Location data gives smart answers to smart cities

Location data gives smart answers to smart cities


Toivo Vajakas, Juri Jagomagi, Mihkel Jarveoja, Ulo


Cities grow – fast and often uncontrollably. The population of urban residents is estimated to increase 85% by 2050. This convergence of people, creation of city conglomerates and suburbanisation causes complex movement of people in and between cities as well as excessive commuting. Are city governments and planners prepared for this turbulence in mobility and problems it could cause for transportation and logistics? Do they foresee the future of cities? One might argue if such a complicated system as evolution of urban areas is predictable in the first place. Leaving aside theories and paradigms then from the practical point of view there is an urgent need for adequate predictions and assessments.

The data and knowledge which is nowadays widely used for strategic decisions is bulky, inadequate, expires quickly and might lead to misinterpretation. We believe that mobile technology is the best solution to handle mobility data and to make cities smarter by using citizen and visitor experience. This knowledge should be available for city planners, decision makers and citizens.

Redefining mobility

While analysing mobility it is rather usual to ask questions like “How many” and “How often”. The answers to those questions tend to be simple but at the same time clearly too primitive to draw any conclusions. Instead we should be curious where do these people come from, where are they heading to and what is most important – why are they there? These are questions that can´t be answered by barely counting people. While defining and explaining mobility, we need to take into account indications such as aggregation, trajectory, intensity, circadian and seasonal variability, relation to topography and infrastructure etc. The knowledge obtained from such combined analyses is especially valuable in terms of planning and managing urban space.

Mobile phones as sensors

There is no need to build additional surveillance infrastructure to carry out this ambition. In digital world we all leave behind footprints which are stored and can be processed. Mobile radio access network produces huge amount of location data. With existing technology it is possible to use all mobile-phone owners as sensors and that could be done almost in real-time, if necessary. A single cell phone, even when it’s idle, registers its location to the network at least once in every hour. What is important is that this function cannot be turned off as long as mobile phone is working. In addition, every data transmission or calling event registers the location of the user. For example a city with one million inhabitants can generate quarter billion location updates per day. No static sensor could provide us so adequate and up-to-date information about citizens’ movement. To make this huge amount of location data valuable and safe asset for city governments and citizens, it has to be generalized and combined with other spatial datasets.

From single user to location semantics – how (do) we do it?

The raw data that is registered in cell phone network reflects single user behaviour and the location of user is determined by the closest mobile antenna. To turn single location event into general pattern of movement, we use probabilistic methods to evaluate users’ trajectories and anchor points (home, work etc.). Mobile location data is often not good enough to calculate accurate single trajectories but its spatial resolution is sufficient for urban planning.

Data mining and different algorithms enable to find correlations between different phenomena, evaluate citizens’ mobility in various aspects and spatial resolutions. One outcome of mobile data enrichment process is origin-destination (O/D) matrices which illustrates people’s movement from one area to other through different aspects of time and speed. To put it simple, you can pick any place – for example problematic traffic junction, and answer to the questions like where all these people, at specific time, come from and are heading to.

Heatmap illustrates the origin of people at the indicated place. Mobile positioning helps to answer the question where do these people come from? Source: Reach-U sample data
Heatmap illustrates the origin of people at the indicated place. Mobile positioning helps to answer the question where do these people come from? Source: Reach-U sample data

Statistical methods estimate hypothetical approximate locations for mobile phones, which deviate from real locations in physical world. Deviations can be reduced by statistical detection of “unreasonable” movements that are most probably wrong. The final analysis is usually conducted by dividing the area of interest into smaller homogeneous regions (e.g. residential area vs city centre vs suburban vs rural) and aggregating the data per region. The aggregation reduces effects of spatial uncertainty and also facilitates privacy preservation.

O/D matrix is excellent tool for assessing the effectiveness of changes made in urban space. Comparing “before” and “after” situation we get reliable feedback to our planning actions and decisions.

Analysing mobile network data alone is not enough to deduct the reasons for mobility. Perhaps the most important question “Why?” remains, but it can be addressed when combining location data with other spatial datasets. Adding semantics to raw location data broadens our view of how urban areas evolve in practise.

Privacy and problems in mobile positioning

Mentioning surveillance and real-time positioning of citizens might sound frightening to a society where privacy is important. As mentioned before, mobile positioning is not for detecting single trajectories. Location data obtained from mobile network won’t be traced back to a single user, is highly secured on a person level and actually never leaves from the mobile operators’ server, where it is stored for technical reason anyway. All results from data processing are shared externally in aggregated mode and this anonymous information does not jeopardize person’s security and privacy. We acknowledge the concern about privacy and realize it is an emerging matter for all modern technologies, but using methodology mentioned above this threat is minimised.

However there are several other issues to consider. First of all, spatial and temporal coverage is limited, which makes it difficult to observe short-term movements. Raw location data contain a lot of noise and distortions which have to be removed with specific algorithms. From our own experience we can say that processing the huge amount of data is another big challenge. Most of common data collecting and processing tools are not working here and integration of data processing and mobile operator servers is time-consuming and expensive.

Case studies

Nowadays mobile positioning is already widely used in different applications but it is clearly underutilized in urban planning and citizen experience. Reach-U has been active in this field for 15 years already. In cooperation with Ericsson and Telia Sonera we have created one of the first positioning services in the world for locating emergency callers in Estonia. Beside public safety and security we have used mobile positioning data in various fields such as network customer experience, intelligent transportation systems, location based advertising, workforce management and other location based services.

DemograftTargeter is one of the application developed and designed by Reach-U. It provides enhanced profiling of citizens by using mobile positioning data. Here are visualized abundance and intensity of citizens in a defined area (black circle) and specific time(8:00-17:00) during one week.
DemograftTargeter is one of the application developed and designed by Reach-U. It provides enhanced profiling of citizens by using mobile positioning data. Here are visualized abundance and intensity of citizens in a defined area (black circle) and specific time(8:00-17:00) during one week.

One of the most extraordinary solutions we have built on mobile positioning data helped to guide lost pilgrims to safety. Every year hundreds of people get lost, and in worst cases die, in the deserts of Saudi Arabia while performing the haj, the once-in-a-lifetime Muslim pilgrimage. Our tracking solution doesn´t need expensive phones, GPS-devices or internet connection and is usable with cheap handsets, which the government would give to the poorest. The solution was implemented within less than a month and was capable to handle emergency requests from more than 90,000 pilgrims.

At the University of Tartu, Estonia, aggregated mobile positioning data has successfully been used in various research topics such as activity spaces, travel behaviour, tourism and segregation. This is definitely a step closer to incorporate mobile positioning data for serving community.


For modern cities it is not anymore privilege but necessity to be smart and to start predicting the future. Due to rapid demographic growth and urbanization, old monitoring and predicting methods tend to fail or produce inadequate and misleading results which lead to unreasonable decision and loss of money, time and citizens’ trust.

Mobile positioning technology integrates citizen experience with state-of-the-art data handling by using already available information and making it more valuable. Desiring or not, all mobile users provide their network operator with almost real-time location data, which makes cell phones one of the best sensors for monitoring users mobility.

Aggregated and contextualized location data can answer questions that really matter. Where to plan big infrastructure changes? How to allocate road maintenance budget? What are the most common moving patterns of tourists? It helps to see reasons behind more or less obvious problems and bottlenecks in urban space.

Although the data is “there” and ready to be used, we need to keep in mind challenges their processing poses. And we need to frame our fields of interest because answering a question is reasonable only if we can really do something with the answer. Smart cities need smart questions.