FINISHED TRANSCRIPT
EIGHTH INTERNET GOVERNANCE FORUM
BALI
BUILDING BRIDGES ‑ ENHANCING MULTI‑STAKEHOLDER COOPERATION FOR GROWTH AND SUSTAINABLE DEVELOPMENT
OCTOBER 22, 2013
11:00 AM
WS 58
MOBILE AND CLOUD COMPUTING IN EMERGING ECONOMIES
********
This text is being provided in a rough draft format. Communication Access Realtime Translation (CART) is provided in order to facilitate communication accessibility and may not be a totally verbatim record of the proceedings.
********
>> JACQUELYN RUFF: Good morning, everyone. Great to see you all here. Mic on? How is that? Good morning, everyone. So this is the Mobile and Cloud Computing in Emerging Economies workshop. It's great to have the very experienced panel that is here today. Looking forward to that.
Let me give you a little bit of the outline of how we'll be set up and set the stage a bit. I'm Jackie Ruff. I'm from Verizon Communication. Our speakers today will be Verena Weber, who is from the OECD, the Organization of Economic Cooperation and Development. She'll do an overview of policy and interstructural challenges related to the service deployment models of Cloud computing, OECD perspective.
The second one is Rohan Samarajivia, the Asia perspective technology challenges. Joao Barros from the University of Portugal, Institute For Telecommunications is going to do a case study or perspective of building an urban digital ecosystem, which is the path to using future technology and Cloud technologies to leverage big data for social benefits in cities.
So we have an analytical framework in Asia, perspective, urban, and so on. We have a remote moderator who will bring in questions. Hopefully we have a large remote audience following all of this in addition to everyone who is here in the room.
Each of the panelists will talk for up to 15 minutes. We thought we would do the whole panel and then have the discussion. We're hoping to address infrastructure, privacy security, and Internet governance, evolution of standards, big data, e‑Health, e‑government, a very wide range. And I am sure that we have terrific experts in the audience as well as here at the table in the front. So it will be useful to at that point your expertise and questions in order to have. Let me say as an introduction I'm with Verizon Communications, and we are a provider of both mobile and Cloud computing. The mobile service we provide is in the U.S. But our country is ‑‑ though not an emerging country, we have mobile now moving into fourth generation wireless using the LTE standard. We now cover all of our country and have done that in the course of several years, actually, remarkably quickly, given that we could build on our existing networks. So I think that some of our experiences are relevant around the world including an emerging economies.
We are a Cloud service provider everywhere. We have large a data center in Brazil and a number of them in Asia and so on. So this is a reason that this topic is of keen interest to us
I would say when we look at mobile and Cloud computing in emerging economies, yes, it is an important trend that is expanding. It is truly a reality. It's not just theoretical, but the enormous potential that could happen over the next five years or so is truly remarkable. If you think about fact that 90% of the world's population is now access ‑‑ now has access availability of mobile coverage. You think of the fourth generation wireless like LTE, which is truly broadband, truly those high speeds. LTE is now being deployed or trialed by more than 400 operators in 124 countries.
You then think about Cloud services, which are happening everywhere, and where the greatest rates of expansion are in emerging economies. I went and looked at some studies and Gardner, the industry analyst, says the greatest expansion is in Asia Pacific, Indonesia, China, and then Latin America, Argentina, Mexico, Brazil.
Cisco, last week, issued a study in which they projected the increase in Cloud between 2012 and 2017. They said it was sort of a different cut. They said it was Middle East and Africa. What's notable is that that's a real collection of emerging economies. So the expansion is not so much in Europe. It's not in the U.S. It is in the rest of the world.
We are in a moment where case studies and policy and issues discussions are very timely. So I think I've covered all the introductory parts. Let me now turn it over to Verena.
>> VERENA WEBER: Thank you very much, Jackie. I'm Verena Weber and I would like to raise some issues while waiting until the PowerPoint is being shown on the screen first of all, I would like to talk about where do we come from when you talk about Cloud computing. That's my first point.
>> JACQUELYNN RUFF: Why don't you stop for a moment here?
>> VERENA WEBER: Sorry about this. We thought we had tested it.
When we talk about Cloud computing, that would be my first point. My second point is why Cloud computing is actually interesting for developing countries. And is thirdly, and finally, I would like to talk about policy areas that we identified during our work. Those are mainly the ones we identified for developed countries. But as I will explain, most are actually the same for developing and emerging countries.
Where are we more or less with the slides?
(Off microphone)
>> Come. Great.
(Applause)
It's great to have an engineer on the panel. Next slide, please. Great.
So let me start by looking back where we come from, actually. I don't know if you remember terminal computing, or maybe if you've even heard about it. That was around the year 1978. Basically the concept was that you would have a terminal that would be connected via network cable to a minute I computer. You had the monitor and a keyboard and that was it. You could connect several of the terminals to the mini computer. So actually you could think of terminal computing as the Cloud before the Cloud.
Next slide, please. Then a bit later, and this is around the early years 2000, we had a Japanese push to have home servers. Here the idea is that basically you have one home server where you could store all kinds of different contents on it, like photos, videos, etc. And you could access it via Internet connection.
The problem was, though, that it didn't work very well. One reason is that at that time upload speeds were really low, so you had just trouble to upload your content on that server. And the other problem was that it actually wasn't that easy to use. So people had trouble. So we saw it in Japan, but not really in many outer countries.
Next slide, please.
And so more recently, we have seen a a rival of this here. This basically operates as a home server that hosts content. Again, you really know how to use that? People that are trained, I mean, for them it's not a big deal. But for, like, an average consumer, or a start‑up small company, that's not that straight forward. Next slide, please.
So the advantage of the Cloud is actually that many Cloud services are pretty easy to use. So they're pretty easy to use by individuals, small companies can use them easily. There's another point I would like to make with that slide. Actually, when you talk about Cloud computing, we talk about a very wide range of different services. So we talk about on the one end software servers, talking about software sales for CRM. Basically where you have a very high level of programmer convenience, everything is up and running. You just put your data in it.
