Ctit brochure 2015 web by University of Twente

SCIENCE FOR A SMART SOCIETY

SLOWING DOWN THE BAD GUYS SMART DISTRIBUTION OF PARCELS THE COGWHEELS OF SMART SOCIETIES PEEKING DEEP INSIDE TWITTER ROBOT BUDDIES AND SMART CUPBOARDS SMART HEALTHCARE FOR THE ELDERLY THE IMPACT AND VALUE OF SENSORS

TOPICS 2015

SLOWING DOWN THE BAD GUYS

SMART DISTRIBUTION OF PARCELS

THE COGWHEELS OF SMART SOCIETIES

PEEKING DEEP INSIDE TWITTER

ROBOT BUDDIES AND SMART CUPBOARDS

SMART HEALTHCARE FOR THE ELDERLY

THE IMPACT AND VALUE OF SENSORS

CTIT CENTRES

CONTACT 22

Are we on the verge of seeing a young science vanish from sight? It may well seem so. Software has always had this strange property that it’s so soft you can’t touch it. You can’t take it into your hands. You can’t see it. Hardware, also known as computers, is moving out of sight as well. When you ask someone how many computers he or she owns, you rarely get the right answer, which is often in the order of ten or so. Just like software, computers are everywhere: your car, your watch, your activity monitor, your TV set, your cell phone, your hearing aid, your modern lightbulb, your door lock, and so on. Oh, and your laptop, of course. Software and hardware are disappearing from sight, but obviously, they’re not vanishing.

CTIT SCIENCE FOR A SMART SOCIETY In fact, ICT has become so successful that it has often become commonplace in a way that we don’t even notice that we’re all swamped by it. It took us decades, but by now society is drowning in ICT, and we don’t even see it. ICT is everywhere and in such an abundance that we can’t even build up a complete picture. What an achievement!

CTIT SCIENCE FOR A SMART SOCIETY

ICT IS EVERYWHERE AND IN SUCH AN ABUNDANCE WE CAN’T EVEN BUILD UP A COMPLETE PICTURE. WHAT AN ACHIEVEMENT!

SMART SOCIETIES: SAFETY & SECURITY And likewise, ICT science may seem to be vanishing from sight. But vanishing from sight isn’t the same as vanishing. On the contrary, there’s a lot of ICT science taking place to let ICT run through our society’s veins and shape the things we do today, and will do tomorrow. ICT science is making our society smart. It lets things happen that we could never have dreamt of. It allows you to do business on the internet, be driven in an autonomous car, and be continuously connected to your loved ones. In a smart society, the supermarket has the products you want, the physician can accurately monitor your health, and energy is no longer wasted. ICT science for smart societies is what makes us tick, that is, CTIT, the research institute that bundles all of the ICT science at the University of Twente. Or should we say that it’s smart societies that make us tick? Or that we actually make smart societies tick? It really doesn’t matter. What matters is the combination of those two words: smart and society. It’s our core business.

AT CTIT WE CONDUCT THE ICT RESEARCH THAT MAKE THE SOCIETIES OF TODAY AND TOMORROW SMART. SOCIETIES ARE IMPORTANT TO US: WE ARE CONCERNED ABOUT THE RELEVANCE OF OUR RESEARCH FOR SOCIETY, WE EMBED OUR SOLUTIONS INTO SOCIETIES AND PREFERABLY IN A WAY THAT IT SEEMS NORMAL TO YOU AND THAT IT FEELS AS IF IT’S SUPPOSED TO BE THAT WAY. In this brochure you’ll find six interviews with scientists working at CTIT. Six interviews each with two people, each duo shedding some light on what makes a society smart. Professor Ariana Need and associate professor Djoerd Hiemstra take you to track the effect of campaigns on cancer by delving into Twitter data. Their research is also an example where ICT science and social science not only meet, but also need each other to solve a societal problem. Likewise, smart logistics that allows us to optimize transportation of goods requires a deep understanding of how potential competitors may still want to interact, while at the same time such interaction needs to be supported by advanced information systems. Professors Jos van Hillegersberg and Roel Wieringa have teamed up to research software solutions that enable competitors to trust each other. A clear example that fits into a smart Every year we highlight a specific topic society is monitoring people’s health and well-being. At the for our symposium. This year it’s ‘Safety same time, a smart society also demands that we rethink and Security’. Safety and Security is one how smart we should make our lives, as explained by of our major research themes. At CTIT we professors Vanessa Evers and Peter-Paul Verbeek. Ethical approach this theme from an ICT and philosophical issues should be taken into account from perspective, but technology by itself can the moment we start designing for a smart society, never bring overall encompassing however difficult it may be to anticipate the outcomes of solutions. We are fortunate to have our own creativity and technologies. colleagues from the social sciences on board so that, together, we can develop a There is an emphasis on the ICT science for smart more holistic approach toward making societies, as science is what drives us. Yet science alone our societies safe and secure. As you’ll will not lead to smart societies, so we always pay a lot of understand, from our perspective, a safe attention to moving the results of science into society. society is a smart society. It’s safe An interview with associate professor Nirvana Meratnia because of the natural embedding of ICT and business developer Alain le Loux illustrates how we solutions that allow us to rightfully feel do this. safe. How we can get to safe and secure societies is touched upon in a separate interview with three of our expert Prof.dr.ir. Maarten van Steen and Drs. Iddo Bante scientists in relevant fields of ICT science.

