a hundred per cent coverage. There could be a bad incident that was missed by a citizen then what do you do? It’s always crowd source plus technology, as in real cameras or sensors that are taking pulse of the situation on the ground. The two together will make it foolproof. Because otherwise there is always a loophole.

Digit: What about security, that’s a major concern right? To present a worst case scenario, you must have seen Die Hard 4 right? Critical infrastructure is now operable wirelessly or through a switch, so maybe the grid goes down, the floodgates are opened, you wouldn’t want that so, when all of these things come onto an electronic or digital system you’ll have hackers who will try to control that, right?

Kavita: That’s an inherent problem. Do we have a solution? We don’t. The thing is, it’s evolving. There are always hackers who can outsmart you no matter how much you evolve your technology.

Digit: So then the question is how much of our critical infrastructure should we put on the grid? 

Kavita: That’s where the security protocols come in. It’s a given that there would be one or two corner cases, which might get missed. But the value that you’re generating out of putting something on a grid probably far exceeds that. 

Digit: So don’t focus on the outliers? 

Kavita: Outliers, yes. I’m not undermining their importance in any way, but the thing is it’s always a trade-off. What’s the value I’m getting out of putting something on the grid? Efficiency, revenue etc. Then what are those outliers and what’s the probability of them happening and what are the downsides of being associated with them? We can always use probability and run analysis to come to a decision. It’s not being forced upon anyone but people are thinking through these anyway. Customers are far more educated.

Digit: Ok that makes sense. Changing topics back to the hardware. Since we also cater to the DIY and the maker community, what role can they play in the grand scheme of IoT, or even the other way, what role can Intel play?

Kavita: One glove doesn’t fit everyone. This is especially true for the Indian market because while there are sophisticated environments like BKC, there an equal number that are looking for very basic, raw, homegrown solutions. The maker community have a lot of ideas, so it’s kind of like crowd sourcing in a way. They need the equipment and tools to develop. So what we did around six months ago, through the President’s Office, we drove an entrepreneurship, something like a maker’s initiative, where about thirty companies were identified. They used Intel embedded solutions and we give them platforms such as this IoT Symposium event we’re at, where they could showcase the solution. That’s an avenue for them to meet with prospective customers. These avenues help them gain visibility. By being part of events like this, you can potentially bring solutions to customers that are looking for them. 

Digit: Is there anything else you would like to add for our readers?

Kavita: Sure, we welcome any new thoughts or ideas they may have. Intel has enough avenues to help them out with their toolkits. They can come in and make use of all the avenues that Intel has. We keep running programs every five to six months such as hackathons like the Intel Embedded Challenge. Intel creates the nucleus of the technology, the core, and then just democratise it. Anyone can build anything on top of it, and Intel holds the IP or the patent to that nucleus only. Beyond that it lets the entrepreneurs thrive and build over it, and wherever required we bring in support from our ecosystem, which are the ODMs, that can bring scale to the entrepreneurs who want to manufacture hundreds or thousands of those units.

Note: At the time of the interview Kavitha's designation was Director, IoT & Smart Cities at Intel, Asia Pacific Japan

Siddharth Parwatay

Siddharth a.k.a. staticsid is a bigger geek than he'd like to admit. Sometimes even to himself.

Digit caters to the largest community of tech buyers, users and enthusiasts in India. The all new Digit in continues the legacy of as one of the largest portals in India committed to technology users and buyers. Digit is also one of the most trusted names when it comes to technology reviews and buying advice and is home to the Digit Test Lab, India's most proficient center for testing and reviewing technology products.

We are about leadership-the 9.9 kind! Building a leading media company out of India.And,grooming new leaders for this promising industry. Protection Status