On the other end, we talk about infrastructure as a service. Infrastructure as a service practically provides you access to a virtual machine. So that means this is very flexible, but it's up to you to develop a platform to develop applications, etc. And when we talk about Cloud computing and several policy benefits and challenges, I think it's important to always refer to what kind of servers are we actually talking about.
Next slide, please. Now, why is Cloud computing particularly interesting for developing countries? So, actually, that's pretty new. Cloud computing allows people and companies in emerging and developing countries to access and benefit from a wide range of very well developed computing resources that would otherwise not necessarily be available in these countries.
And the other main benefit is that actually you only buy the quantity you need. If you have a small business, you buy a small amount. If you increase your activities, you can up the scale at a very fast pace, actually.
So basically we see a couple of benefits for very different stakeholder groups. So individuals and developing and emerging countries, they mainly benefit from software services. If you think of e‑mails or Google docs, etc. Also, about platforms that are already assigned for mobile devices, if you think about a platform where you can upload your data.
Then for businesses and emerging and developing countries, the Cloud is very interesting because those businesses in these countries have normally stronger financial constraints than businesses in more developed countries. What a Cloud does is that it allows it to buy computing as a pay-as-you-go model. So that actually means that you don't have up front capital investments. If you have an idea and you have the infrastructure and the Internet connection up and running, basically it doesn't cost you that much money to get access to computing resources. That's a huge advantage.
Finally, for governments we see in emerging and developing countries, governments rely on IT infrastructures and confronted with the need to upgrade infrastructure and extend the infrastructure. Cloud services can actually provide access to that infrastructure in a very cost and also energy‑efficient way. We see that in many OCD countries. I think the U.S. is leading in that area, where you have a mandate to use Cloud services now instead of just building up additional at that time center.
Next slide, please.
So at OCD we did some work on Cloud computing and we identified a whole range of different policy areas. I won't cover them all today. But this is like the overview. So it goes from how to spur the use of Cloud computing to the promotion of standards, we don't really have statistics to measure that. The link between Cloud computing and big data, we have several workshops at IGF here. Then what are the kind of infrastructure requirements and infrastructure policy requirements that we have. Then a whole lot of policy areas that are tied to the cross‑border flow of data, including privacy and security issues.
What I would like to do in the next couple of minutes is mainly talk about the areas of how to spur the use of Cloud computing, how standard ‑‑ the promotion of standards are important, broadband infrastructure requirements, and I would like to end with privacy and security.
Next slide, please.
(Off microphone)
We are sorry for our remote participants. Are you able to tell when there are any? How can governments spur the use of Cloud computing? One of the most important areas is that we actually act as lead users of Cloud computing. What we can see is that actually many governments are relying on all different Cloud computing services. So some examples are here. For example, the platform from NASA is a very good infrastructure when you think about platform as a service, basically why you have a platform of the Cloud computing and you develop your own applications on that platform. We see applications in the area of software development environments or education programs.
Finally, there are really a lost software services ‑‑ Cloud computing services that are used by governments, starting from human resource management to knowledge management support, or legal case management systems.
So basically many, many service and applications for governments are already there, and they are used in these countries.
Next slide, please. Now, one main issue that we identified when it comes to Cloud computing is standardization and sometimes the lack of standards. And this is especially true for platform as a service. What happens is that actually the Cloud provider provides you with the application programming interface. You're developing your applications, but then moving those applications from one provider to the other is not that easy. So it would be very important to have open standards in that area. Governments have a role to play in encouraging the development of operation standards, not developing them themselves, but encouraging the development and we see that actually currently there are quite some standards organizations that are working exactly on that issue.
Next slide, please.
Now, let me come to infrastructure. I want to make we're talking about quality of service later on, I think. So everything about businesses in emerging and developing countries is relying more and more on Cloud computing. One, you need a good Internet connection and you need electricity supply to do that. But second, you want to upload content. You want to upload applications. So what's actually important is to have symmetric bandwidth. You have cable at the very left, then DSL in the middle and fiber to the home on the right. We can only see high upload speeds for fiber to the home.
So that means that we have to promote the deployment of high speed Internet infrastructure in order to guarantee symmetric upload speeds.
Now, because this is a workshop where we talk about mobile Cloud computing, we see that especially in emerging and developing countries more and more Cloud computing services will need to become mobile and operated by mobile devices.
If you want to deploy high speed mobile networks, you need a backbone infrastructure that is performing enough to do that. Currently we see that most LTE networks, Jackie talked about LTE networks a second ago, mainly rely on fiber. This is very important.
What we see at the moment is that there are lots of data caps by mobile operators. This is really a problem in many, many countries. So we need to look for innovative solutions there next slide, please
I brought one example from France. This is to say that France has a very good fixed network in the capitol. So what I do is they bring in fixed and mobile networks together. On my phone as soon as I'm reaching a free box, which is on the right of the slide, I can get wifi access. Wherever the fixed slide, I can hook up to that box. What I do is basically shifting traffic from the mobile network to the fixed network, which is a pretty smart idea
My last point regarding infrastructure is the hosting of content and where the content is actually hosted. And I brought a map with me. That's the work of a colleague of mine. That's a company that is track where the most popular one million websites in the world are currently hosted. And so the countries that are in red are the countries where more content is hosted outside a country than inside the country.
In blue you have the countries where more content is hosted inside the country. So you can see a pattern, especially when you look at Africa. Currently we see that not a lot of content is hosted there. If you want to make Cloud computing work, it would be good to start hosting some content in emerging and developing countries because this allows you to keep traffic local.
Next slide, please.
Finally, I would like to end with one important additional policy issue, which is privacy and security. And I see that we currently have a lot of challenges. The key issues are how can I protect the data on the Cloud, what ‑‑ how is the contract set up? If you see you're an individual or small company, basically you can't modify the terms of the Cloud computing contract. So it's take it or leave it.