SYMPOSIUM: SAFETY & SECURITY

CTIT SCIENCE FOR A SMART SOCIETY

SYMPOSIUM: SAFETY & SECURITY

SAFETY AND SECURITY SLOWING DOWN THE BAD GUYS ALTHOUGH BANKS ARE THE PRINCIPAL TARGETS FOR CYBERCRIMINALS, MANY OTHER SYSTEMS ARE VULNERABLE TO ATTACKS, WITH POTENTIALLY HUGE CONSEQUENCES FOR SOCIETY. COMBATING HACKERS ALSO CHANGES ICT RESEARCH IN A FUNDAMENTAL WAY.

‘All computer systems that involve big data, let alone the internet, are simply too large to describe’, says Raymond Veldhuis, professor of Signals and Systems. ‘Therefore, we must use statistical methods and pattern recognition to make sense of what is happening, in order to be able to respond to it. This applies to many cases, ranging from recognizing suspicious credit card transactions to detecting the behaviour of people in crowds.’ As a result ICT is more and more becoming an empirical science, which is as much about analyzing what is happening within systems as about designing them. A clear example, put forward by professor of Network Operations and Management Aiko Pras, is the practice of DDoSing, saturating websites with information requests so they break down under the load. There are websites that offer DDos as a service. For as little as 5 euros per month it is possible to attack all kinds of victims, such as banks, schools and webshops.

ALL COMPUTER SYSTEMS THAT INVOLVE BIG DATA, LET ALONE THE INTERNET, ARE SIMPLY TOO LARGE TO DESCRIBE. - Prof. Raymond Veldhuis ‘Many researchers content themselves with establishing the scale of attacks and trying to counter them’, Pras comments. ‘We are also interested in the motives behind them. So in one case we anonymously hired hackers to attack our own servers, so we could study their methods. In another case we found clear correlations between attacks on a school network and the examination timetable, enabling us to find out which exams the attackers were reluctant to take and thus narrowing down the list of suspects.’

DESIGN FOR FAILURE If ICT professionals are puzzled by the sheer complexity of the systems they have built, it is no wonder that the general audience is often clueless. People agree to any user license and click on every exciting looking link, until something goes seriously wrong. They will blame someone else, because they feel no responsibility for things they do not comprehend. This does not mean that it is useless to try to raise consumer awareness about privacy and computer security, but it makes clear that designers should not count on it. ‘Banks know that some ten percent of their clients’ computers contain malicious software’, says Boudewijn Haverkort, professor of Design and Analysis of Communication Systems. ‘Bank security systems treat this as a given fact and are built to cope with it. This approach is used for hardware as well: you assume that some components in your server room will break and design the whole so that this failure will not interrupt your business processes.’ Another design approach to enhance online safety and security is ‘privacy-enhancing technology’. This involves encrypting user information such as mobile phone numbers or fingerprints, in a way that still makes unique identification possible but obscures the original data, thus preventing the accidental revelation of sensitive information.

LOOPHOLES The possible sources of mayhem emanating from digital systems are countless. Energy supply, stock markets, traffic management, hardly any vital system in today’s society (and tomorrow’s smart society) is unconnected. Their complexity and as a result vulnerability is increasing. Since charting potential problems in full is impossible, the basis for protection lies in statistical and empirical analysis of weaknesses. This calls for academic knowledge of a different kind than is traditionally pursued in ICT research.

BANKS KNOW THAT SOME TEN PERCENT OF THEIR CLIENTS’ COMPUTERS CONTAIN MALICIOUS SOFTWARE. - Prof. Boudewijn Haverkort

THERE ARE WEBSITES THAT OFFER DDOS AS A SERVICE. FOR AS LITTLE AS 5 EUROS PER MONTH IT IS POSSIBLE TO ATTACK ALL KINDS OF VICTIMS, SUCH AS BANKS, SCHOOLS AND WEBSHOPS. - Prof. Aiko Pras ‘In order to be able to analyze the phenomena you are trying to tackle, it is essential to have deep knowledge of modelling’, says Veldhuis. ‘Like physicists studying natural phenomena and fitting them into a mathematical model, we must study and model online phenomena, for instance the way malicious software spreads through a network or the behaviour of a botnet.’ Take, for instance, a water treatment facility. The different components are all equipped with plenty of sensors measuring the water quality, cameras guard the facility’s perimeter; a rigorous protocol defines which employee has the right to undertake which action. Still, there will be plenty of loopholes, like somebody leaving his spot for a short break without logging out, enabling others to access or even modify information. An adequate model of the actual situation may find such weaknesses. Safety and security of ICT systems are so interlaced with physical reality that the two can no longer be dealt with separately.