At the moment there are no guarantees that you get data back. Once the data is lost, if the Cloud provider has a problem or it's going out of business, in many, many cases you're not covered. So other issues are how to handle data in the Cloud in a secure manner. Then how to determine who is if the data is in another country, which law is responsible for that.
Then the current issues which law enforcement bodies have actually access to the data. Is this transparent or not? Finally, how do we enter equivalent data protection of data held in the Cloud and data stored at a PC at home.
I would like to end with that globally harmonized approach certainly would help to address all of these problems.
Thank you very much.
>> JACQUELYNN RUFF: Let me just do a quick transition to thank you for all those comments and the overview and so on. As I looked at your slide about the issues to think about, two observations came to my own mind. One is the number of workshops and sessions here on those very topics. So it's great. That's what the Internet Governance Forum is all about.
What’s often said about Cloud services is that they are expanding dramatically, but it's a continuum. At least from the position of a company like ours that has provided global enterprise services of many types for a long time, many of those same issues of how to protect at that time, how to manage with different laws and different countries, are things that have been dealt with in various ways through contracts or just wait that you set up systems for a long time.
So there is actually a good history to draw on. We're not starting from the beginning, but it's how do we go to the next phase as all this becomes more prevalent.
JACQUELYNN RUFF: Rohan, I'll hand it to you.
ROHAN SAMARAJIVIA: Very optimistic water on this. So I'm going to be pouring some cold water on this shroud.
On the other hand, I want to declare to you there was a hand, not only am I the hair club president, I'm also a satisfied client. This was a guy who had got plugs in his bald head. I can assure you this is not hair club product. This is original; right? So why I say that is because I run a regional think tank called LIRNEasia that does work and analysis in this area. We are essentially users of the Cloud, not of the high end Cloud services, but we basically function without a specific dependence on a particular server in a particular place. We have a website that we use as a filing cabinet for all our documents. We use Google docs, etc., etc.
So we really have an investment in this we are very clear about the difficulties of using it, because we use it not in places with FTTH, but in places that are the least connected places in the planet.
If you could move on to the next one. This is a report that will be coming out within the next few months it's for Cloud computing, particularly for small and medium enterprises or microenterprises, which is our angle. We are looking at poor market solutions. When we look at that, you can see that for the ‑‑ there are five dimensions of quality performance that we are looking at, download speeds, upload latency or return trip time, jitter and packet loss.
Our colleagues have developed software solutions and we have done testing in a large number of Asian cities. And I say cities or locations because the results that we get are not generalizable across entire countries. What you get in Bali is not necessarily what you will get in another country, but particularly in a large country like Indonesia, a lot of discussion about broadband quality. Remember we started in 2007, is about download speeds. That's sort of the main dimension that people use.
But instead of that, we looked at all these five dimensions and we also tested across three domains. Within the ISP network, that is within the country and within the same operator, you're downloading and uploading files, doing all the tests, and then within the country but not on your ISP, and then outside, the international.
So we were looking at all these. We were using it as a diagnostic tool that regulators could use to identify the problems and prioritize their regulatory solutions.
Now, at the bottom part of this slide, you see the same slide set. You'll see that when we say three marks, that's very important. For example, for browsing, etc., etc., you see that upload is not very important. But when you come down to the Cloud services segment, we are looking at data, storage, management information systems and so on, the upload becomes more important.
Latency becomes extremely important with Cloud services. So this sort of gives you a sense of the new ways in which we have to look at quality and the problems that we experience as users, as small businesses in these areas. So let's look at latency. In comparing within the ISP within the country, and international, illustrative Asian cities, I'm not expecting you to be able to see the cities. We've done this test in Bangkok, in Pakistan, in Jakarta, and Indonesia.
That red line is not fantastic, but it is what's important for most applications. This is the regulatory standard that was first adopted by Singapore's IDA and now has been adopted by India TRI, Bangladesh and so on. The regulatory agencies have adopted the 300 millisecond rule. Maybe we had something to do with it engaged with all these people.
What you see from this big picture, and this is illustrative; right? I'm not trying to prove that there is a problem in every single circumstance everywhere, but one research manager who did this is Shazna Zuhyle, which is sitting over there. If anybody wants any details, you can talk to Shazna rather than talk to me about this stuff.
You can see that performance deteriorates in the international domain. Things are going a longer distance; right? But then do you want to handshake with the remote server or not? Whether it's long distance or not. So here is a question for Cloud users. If you are using things inside your own country, you're likely to have better handshakes and better performance, lower latency than if you're going international.
I will talk a little bit about what we are separately doing with the United Nations economic and social commission for the Asia Pacific to try to address this problem, because this problem is caused by not some inherent qualities of moving data from one place to another. This is caused by the fact that in Asia we don't have enough cables, the cables are IP transit about ten times more than in other places and so on. We don't have enough redundancy. We've got choke points in Egypt, Swiss. We've got choke points in the Taiwan Strait and so on. What matters to us is this is the situation that somebody wanting to do Cloud today has to face. This is the reality of Asia.
This is important for small businesses, but it is also affecting the large businesses who want to do Cloud, right?
Let's look at download and upload speeds. On the left you see download speeds. On the right you see upload speeds. My right. Is it your right? I don't know. You figure it out.
(Laughter)
Your left is download and your right is upload; right? You can again see, now, here in the other one, lower was better because you want lower latency. Here higher is better. You can see that upload is pretty much across the board, and we are now here using examples from these countries, etc. A lot of places. Not big countries, small countries, all kinds of places; right?
So we have a problem. You can see.
Download performance. It's kind of important to download something. Clearly if you got all your documents on the Cloud, which is something I always experience, because I have no idea what's in my computer. Even the thing that I wrote, I have to sometimes get it off my website; right? So I'm downloading things from obscure places. What's my situation?
Again, it's not so dramatic, but you can see that performance generally tends to drop. Right? So this is the hard, harsh reality of ‑‑ I'm done with the slides ‑‑ the reality of what exists here. And the issue is that we are looking at a chain. Let's use the metaphor for chain. And a chain is as strong as its weakest link. In some cases the weakest link is international. There are problems around international. In the sense, as I said, Asia, though it's the largest continent, tends to behave, just like to give you an illustration, Bangladesh is a country of about 150 million people. Until 2012 Bangladesh's connectivity was through one international submarine cable. It was landed in Bangladesh in 2004. Miramar is connected to the world with CMA 3, which is also at the end of its useful life.
So you can see in some cases that is the weak link.
We are also talking about mobile. And we have got survey data that I'm not going to share with you because I don't have the time. Where we are looking at small businesses, not small. We're looking at microenterprises, below ten employees, right? Looking at how they are using ICT's. We don't ask them if they use Cloud. We ask them if they use ICT's. Almost universally everybody is using mobile devices. If not they are using wireless connectivity.
Now, in some cases that is the condition point because the governments have not given out enough frequencies. Miramar is a country with one of the lowest penetrations in the world today. They have a target of going from 10 SIMS or 10 mobile connections to 80 in five years. I think it's an achievable target. I'm doing my best to help them achieve it. The problem is they have been given five megahertz to do this. I can see people shaking their head. This is the reality. This is the problem.
So in some cases, even if you fix the weak link that is international, we'll have a problem in the local access network.
If I'm going to be a real Cloud computing guy, I want to be able to have access to my information all the time.
In Pakistan, there have been times when we've had 18 hours of 24 hours in the day, 18 hours the grid doesn't give me power. So that means this unannounced load shedding. That means you have to back up and back up and back up. You got generators kicking in, diesel, the only lot of issues. Your entire network will go down if they are not capable of that backup. Electricity is a big problem. In the entire region, which is about 2 billion people, there are only two countries that don't have regular load shedding exercises. I don't know if you know what load shedding is. I learned these things after I came back. That is Sri Lanka.
Then thirdly and lastly, we have now got another layer of legal issues. We've talked about what if the government comes and takes off with your computer? Governments have been known to do that, because they are investigating whether you paid your taxes or whether you have been subversive or whatever. They come off with your computer. You can have some precautions for this, you know, like always make sure you're backed up and the backup is kept somewhere else other than in your office and so on.
Now, what if we have three ‑‑ a minimum of three and maybe four countries or seven countries not necessarily we know all of them, who are involved in keeping my at that time. The country that I live in -- and actually I have a policy that I would much rather not keep my data in my own country. I love my country so much that I keep my data somewhere else. That is only real rule that my staff have been given. Go keep it somewhere else. Don't ever keep it here; right? So if they come to get me, they can just get me and not my data; right?
So now, do I know where this other data is? If I go to Cloud, it will be all over the place. And do the laws applying to all those places apply to my data? And will they take my data?
Now, you know this used to be a nice dinner conversation. It's no longer a nice dinner confession. So now we are all excited. I mean, you can debate. Is this reality? Is this perception? But I can tell you that American companies supplying Cloud services are going to be met with a lot of skepticism in Asia, except we don't know where to put our data. I mean, Russia sounds – wow. That sounds like a nice option. How about Hong Kong? No? And so on.
So this has become an issue. So that is, you would say, an interesting place that we are in. We all want to do Cloud and just to conclude, in some other work that I'm doing, we have been working at the resilience of these networks, of disaster response and things like that. We actually are recommending that people go to the Cloud, because in the end if there is an earthquake, it would smash us down every building, the Cloud is safe.
But on the other hand, if some government is swooping in there or shutting down your networks or putting viruses into your data, maybe it's not such a good idea. I come from here. We know trouble. We know how to deal with trouble, which is multiple sources, multiple redundancy problems, which is what I think we'll he have to do with regard to Cloud.
Thank you.
>> JACQUELYNN RUFF: Great. Thank you, Rohan. Let me do an environmental check here. When we first came into this room, it was really freezing, so we turned off all of the air conditioning. I notice one of our participants over here fanning himself. Should we turn the air conditioning back on? Is that possible, Barbara? Okay. Because I think it is getting a little stuffy. If we can't, that's fine. But, Joao, let's hear about the cities in Portugal.
>> Joao Barros, when I tell you the story of what several countries have done in Portugal and are continuing to do in using Cloud computing and Internet things and connected transportation to improve the quality of life in our city our city is a natural laboratory for anything what has to do with integrating information and communication technologies into everyday life, because it's not too small and it's not too large. It's a midsized European city in the north of Portugal. But already has a multimodal transportation system. So the buses and the trains and the metro are integrated with a single card. You can use all of these different modes of transportation. Statement, we have a fiber optical Internet backbone that belongs to the municipality. Of course we also have the commercial ones that belong to the different telecom operators. But the fact that our municipality has three rings of fiber open I can fibers, allows us to before we bother the telecom operators who has service has to be on 99.99% of the time. We are able to test things in the municipal network before migrating them into commercial applications.
It's a city of merchants and sailors. We've in the current situation with Euro crisis and so on is very welcome, created a lot of jobs. We have 31 square kilometers with real life and end users which we can run these experiments and pilot trials.
One of the main is the University of Portugal has 31,000 students in all the different schools, so we have faculty covering not only engineering, but also humanity, social sciences, medicine, etc., and this is very important for our project as I will show in a minute.
So the future cities project was selected by the European Commission. We submitted a proposal together with a number of partners, not just from academia, but also from the city, local government, end-user communities. It was actually quite fortunate because the acceptance ratio was 7%, so more than 300 projects applied. Our center for future cities is currently funded by the EU. We follow six very simple steps to implement our future cities center.
So we want to gather into disciplinary people and we started with the Internet, but then added infrastructure, and it's the end users who are at the center of all our efforts.