CTIT SCIENCE FOR A SMART SOCIETY

SMART LOGISTICS COULD DRIVE DOWN THE (ENVIRONMENTAL) COSTS OF PARCEL DISTRIBUTION. BUT HOW DO YOU DESIGN SOFTWARE THAT MAKES COMPETITORS WANT TO WORK TOGETHER?

participants will be reluctant to reveal their priorities in full. Secondly, the joint system must not only be able to communicate with the participants’ software, but must do it in such a way that it operates efficiently without revealing how it achieves its results, since this would Partly due to booming e-commerce, which allows involve disclosing details about packages. On the other consumers to order directly from wholesalers or even producers, the average packet size of goods deliveries is hand, the participants must trust that the system deals with all of them fairly.’ shrinking. As a result more and more goods are delivered by road, because transport by rail and water typically involves larger loads. This development may be Internal security is not the only issue here, Wieringa efficient for individual deliveries, but for the system as a adds. One may be pretty sure that such a system will come under attack from hackers, both with commercial whole it is not – and that’s not even accounting for the and military interests. So, ICT-security, until now not a higher environmental costs of road transport. great concern among logistics companies, will become a central focus for a synchromodal system. This is where synchromodal logistics kick in, says Jos van Hillegersberg, professor of Design and Implementation of Information Systems. ‘The idea is to SKIPPERS postpone the decision on which mode of transport to So far, some small-scale experiments have been carried use as long as possible. This enables you to gather out to test the concept. Van Hillegersberg: ‘We use larger loads with the same destination and then serious gaming for this, because in this way managers dynamically decide how to ship them.’ can gain insights in a very imaginable way. For instance, we did a game about inland shipping. Typically vessels The synchromodal approach leads to a more complex have to navigate past several points to collect loads logistic system. Traditionally, logistic networks are before they take on a longer journey.’ supposed to be fixed and planning is about efficient transport along existing lines. Synchromodality implies THE SYNCHROMODAL APPROACH LEADS TO A MORE COMPLEX dynamic lines. Customers will close a contract for the LOGISTIC SYSTEM. delivery of goods without knowing by which means the - Prof. Jos van Hilligersberg package will travel. The development of dynamic planning is helped significantly by the ubiquitous implementation of sensors, which enable real-time tracking of parcels. The ample availability of data makes it possible to base pilot projects on actual information rather than on simulations. This, in turn, leads to a higher credibility of results and hence better chances for actual implementation. After all, synchromodal logistics call for a seriously different organization of the whole sector, which will happen only if the benefits are clear.

THE PARTICIPANTS MUST TRUST THE SYSTEM. - Prof. Roel Wieringa ‘In the first round it was every man for himself – the current situation. This leads to waiting times, as a vessel may arrive at a dock where another vessel is moored at that time. The second round provided direct feedback. This gave skippers the opportunity to adjust their route, if they knew somebody else was heading to a dock they were planning to call in at. The third round provided coordination, assigning everybody a route.’

REVEALING As it takes quite a lot of parcels to fill a train or a boat, synchromodal logistics demand that competitors work together. However, as efficient logistics form their core business, they will want to prevent information about their own systems leaking to others through the joint system.

While the skippers could see that coordination led to the optimal collective performance, inevitably some got a better route than others. This should be compensated for in the course of time, so everybody feels the system is fair.

In a simple game like this one, fairness is more or less calculable. But once it is about distributing and bundling packages, perhaps even speaking to clients about ‘From the point of view of software design this is a delaying deliveries, the complexity may rise to challenge’, comments professor Roel Wieringa, who leads the research group on Services, Cybersecurity and unexpected heights. At that point, Wieringa observes, software engineering overlaps with engendering trust in Safety. ‘First, it is already difficult to define the future users. requirements of the system, because intended

SMART DISTRIBUTION OF PARCELS

CTIT SCIENCE FOR A SMART SOCIETY

THE COGWHEELS OF SMART SOCIETIES EVER LARGER AMOUNTS OF DATA, STRUNG TOGETHER IN COMPLEX NETWORKS, ARE GATHERED IN TODAY’S WORLD. SMART ALGORITHMS ARE NECESSARY TO MAKE SENSE OF IT ALL. Google’s imperium is built on it; an algorithm that is capable of determining which node in a complex network of information is the most important. This is the website that will end up at the top of its search results. Simultaneously another algorithm analyzes a different network of information to calculate which advertisement to show with the search result. The algorithms that Google, Facebook and the like use for such analyzes are proprietary. But more in general,

research on algorithms for complex network analysis is an important subject for academics who study smart societies – which, after all, are complex networks with enormous flows of information. Determining which information is relevant and how to use it efficiently is crucial. But not trivial. ‘Finding efficient algorithms is like solving puzzles’, says Jaco van de Pol, professor of Formal Methods and Tools. ‘Let me give an example. We had a network problem, which according to theory could be solved only step by step. You could not speed things up by dividing the problem into parts that could be solved in parallel. However, we decided to try anyway. In the end, we found a combination of serial and parallel calculations that enabled us to find a solution faster anyway. That is very rewarding, of course.’

EXTREMIST THOUGHTS The applications of smart network algorithms are manifold: finding the balance between capacity and safety in railroad networks, for instance, or determining which processors in a smartphone to use in which case, in order to maximize battery life. ‘Or to find out how extremist thoughts spread on Twitter’, adds Nelly Litvak, who is an associate professor specialized in algorithms for dynamic networks. There is a huge difference between fixed and dynamic networks, Litvak explains. In a fixed network such as railroads, it is possible to zoom in on specific places and events. What would be the effect of, for instance, adding an extra switch at a certain junction, or of imposing extra security measures at black spots in a trajectory? In dynamic networks, as found in social media, zooming in is meaningless – the moment you do it, the situation may already have changed, because new friend connections were made and others severed. Still, it is often possible to make meaningful statements and even predictions about the network as a whole, for instance, whether the latest hype on Facebook arose spontaneously or was directed. Ideally, such research should combine automated analysis with surveys of the actual situation to corroborate results, but such a multidisciplinary approach has yet to be developed. As fixed networks grow in size it often becomes impossible to analyze them in full, making the statistical methods from dynamic networks more attractive. Litvak: ‘Often enough you can’t do a precise analysis, as time and computing power are limited. In those cases a quick and rough estimate suffices. Of course, you then need algorithms that know how to find a decent estimate.’

DEEPLY HIDDEN Both Litvak and Van de Pol can easily provide more examples for application of their research, ranging from efficient design of wireless sensor networks to effective government information campaigns. However, they understand that the final destination of their fundamental work may just as well be in territories as yet uncharted. After all, research on network algorithms is an abstract art, independent of application. It is about nodes, relationships between them and events propagating through them. Lots of theoretical analysis is possible, though in the end it is always about getting things to work. A little shortcut in an existing algorithm may fit it to a certain practical problem, thereby making it perform better in that specific case. Litvak stresses the importance of better algorithms by comparing them to hardware developments, which

FINDING EFFICIENT ALGORITHMS IS LIKE SOLVING PUZZLES. - Prof. Jaco van de Pol

OFTEN A MODERN ALGORITHM ON A COMPUTER FROM THE NINETIES WILL SOLVE A PROBLEM FASTER THAT A NINETIES ALGORITHM ON A MODERN COMPUTER. - Dr. Nelly Litvak often enough get most credit for advancements in computing power: ‘Often a modern algorithm on a computer from the nineties will solve a problem faster than a nineties algorithm on a modern computer.’ Van de Pol sees algorithms as the unreasonably effective cogwheels of smart societies, deeply hidden in all devices, but still their driving force. The trick is to get them to run as smoothly as possible.

CTIT SCIENCE FOR A SMART SOCIETY

TWITTER IS A GREAT MEDIUM FOR PUBLIC AWARENESS CAMPAIGNS - OR SO IT IS BELIEVED. A COMBINATION OF SMART ALGORITHMS AND SOCIAL SCIENCES PROVIDES INSIGHT INTO THE EFFECTIVENESS. Every second Twitter generates troves of data, as people all around the world voice their opinions on every subject imaginable. Often they even include hashtags in their tweets, providing a useful categorization of the content. No wonder research groups around the globe like to sieve through this data. But professor Ariana Need and associate professor Djoerd Hiemstra have a significant advantage: the cooperation of Twitter itself. They have no need to scrape the data from the internet, but can ask for it directly, saving a lot of time and effort.

preferable to have decent results for every country rather than an optimal overall result that completely misstates the results for Belgium.’ The analysis of Tweets is mainly based on linguistics. Language use differs between age groups and gender. ‘School’ is a word mostly used by youngsters, while ‘daughter’ is more likely to be used by someone over thirty. ‘Beer’ is a male word, ‘sweet’ is female, and so on. Naturally, these classifications are derived from checks against actual data, not from researchers’ intuition.

For the Dutch language there is even a website Their research focuses on the effectiveness of Twitter for (tweetgenie.nl) where visitors can give their Twitter public health campaigns, concentrating on four subjects: address. The site then guesses their gender and age, cancer research, vaccinations, skin cancer awareness subsequently asking for correction if it is wrong. and charity donations. The aim is to establish how such In this playful way visitors help to refine the algorithm. campaigns work out. This in turn may help governments Hiemstra: ‘We’re quite good in predicting the gender and and charities to improve their operations. visitors under 16, because their language is really ‘To achieve this we combine analysis of tweets with other different. For other age groups it is more difficult.’ data’, says Need, who studies sociology of public governance. ‘For instance, we study how information about a certain donation campaign spreads through Other methods for analysis are based on GPS coordinates Twitter and compare the results with information about and smilies that some users add to their tweets. By actual donations that we get from the charity involved.’ linking these to words in the tweets and then searching other tweets with the same words or hashtags, it Ideally, the analysis not only says something about the becomes possible to make educated guesses about propagation of information in general, but also discerns people’s location or mood. opinion leaders and response among different groups. For example, an international campaign like Movember, #VACCINATION the goal of which is to rise funds and awareness for In the past, social research of the kind Need is interested men’s health, would like to know whether there are in, would ideally be based on data retrieved from census differences in the reception of its messages in the records. However, many countries no longer hold countries it targets. censuses. Twitter is a great alternative, she says, not in the least because you can retrieve information any moment you want, whereas a census is by definition an WE CAN DEFINITELY SAY SOMETHING ABOUT FACTORS arbitrary snapshot. FOR SUCCESS AND FAILURE. Actually, the information is even richer. A census and - Prof. Ariana Need information from public health agencies could say something about vaccination rates among different THE GOAL IS NOT TO FIND A NEEDLE IN A HAYSTACK, BUT groups. Twitter adds the layer of motivation, for both TO MAKE AN INVENTORY OF THE WHOLE HAYSTACK. those who have their children vaccinated and those who - Dr. Djoerd Hiemstra do not. It is even possible to follow the interaction of campaign and contra-campaign. On the most basic level this translates to questions like: how many happy smilies HAYSTACK accompany the #vaccination hashtag at the beginning of ‘This calls for smart software’, says Hiemstra, whose the campaign and how does this develop during the research is mostly about search algorithms. For him, the course of time? special interest in the Twitter project is that the goal is ‘It’s great to be able to found social research on copious not to find a needle in a haystack, but to make an inventory of the whole haystack. ‘Traditionally, we would data’, says Need. ‘Of course, this doesn’t mean we can try to get the best average categorization, but in projects predict how effective a campaign will be, but we can definitely say something about factors for success and like this it is more important to avoid large deviations. In failure.’ the Movember campaign, for instance, it would be