Porto has a school of architecture, those thinking about cities and also artistic perspective would certainly enriches very much our activities.
So step 1, we form into disciplinary teams with researchers from all of these areas, but also companies, I'll see in a minute. So step 2, we are building world class for your ban experiments. So we are 'em employing our devices and our solutions in the city itself, not in the university campus. I also like to tell our researchers that real life is outside of the campus. So we try doing out as much as possible. And I'll give a number of examples in a minute. Step 3 we work closely with the end users from day one. It's very important to do so right when you're writing the proposal for a project, because then they feel that they have been involved in the definition of the experiments that are going to be done. And as one of our firemen who works with us told the television, the researchers never made him feel like a guinea pig, which is something we should not underestimate. Very often researchers come with fully fledged projects that never actually address the concerns of the end users from the start
Finally, we follow an open data set policy. Very often we bump into problems, particularly for commercial interests, because some companies have exclusive rights on certain types of data that actually belong to the public, because they've been gathered with taxpayer money. That's one of the issues I'm currently solving. For example, the position of every bus in Porta belongs to a single company. We have several startups, one to use the data to provide innovative data to users. We need to break that deadlock
We have a number of industry partners already involved, so that was step 5. Not just large companies, but also many of these startups around our university. There's a hidden agenda. The hidden agenda is to allow these startups to showcase their solutions in our city to these big corporations so that we can get also Portuguese companies into the value chains of large companies like IBM, Microsoft, or Cisco.
And so ultimately we want to bring the results out to the world, not just in the form of scientific publications, which we do very well, but also in terms of products and services that we can export to other cities around the world.
So let me give you the success equation after several years working under this model we were able to condense into these four aspects. One, you need the research question that the research hers can feel passionate about, but at the same time you need the business case for the companies who are involved. Either you allow these companies to increase their revenues or to save on their costs. Those are the only two business cases that you can make, and another lesson to be learned, especially for academics like me is indeed those are the arguments you have to bring to corporations to be able to get their important contributions. Then finally you need real user benefits. The only way to figure out what those are is to talk with your end users or customers directly every day.
Finally, what is often very underestimated also by academics is the importance of actually having political support for what you are doing. So we are very fortunate to have a Mayer who believed in the project and we have a new mayor came in and he put the center for future cities for one of the cornerstone of his political program. So we're fortunate that we now have continued political support for this
So this was the basic framework, but I wanted to give you hands‑on experience. So this is a project with the firemen. And you can see they're in their operations actually using wearable technologies. These are connected in the mesh to the chief with a tablet can actually see the life signs of the different firemen. Why is this important? Because very often in operations the collapse or heat stroke because they don't realize they are suffocating or their heart is going to fail. With these wearable sensors and with the iPad application, we can actually see them in their context and give orders like come away if they're entering a dangerous zone or risk of actually something happening to them.
Now, this T‑shirt is actually commercial medical device. This is the inventor, and it's ‑‑ has many, many uses because it allows you to give your entire heart wave or electrocardiogram in a nonintrusive way. It has small electrodes and a small Bluetooth transmitter. This Bluetooth transmitter connects to the Smartphone. So we are using this also with policeman. We are using this with drivers. And the Smartphones upload the information to the Cloud so that you actually get a lot of data
Now, the first questions, of course, how do we ensure privacy rights and. The protection agency in Portugal has a number of rules we are following. Here it really helps to have social scientists also involved because psychologists, for example, know very well how to work with confidential data and private data
Another system we developed which you will see next is this sensor system, which is deployed in several buildings, which if you see any of these buildings, actually has arrows. This doesn't, for when an emergency happens. The problem is that they're static, and so you can actually be running into harm's way by following the arrows that have been put there. So these ones are dynamic. So the sensor can communicate with each other wirelessly. They know where what is happening inside the building. And they give you precise LED instructions or where to go as opposed to the static arrows which people often don't trust precisely because they know they've been put before you knew where harm's way was.
Now, we also work extensively with our harbor. It's an urban harbor close to the center of Portugal. You see these trucks that carry the containers back and forth. They are all now connected wirelessly using the new standard, the 80211 P standard, which allows vehicle to vehicle communication. This allows us to take continuously enormous amounts of data, again, from the harbor to the Cloud to help manage operations. Very often you have traffic congestion here because the ship arrives and all the trucks are there. And now we're able to actually give precise indications to each one of the truck drivers where they should go or where they should stop. This increases through put, reduces fuel consumption, reduces CO 2 emissions and also controlled from the Cloud.
We move from the harbor to the actual city, and the idea is the same. You have all sorts of vehicles that are connected to each other. You have people with Smartphones. And the data goes to the Cloud and is used then to provide services.
Now, how are you able to put enough vehicles, critical mass of vehicles to be able to actually get data from the whole city? Well, we started with the taxi company, which also has the advantage over private vehicles that are constantly circulating. We have more than 500 vehicles which are connected to the Internet. They have these boxes that are multi-networks, so they connect not only to cellular and wifi, but also with the standard among each other and this way we are able to leverage the existing wifi capacity. Basically the buses and the taxis, as they go along, they form a vehicular mesh network and they basically cover the whole city, communicating with the cheapest available access to the Internet.
Now, as it comes, not all data is real time data. Some data is delayed. So the box makes locally decisions which data it must send immediately, in which case it might use the more expensive cellular link, or which data can store and send later and it will get free wifi. Here you see the real position of the taxi cabs and the area that we are able to cover with the vehicular mesh which is basically the center of the Porto.
Now, how do they connect to the Internet in well, one of our telecom operators has 40,000 wifi hot spots. They basically turned every router in people's homes into a private channel for people to use and a public one for other people to use. And they opened all of these wifi hot spots for our data communication from the vehicles. We actually have in real time a wifi map of the city we're continuously assessing the quality of service that we provide in every part of the city from our vehicles, and the vehicles use this then to make decision of when to upload the data
When you put together the red dots, and the green dots, which are the public buses, you see we have an urban scanner that is basically gathering information the whole time. And this is what you get in terms of sampling rates. So within 30 minutes you see some of the regions are light green. Those are the regions where we get samples every ten seconds, and then others are darker because we don't have that many taxis and buses there, but most of the inner city of Porto is actually well covered by these vehicles.
And so one last example, we used these technologies to study driver stress. We have the biomedical data from the driver, but we also have information from the sensors from the vehicle. We combined this into a function that tells us if something happened or not. Green means you are relaxed and red means you are stressed.
This is the trip of one of my Ph.D. students from the university down to his village. And you can see he is generally green. But when you see the whole trip and actually you can see here was a hot spot, something happened here. So as it happens, someone tried to overtake him from the right, and you can switch to 3 D and see the speed. And so you see he had to brake over there. Then you could look at each of these points and see what happened. And you can see here, for example, heart rate 109 at 81 kilometers per hour. Then you go to the other point and you see heart rate 88, but he was actually speeding at 99.94 per hour.
One of one special case, but actually you can go beyond this by gathering data from a lot of users. If everybody gets nervous at the same spot, this means that the city has to do something about that spot. We went even further and actually studied the stress of about 40 bus drivers. So before the psychologists, when they wanted to assess stress of bus drivers, they ask a questionnaire before and a questionnaire after. They didn't get much because the bus driver would say no, this was a normal trip, nothing happened.
Now, actually the psychologist comes with a tablet that shows the trajectory, and he can zoom in or zoom out and say here, here, and here. The bus driver will remember, yeah, I braked or I was in a traffic jam. You're able to make real recommendations to the city and to the bus company in how to reduce the stress factors of bus drivers, and more relaxed drivers means less accidents and also more comfort when you are traveling because they drive in a smoother way.
And so we want to find the maximum intersection between the research business, the business interests, the user interests, and the political interests. And we have found actually, that the intersection is much larger than people think.
The only problem is that all these constituencies speak completely different languages. You need a lot of people that have spent some time in the different areas and are able to speak the language so that you can actually translate what people talk about when, for example, they talk about value creation or when they talk about knowledge building, all of these words have different meanings for these different people.
So we're building our living lab. And I look forward to your feedback. Thank you very much.
>> JACQUELYNN RUFF: Thank you to all of our panelists so far. You will agree this is a very good collection of our reports.
(Applause)
To our audience, do we have anyone participating remotely? Do we have questions? Comments? Raise your hand. There is one right here. We'll get the ball rolling.
>> Just a quick question. I didn't quite understand on the cities, you're gathering the data from the taxi drivers and the buses. What kind of data and what is it being used for?
>> JAOU BARROS: Basically we have access to the sensors in the vehicles. We also placed sensors, not just in the vehicles, but outside in the city. These sensors have just a wifi, so they're not connected over cellular. But whenever a vehicle goes around, it can ping the sensor and get the data. And so this way we use these vehicles to gather data not just about the vehicles themselves or the drivers themselves, but actually what is going on in the city.
I think right now we are still at the surface of what we can do with this information, so the whole first year of the project was about building the actual infrastructure, also building the Cloud infrastructure, how does it access the data. All of that has to be implemented.
The taxi company was able to reduce by ten times the time it takes from the minute you call the taxi until it's in your front door just by using the real position of the taxi. The taxi driver has a new service they loved which they came up with, which they can see in real time actually how many vehicles are in each one of the taxi stands around the city. Now with machine learning, we even tell them what is the expected waiting time in the different parts of the city, depending on where they go. This gets updated in real time. So this way they don't have to drive around. They just look at their onboard computer and it tells them where is the smartest place to go. Of course many of them still think they are smarter than the computer.
(Laughter)
Those are the types of services. Just one more example of the public bus company entered the whole project with ‑‑ because they wanted to provide free wifi to people. They did the pilot of one line for a whole year that had Internet. The customer satisfaction increased by 20% just because of Internet access. They can't get this sort of improvement any other way. At the same time, if you're able to use this data to speed up the bus fleet by one
>> My name is Ramadan, I'm from CMC India. I was very interested in the fantastic presentation on future cities. We are engaged in a number of products of this kind India is very different than in Portugal. We implemented an intelligent transport system for 5,000 buses in a remote city outside Bangor. We had a number of issues to deal with. Number one, you wanted to actually provide a facility by which at the bus stops the people know exactly what time a bus is going to come and what time is going to leave within an estimated time of arrival of plus or minus two minutes, for it to be meaningful. The second is it had to be a reliable system, because you had a command system which would then many buses in India and old and new. We are trying to put GPS‑based. They do not have a uniform sometime because of the sensor equipment that varies from bus to bus even in terms of the power connections that are available.
So the third problem was did anyone know where the bus stops were. In a country like India, for example, you have a bus stop which is supposed to be a bus stop and then they say it is abandon bus stop because all the people are standing under a banyan tree for the shade. When you're looking at cities and two types of cities, emerging cities like India that he was talking about, and the other is various other cities. What is your involvement in trying to do future cities in emerging cities in geography? GPS is easily available. You have Cloud, which is going to make all of this. So Cloud, mobile computing, GPS, become very relevant for future cities. So what's been your experience in collaboration? I'd like to discuss with you means of collaborating with your university.
>> JAOU BARROS: Thank you so much. That's a really great question. I appreciate your feedback.
I actually had a very similar question just a month ago. I was in Brazil and I was presenting at the Brazil similar podium for telecommunications. I showed some of the results. Immediately the question was well, how much this cost? You already have great infrastructure. Our population doesn't have the same level of communication. How do we translate some of the things you're too long in our country?
I should tell you the first service we had in intelligent transportation systems in Porto was actually based on SMS. It was a system when you are at the bus stop, you send an SMS and you get a reply of how long you will be waiting for the bus. The only problem, of course, is that the estimate ‑‑ if the estimate is longer than five minutes, then five minutes later you're sending another SMS to ask for an update on your estimate.