PEEKING DEEP INSIDE TWITTER

CTIT SCIENCE FOR A SMART SOCIETY

ROBOT BUDDIES AND SMART CUPBOARDS

SMART AGENTS, BOTH VIRTUAL AND PHYSICAL, WILL PLAY INCREASINGLY LARGE ROLES IN PEOPLE’S LIVES. IT IS IMPORTANT TO DESIGN THEM IN WAYS THAT MAKE THEM FIT SEAMLESSLY INTO SOCIETY.

about theorizing, but I prefer empirical philosophy, where you develop your ideas through observation.’ There is plenty to observe in human-robot interaction, as much on the practical as on the philosophical level. It Just as people are getting used to real-life shops and is important to do both, say Evers and Verbeek. cocktail parties being replaced by virtual agents telling Philosophy is not something that kicks in as a distant them which books to buy and which friends to like, observer after technology has been created, but needs these agents are moving into the real world, in the shape of robots. Visitors to the University of Twente may to be present during the design process. Designers encounter the Campus Robot FROG, greeting them and must be aware of their works’ consequences, at the deepest level. offering information, but this is just the beginning. ‘We will see many more such agents everywhere in the not too distant future’, says Vanessa Evers, professor of Human Media Interaction. Her research focuses on technology that enables robots to interact with humans in a social way, as humans would among each other. One of the studies currently carried out at Evers’ lab involves human-robot symbiosis in learning situations. In this case the teacher is not being replaced by a robot but the machine is given to a child as a learning buddy. The idea is that children comprehend things more easily when they have to explain the matter to another child, or in this case a robot. Evers: ‘We are interested in finding out whether the child establishes a social bond with the robot and, if so, whether this actually benefits the learning experience.’ The approach of this research is thorough. Many hours of video are shot while children perform tasks, some just with an iPad, others assisted by their robot buddy. Specialists in education and behavioural sciences assist in both quantitative and qualitative analysis. The ultimate aim is to design educational methods that optimize learning results.

MURDER MYSTERY In another experiment students were to solve a ‘murder mystery’ with a robot acting as a go-between to gather information from different group members. The idea was to observe the behaviour of humans towards the robot, depending on the robot’s behaviour. ‘We’d like to know which types of robot behaviour humans feel most comfortable with’, Evers explains. ‘Many of these are on the subconscious level, such as how to approach a human and which distance to keep. In the future you’d want robots who understand these rules irrespective of the task you want them to perform.’ ‘From my point of view the interesting angle of these studies is what they say about our relationship with technology’, comments Peter-Paul Verbeek, professor of Philosophy of Technology, who works with Evers in the recently opened DesignLab, aiming to integrate broader aspects into technological design processes ‘The robot in this case does not change a situation by replacing human tasks, but by altering them. Philosophy is often

WE’D LIKE TO KNOW WHICH TYPES OF ROBOT BEHAVIOUR HUMANS FEEL MOST COMFORTABLE WITH. - Prof. Vanessa Evers

SMART SOCIETIES ARE ABOUT OFFERING TECHNOLOGY IN A SENSIBLE WAY, NOT ABOUT INSERTING TECHNOLOGY WHEREVER IT IS POSSIBLE. - Prof. Peter-Paul Verbeek

PERSONAL CONVERSATIONS For instance, projects to encourage the elderly to continue living on their own often feature outfitting their homes with sensors or therapeutic robots. In this way it becomes possible to automatically monitor their wellbeing. Offering robot assisted physical or cognitive therapy is an interesting method to establish whether people are adhering to therapeutic exercise routines. However, though this approach makes sense, it also is a potentially serious invasion of privacy. An incident from a recent project still strikes Evers: ‘The robot turned on a couple of times a day to go through a few simple cognitive exercises with the subject. For this purpose the robot had limited vocal responses, sufficient to guide the exercises. However, the test subject confided a lot of personal information to the robot before starting the exercises, how they were feeling, why they missed their deceased spouse and so on. The robot was simply not equipped to respond appropriately. These highly personal conversations were recorded by the robot. The subject knew this, but we had not anticipated this effect.’ A small observation like this shows the moral dimension technology may acquire. Outfitting senior’s homes with sensors may be a smart way to check on their wellbeing, but it may unintentionally lead to emotional voyeurism of a bereaved elder. According to Verbeek it shows that sometimes it might be wiser to refrain from implementing certain technologies. ‘Smart societies are about offering technology in a sensible way, not about inserting technology wherever it is possible.’