But students use the SMS systems all the time. They still do it. Not everybody has a Smartphone. But, in fact, in developing countries more and more people have Smartphones. As you said, GPS is very cheap now. We believe that for developing countries the way to go for this type of applications is more and more actually based on wifi. This is because many people have Smartphones but not many people actually have cell data plans. This is true not just in the developing world but also in our country. We have all the applications that do crowd sourcing, which means they get information from the Cloud to upload to the Cloud. These applications actually choose all the cheapest Internet access to upload the data and distinguish between what is data that is delayed and what is real time.
There's a lot of intelligence in there, and I would be very happy to talk about that with you afterwards.
>> I'm your neighbor. I'm ISOC ambassador here. I'm working on something similar. I believe that your network will be able to communicate if you have a lot of traffic with cars, because the cars ‑‑ when the traffic goes down, when you have a few vehicles, I think that's routing protocol doesn't work good enough when the traffic goes down. We are working on something similar, which is VDN, which is vehicular delay networks. Are you using some kind of protocols in your smart cities?
>> JAOU BARROS: Thank you very much for the question. A general question with viability of vehicular networks and the kind of applications that they support, well, it all depends on the density of the vehicles, but on the density of the access points. If you have an access point on every street, then you don't need communication between the vehicles. We have realized that actually it's possible to use multi-hot communication. We treated a start‑up company that commercializing. You have multi‑hop communications depending on the density of the vehicles and the density of the access points. The delayed tolerant access at some point in time is very important. In fact, one of the challenges with Cloud that I would like to emphasize is really the latency, which was also emphasized in the previous talk, because for many of these applications, for example, in intelligent transportation systems, you really have to make local decision with local information. So you cannot simply rely on upload go everything from the Cloud and then waiting 200, 300, or 400 milliseconds or sometimes more than for feedback to the user or feedback to the cars or the buses.
So in our experiments we have realized you actually need intelligent devices. Fortunately, because the electronics are so cheap today, it's actually possible to have boxes to put in the vehicles and also in the developing world.
>> JACQUELYNN RUFF: So I think we had a question about second row and then at the front row here. And then back.
>> Hello. I came from Indonesia. And then we support your work. Asking about the protocol, what protocol to you use?
>> JAOU BARROS: That's also an ongoing question. So we implemented IPv6, but most of them are in IPv4. We've also been working with Cisco on their list for protocol which allows us to do this translation more easily. In fact, within the network we don't use IP addresses for routing. We use the geographical location of the vehicles. We use GPS coordinates. It often doesn't matter which vehicle carries the data to the access point, just the one that is closest to the access points. By using the positions for routing, we're able to achieve that.
So the vehicles are anonymized. I believe 1609 standard forces you to keep changing the addresses of the vehicles so that you can actually not track them as they communicate and change from access point to access point.
>> JACQUELYNN RUFF: So I think there's a question here in the front row.
>> I'm from Singapore. I am an ISOC ambassador. I'd like to turn the dialogue away from Portugal, a little bit unbalanced, and perhaps Rohan and Verena can approach my question. I've got a policy question. You say that we need to find a globally harmonized rules for deployment for networks as well as for Cloud computing. Rohan talked a lot about latency and the difference in quality of standards. We need to look for new way to approach the quality of service that we get from networks. The thing is, I work in policy, and the mobile computing advocates request these improved latency networks speeds and we want to host data locally to improve upload and download times. The trouble is that Cloud computing companies are actively opposing data localization, because they say it's expensive. If I have to host the data locally, every single emerging economy that I want to go to, I have to build the data center there. So my question, perhaps, to the whole entire panel, given that I know that Joao also mentioned this research has to be interdisciplinary, what kind of dialogue or argument could be made to get these two groups work on diametrically opposing locations on the fence? How could we get them talking?
ROHAN SAMARAJIVIA: We could bring a third group to the table. That is I don't think those are only two positions; right? Very clearly even in just in terms of redundance and reliability factors, you don't want sort of legal compulsions to keep your data in a particular country.
The government would probably like that, because that would give them complete legal authority. We've had legal cases on these issues. For example, there was an ongoing case where the US was level he willing a daily fine because they were not were not responding to US data requests. There was a Canada bank. These kinds of things would suggest that the knee‑jerk response to the prism dispute, which is to bring all the data into your country and ‑‑ the United States is at least saying we will give safeguards to our own citizens. For us, we are fair game. Anything goes when you're not a U.S. citizen. But our countries, I think, we don't have even that level of protection. Our governments are saying my citizens. If you bring this Internet thinking into this issue, is distribution, mesh networks, is not keeping all your eggs in one basket; right?
So we should be looking at a situation where data is all over the place. In no place feel the data be complete. That is what I'm looking for. If they want to go into my data because they've got a data center there, they'll get a lot of partial pieces and they won't able to put the whole picture together. Only I will be able to put the whole picture together. So I think we need to sort of get beyond these two dichotomies of Amazon or whoever wants to keep all the data in the U.S. And we know that we have no protection against the U.S. international security agency.
What I would suggest to them is release your data, so if the NSA wants to look, they only get to look at pieces, not the whole thing. Maybe that is a creative way of approaching this problem.
>> Actually, I'm German, and Germany has the data is the strongest data protection laws ever. So even within the year Germany poses all kinds of problems when you try to come up with even with a solution for Europe only; right? And I was at a conference and I talked to an end provider. And I'm like wait a minute, the Cloud service is in Germany and end provider. Basically the solution they end is that actually the data centers are located in Germany. So. It doesn't make pure sense to localize all the data centers, but if you have a very strong privacy law, that also makes Cloud providers move some centers to this country. This is what happened in Germany. Distributing the data to different data centers in the world where you can see that you have at least some data protection. That's certainly a good idea.
Then you talked about Cloud providers that don't want to move to emerging companies. The U.S. companies, Amazon, etc., Google. In England there are some small companies that are providing hosting services. When you ask them, how do you do that, because you don't have the scale to do that? You don't have the scale as Amazon, Google. They tell me Amazon is not that cheap and it is possible to find a business case.
So the idea here is not only rely on the biggies, but also to approach smaller companies, and to also try to develop something really within the country within the help of foreign companies at the beginning, but later on more of a local network.
JAOU BARROS: When it comes to infrastructure and infrastructure development, if you keep the traffic local, you reduce your latency. For that you need Internet exchange points in the country. We see there are many, many countries in the world that don't have Internet exchange points. It's cheap. It costs around $20,000 more or less. So if you think for a regional for a whole country, so that's the cost that's shared among all telecom operators. Basically you have a room where they plug in their cables and exchange the traffic locally or regional. So if you distribute those costs, that's really not a lot per operator and does allow you to keep the traffic local. I think that's a really important point when it comes to infrastructure. I think ultimately the only way is to increase the trust among natures. I think Europe is, if I may give one ‑‑ I mean, because Europe has been so beaten up in the last couple of years. So let me tell you one of the things I find remarkable and I served a number of committees with European Commission. Indeed, you see these 27 countries sitting in the same room and you think there's no way that ever any consensus is going to come out of it. In the end some of the basic things like some of which my colleagues addressed have been successfully worked out in multinational panels. It takes time. That's a very good point. The fact is that technology is advancing so rapidly that the pace of legislation is way too slow ever to catch up with this development. That's a serious problem.
>> Thank you. I'm from GSMA. My question was initially for Joao, but I'll try to phrase it in a more general way using your example of cities. So if we put this behind the location of the service for a moment, going back to your example, as you move away from the pilot from the taxi drivers and the bus drivers towards this scenario where potentially every citizen of Porto and going globally is connected in real time and sends their location to the Cloud wherever the server may be based, what are the challenges and how do we ensure that people's privacy is preserved and how to give them control over who connects that location and the personal data? What are the challenges you're facing and how do you see these challenges being resolved? Thanks.
>> JAOU BARROS: So this is a huge challenge. It varies a lot depending on what population or groups of population you are working with. Let me give you several examples, including normal end users, as you mentioned. So with the bus company, every the bus company owns the whole fleet and the bus drivers are employees. So you convince the management. The management talks with the unions, very important. We had to get the unions on board. That was a huge task because we have to show them that what we're using the data to improve their quality of life and not to give the management more arguments to cut down their rights. So that was a huge problem.
But in the end it works out more easily because if the administration says okay put devices in all the whole fleet, then it gets done.
With the taxi drivers, it's totally different. It's a cooperative and you're actually talking with 400 microcompanies, and the taxi driver is totally free to switch on or switch off the box. Some of them switch it off because they feel that it's draining on their battery. We are able to prove that it doesn't, but still we give everyone the freedom to opt in. Also the taxi drivers, it gets installed, but they decide whether they keep it on or not. It's a learning process until you find the right incentives no them. You have taxi drivers are 62 and some that are 22. Their relationship with Internet is really different. That's when the social scientists and psychologists get interesting results. They can look at a cohort that has Internet and what happens over different generations.
To give you an example, before they had radio which was constantly on sending messages and some of them are very happy because now it's silence and they don't hear anything in the car and it's very relaxed. Others are feeling lonely, because the social atmosphere they had from the broadcast radio is gone. These are important aspects.
Regarding their data and their position, they take it very lightly because they know for me to get business, the taxi central has to know where I am. Since it's directly related to their business. We have been work with truck drivers. They want to be able to decide on their routes on their own without having management or anyone tell them how too long from customer A to customer B.
For private users, our application is called Sense My City. It works on Smartphone that connects to the vehicle use you go the onboard diagnostic unit. If you Google Porto Future Cities, you can find all this information. It has two buttons. It has a play button and pause button. So that you know when you are recording and sending data to the loud and pause when you don't want to. Again, it's completely opt in.
There's an interface that allows you to see all the data industries that were gathered, and what the data was. What I feel is that end users not only know how much you can actually learn from that data. So it's no risk for us to put all the sensor readings and everything available to them, because at the end of the day it is how much knowledge you can extract from them. It is our obligation to educate people what it means when they press play, how much can we actually learn about them.
So we're following the data protection rules, which means everything is anonymized. Only one researcher is able to say Number 347 is Mr. Pedro and Number 307 is Mr. Silva. But we follow policies on access to the database. Everything is logged so we know exactly who gets in and gets access to what data.
Are we able to prevent all the leaks? Of course not. But it's an ongoing experiment. So far we haven't had any complaints from the end users about their privacy.
>> I'll tell you when you'll get complaints, which is with that stress measurement. I think everybody here would be freaked out. That somebody somewhere is going to know how stressed I was in a particular situation. And there you were actually looking at individuals, not anonymized. I was intrigued that the buses are anonymized, but maybe the drivers are not, because for the stress analysis you can't be anonymized.
We will be participating in a big data session on the ‑‑ day after tomorrow. We'll be talking about big data and the privacy issues. The standard informant consent model doesn't really work with big data. What you need is something a little bit more flexible and useful. To me, anonymized would be extremely important as a safeguard. We know there are some kinds of analysis that won't be possible if you fully a Monday myself. Many of them in particular with these T‑shirts where over a course of several days we have full ECG analysis, so not just one ECG analysis when you go to the doctor. Continuously. We already identified four arrythmias that people were unaware of. And this is a benefit to them. It's expensive. You have a cardiologist looking at data over several he days. So we tried to make it win/win. When I mentioned end user benefits, we listened to them from the beginning and we say so what we asked them.
********
This text is being provided in a rough draft format. Communication Access Realtime Translation (CART) is provided in order to facilitate communication accessibility and may not be a totally verbatim record of the proceedings.
********