CTIT SCIENCE FOR A SMART SOCIETY

SMART HEALTHCARE FOR THE ELDERLY AN AGEING POPULATION INEVITABLY LEADS TO RISING HEALTHCARE COSTS – UNLESS SMART TECHNOLOGIES CAN KEEP PEOPLE HEALTHIER FOR A PROLONGED PERIOD OF TIME. Given that healthcare, and its related costs, is among the most pressing issues contemporary (western) society faces, it is small wonder that this field is a prominent test bed for smart technologies. Miriam Vollenbroek, professor of Technology Supported Cognitive Training and working for Roessingh Research and Development, concentrates on smart feedback systems for people with chronic disadvantages, such as the elderly. ‘Our healthcare systems are generally based on the thought that you go to a doctor only once you experience severe problems’, she explains. ‘However, those problems build up. If you could detect a physical or mental decline at an earlier stage, you could take measures to prevent further deterioration of your condition.’

Monitoring someone’s physical condition can be pretty straightforward. Smartphones these days mostly contain a motion sensor, primarily meant for detecting the orientation of the screen. This sensor can also be used to monitor the amount of physical exercise taken by its wearer. ‘This is not straightforward’, says Dirk Heylen, professor of Socially Intelligent Computing: ‘If you know nothing about the context, it is impossible to interpret the enormous amount of data such a sensor collects. So you will need to perform a lot of statistical analyses to make sense of it, assuming of course that the data collected is actually workable.’

SCRABBLE Once you have analyzed what the sensor has detected, the really smart part starts: how to automatically encourage people to change their behaviour if the data, for instance, shows that they are inactive for too long a period. Simply sending a text message ‘go out for a walk, now!’ is unlikely to work. Prompts must fit into personal and social circumstances.

Dirk Heylen: ‘The great benefit of big data is that it enables personalization. In the past we had just enough data to find an average for all people, or at best to distinguish between, say, five types of people. Now we can distinguish between people on an individual level and adapt prompts accordingly. On the other hand, this raises the expectations of the system’s accuracy.’

‘Well, at least we noticed’, she smiles. ‘But this incident actually points at a serious bottleneck. By providing many elderly with a monitoring system we risk overloading healthcare professionals with data. So we will also need to build systems that reduce the data generated by thousands of elderly in a meaningful way for their therapists.’

In other words, the massive availability of data not only triggers more analysis, it also raises the bar for the outcome. Systems must be adaptive, constantly adjusting the way they send out their prompts to user feedback, for instance by measuring how often the advice is followed and whether this is deteriorating in the course of time – and if so, whether the individual responds better to more aggressive or more subtle prompts.

Apart from fine-tuning the interplay between end user, therapist and system, there are plenty of other open questions, for instance how to take into account social interactions between elderly and to bring mental health factors like stress into the equation. Heylen: ‘There is certainly no lack of challenges in this field.’

‘One of the things we are exploring right now is gamification’, says Vollenbroek. ‘Elderly people generally like mild competition, showing for instance how they are performing compared to others they know. Some people like puzzles and may like gaining access to new mini games in exchange for exercises. Or they could earn better letters in a scrabble game. In the future I can even imagine people obtaining actual discounts at shops as a reward for completing certain tasks.’

DOG Currently a pilot with a thousand elderly people in the Twente region is running to test this gamification approach, soon to be extended to 44,000 people. This should not only show the technical performance of the systems Vollenbroek and Heylen are proposing, but also their acceptation by end users and healthcare professionals.

OUR PRELIMINARY FINDINGS ARE THAT ELDERLY PEOPLE ARE ACTIVELY ENGAGED. - Prof. Miriam Vollenbroek

THE GREAT BENEFIT OF BIG DATA IS THAT IT ENABLES PERSONALIZATION. - Prof. Dirk Heylen ‘Our preliminary findings are that elderly people are actively engaged’, Vollenbroek observes. ‘They may even be disappointed when they realize their therapist is only looking at their results occasionally. In one case we found that someone had attached the device to his dog, just to test if that would trigger attention.’

CTIT SCIENCE FOR A SMART SOCIETY

THE IMPACT AND VALUE OF SENSORS RESEARCH IN ICT MAY TAKE ON MANY FORMS, FROM THE HIGHLY THEORETICAL TO THE MOST PRACTICAL, BUT THE ULTIMATE GOAL ALWAYS IS CREATING SMARTER, SAFER, HEALTHIER, MORE SUSTAINABLE, AND MORE COMFORTABLE SOCIETIES.

‘After the oil spill disaster in the Gulf of Mexico we thought this could have been prevented or at least the damage could have been minimized with the smart sensing technology that we were working on’, says Nirvana Meratnia, associate professor of pervasive systems. ‘Wireless sensor networks could monitor the oil rig and platform, and timely detect malfunctioning and leakage of the pipes.’ ‘The underwater environment is difficult, harsh and electronic-unfriendly’, she continues. ‘So, before we could use our smart sensing systems, which are meant to operate in air, under water as well, we needed to do more research. The existing commercial underwater monitoring systems are large, expensive, and difficult to deploy. They also do not offer the flexibility that research requires to test different solutions. So we decided to develop our own underwater sensing, communication, and processing platform.’

Meratnia shows the underwater sensor node, which is a steel tube filled with electronics. The node, the cost of which is much less than off-the-shelf systems, contains various sensors to measure water quality and the condition of the structure to which it is attached. There is also a processor to analyze the data locally and an antenna to communicate (acoustically) with other nodes and a base station at the surface. The node has already been tested in a diving centre and a lake in Enschede and a fjord in Norway. After some modifications to the design it is now entering a phase to become part of a pan-European underwater test bed.

MARKET DEVELOPMENTS These underwater nodes are a perfect example of research that immediately shows its market potential, according to Alain le Loux, whose task is to steer technological developments towards valorization. This can take on various forms, like the sale of a patent or the establishment of a start-up company. ‘Usually I become involved once a development is getting close to market application’, le Loux says. ‘In this case, for instance, we have established contact with several engineering companies that could be interested in employing the technology in their operations. I have also alerted one investor known to participate in developments like these, who could become involved should we decide to spin off the probes into a company.’ ‘Supporting valorization is not only about taking research to market, but also about providing feedback to researchers on market developments’, le Loux continues. ‘For example, when I came across sensors that function without batteries, I forwarded it to my colleagues in research, because it might be relevant for what they are developing. By the way, as far as I know we are the only ones working on underwater communications in this way, so we are talking about really unique and hence valuable technology.’

START-UP The underwater platform is just one, fairly preliminary example of the market ready technology coming out of ICT research. Another example, in full swing, is Smart Signs, which provides indoor messaging and guidance using smart sensors. These can be used in hospitals to react to visitors and guide them through the building, or in flexible work spaces to indicate where there is a free desk. The company was started four years ago and has since grown to employ eight people. ‘In this case I coach the company and in the beginning even did some project management’, le Loux says, to explain the intense process that bringing a development to market is. ‘Furthermore I help finding subsidies and investors. Getting support from the European Union also asks for the involvement of industry into an academic project at an early stage. The intensive interplay between research and

valorization cuts both ways, because it helps getting more funding for fundamental research, while at the same time widening the path to applications. ‘Today’s society is unthinkable without ICT. It has become an essential part of societies and is embedded in all aspects of life’, Meratnia concludes. ‘Through our research, we hope to show the great potential and importance of ICT research and its economical, societal, and environmental impacts.’

AFTER THE OIL SPILL DISASTER IN THE GULF OF MEXICO WE THOUGHT THIS COULD HAVE BEEN PREVENTED OR AT LEAST THE DAMAGE COULD HAVE BEEN MINIMIZED WITH THE SMART SENSING TECHNOLOGY THAT WE WERE WORKING ON. - Dr. Nirvana Meratnia

AS FAR AS I KNOW, WE ARE THE ONLY ONES WORKING ON UNDERWATER COMMUNICATIONS IN THIS WAY, SO WE ARE TALKING ABOUT REALLY UNIQUE AND HENCE VALUABLE TECHNOLOGY. - Alain le Loux, MSc MBA

CTIT SCIENCE FOR A SMART SOCIETY

CTIT CENTRES CENTRE FOR WIRELESS AND SENSOR SYSTEMS (WISE)

CENTRE FOR GREEN ICT

The Centre for Wireless and Sensor Systems aims its research at wireless networked systems for sensing and control. Typically, such systems operate in the context of Cyber-Physical Systems, which integrate networking, computation, and physical processes. The research of the centre ranges from radio transmission systems, and wireless network systems to sensing and control, always in close connection with the (extremely demanding) applications. The WiSe centre is focusing its research on systems that are highly distributed and that can potentially scale to an extremely high number of nodes. The centre will focus its research on a few application areas where the specific requirements and environmental conditions of the application put a very strong demand on the supporting system.

The research in the Centre for Green ICT evolves around two topics: efficiency of ICT and using ICT to improve the efficiency of Smart Energy Grids. Next to ICT, in both topics modelling and (on-line) optimization play an important role. The following scientific challenges are defined: • Energy-efficient processing platforms and efficient control software • Smart control systems for flexible energy networks • Resilience in smart grids • Dependable communication networks for smart grids The focus for the coming years within the centre will be on energy-autonomous systems. This includes autonomous ICT systems, as well as autonomous energy related systems.

CENTRE FOR SAFETY AND SECURITY IN SMART SOCIETIES (C.S4)

CENTRE FOR SERVICE AND INTERACTION ROBOTICS (COSIR)

The CoSIR research agenda is focused on the following themes: The Centre for Safety and Security in Smart Societies (C.S4) aims to make the physical world and cyber space • Research methods to investigate real-world use of robot prototypes; a safer place by enhancing (cyber) security, public safety • Novel design of service robots; and system dependability. • Philosophical, social, legal and ethical aspects of human-robot interaction. C.S4 combines core ICT research with the social and behavioural sciences (psychology, sociology, ethics and For the short term this implies the set-up of experiments philosophy) in order to design high-tech systems and dealing with human-robot interaction. It implies solutions tuned to human behaviour and needs. attention for embodied learning and cognition, Dependable ICT will provide the core technology to philosophical, social, legal and ethical aspects of it. detect risks and threats and to protect against these, Furthermore, design methods and prototypes are being whereas the social sciences help to understand the developed to explore new solutions to robots and robot behaviour of users and potential offenders or attackers components, both physical hardware and software. to design technology that is effective, efficient and proportional. Within C.S4, the following themes are actively worked upon: • Cyber-security, including network security, cryptography, and the combat of cyber crime; • Public safety, including intelligent surveillance and access control, and crowd management; • Advanced system modelling and analysis techniques to predict and prevent future faulty operation or maintenance actions of ICT systems; • Dependable system and network design, including architecture and protocols, and reconfiguration and fault-masking techniques to make system more resilient and fault-tolerant. Note that the above themes are interconnected, e.g., activities on social media can arouse uproar that affects public safety, or a cyber attack on an industrial control system can affect public safety. Similarly, an ill-designed e-commerce system can cause monetary loss for companies or a government, or a faulty embedded system might lead to reliability problems in cars.

CENTRE FOR MONITORING AND COACHING (CMC) The Centre for Monitoring and Coaching is a multidisciplinary CTIT/UT research centre that aims to develop smart, innovative technological solutions for applied questions on health and well-being in collaboration with users and healthcare professionals. The main building blocks of systems that enable smart monitoring and coaching are sensing, processing and reasoning, and human interactions. Some of the scientific challenges that the centre addresses in developing smart systems are: • Ambulatory physiological and environmental sensing using wireless sensor networks as input to smart monitoring systems needs to be designed and combined with e-diaries, questionnaires and other subjective sensing (e.g. mood). • User interaction with systems should be friendly, intuitive and persuasive, able to adapt to the environment or momentary context, and should enable efficient and personalized human-computer interaction. • The system should effectively support selfmanagement and behaviour change in people with (chronic) physical and mental conditions, by integrating personal characteristics, preferences and behavioural theories in the technology design process.

CTIT SCIENCE FOR A SMART SOCIETY

CENTRE FOR ARRAY TECHNOLOGY (CAT) CAT is an expertise centre for antenna array technology and processing. Beamforming is the essential functionality at the heart of communication, (radar) sensor and imaging systems which combines signals from an arbitrary number of antenna elements into a limited number of high quality directional beams. Antenna arrays not only replace mechanically steered dish antennas but also offer opportunities to develop new techniques like e.g. multi-beaming and adaptive beamforming, exploiting digital processing capacities. Within the CAT, the focus is on ‘power efficiency’, ’flexibility’ and ‘Electro Magnetic Interference (EMI) robustness’ in the context of antenna arrays. Research on antenna arrays comprises both the analogue and digital domains and requires an integrated approach to discover solutions for multiple applications in e.g. telecommunications, radar and radio astronomy.

CENTRE FOR SUSTAINABLE SUPPLY CHAIN INNOVATION (SSI) The Centre for Sustainable Supply Chain Innovation (SSI) conducts internationally leading interdisciplinary research focused at the design and implementation of sustainable innovation of global supply chains. SSI is achieved by enhancing and implementing breakthrough ICT, operations and coordination methods for planning and control, and organizational theories in complex multi-actor settings. SSI conducts research projects following design science and action based research methods, with a strong foothold in Operations Research and Information Systems, while closely collaborating with industry, knowledge institutes and government agencies. The overarching challenge in the next decades will be to design supply chains that meet future economic, environmental and social requirements, in short that are “people, planet, profit” oriented.

CENTRE FOR HEALTHCARE OPERATIONS IMPROVEMENT & RESEARCH (CHOIR) The Centre for Healthcare Operations Improvement & Research (CHOIR) supports healthcare institutions to organize their business processes to provide highquality care, at balanced costs. Examples are the development and application of information technology and the optimization of logistical processes with stateof-the-art mathematical modelling and computer simulation. These business process improvements often result in lower work pressure and higher employee retention. CHOIR is a multidisciplinary collaboration, combining the following research areas: • Operations research and Business analytics • Logistics and operations management • Purchasing management • Information technology and change management • Quality and safety management The researchers involved have a broad expertise, originating from a tradition of research and innovation in industry, telecommunications, transportation and ICT.

CONTACT CTIT Address: University of Twente, CTIT P.O. Box 217 7500 AE Enschede Telephone: +31 (0) 53 489 8031 / 489 3994 E-mail: office@ctit.utwente.nl Website: www.ctit.utwente.nl

VISITING ADDRESS CTIT is located on the campus of the University of Twente, building Zilverling, Drienerlolaan 5, Enschede.

CREDITS Texts Christian Jongeneel Photography Gijs van Ouwerkerk Design Eric van der Wal; Station Noord

University of Twente, CTIT P.O. Box 217 7500 AE Enschede T +31 (0)53 489 8031 / 489 3994 office@ctit.utwente.nl www.ctit.utwente.nl