With more than half of humanity living in cities, there is an urgent need to improve urban planning, design and management.
Until now, policy makers and planners have struggled to keep up with trends. Changes were too fast, and variables too dynamic, for conventional surveys or censuses to capture them adequately.
The new field of big data analysis allows a better understanding of how large numbers of people interact with their built environment.
With lots of number crunching, industry-research collaborations and some imagination, previously unseen ‘big picture’-level perspectives can emerge.
This capability is going to be critical for South Asia. Already, three out of 10 South Asians live in crowded cities, and a massive rise is expected in the coming decades. Between 2014 and 2050, for example, India is projected to add 404 million city-dwellers. 
Business as usual will only worsen the current urban chaos. Haphazard planning based on obsolete or patchy data is no longer an option.
Governments are responding to this challenge. In mid 2014, India announced an ambitious ‘100 Smart Cities’ programme. Under this, state capitals, as well as many tourist and heritage cities are to have a rapid upgrade of urban infrastructure. 
There is no universal definition of smart cities: it can mean smart utilities, smart housing, smart mobility or smart design. India’s smart city programme starts by asking why so many people migrate to cities — primarily, they are in search of employment or economic activities, and a better quality of life. How can everyone’s expectations be met without overwhelming systems? 
It is relatively easier to build totally new smart cities. But existing cities that have organically developed need better management more urgently. Our ‘stupid cities’ are simply too unhealthy.
Smart cities use information and communications technologies to enhance feedback loops within the complex systems that make up a city. If processed properly, the steady flow of data can vastly improve the design of ‘hard’ physical environment and the provision of ‘soft’ services to citizens.
“Business as usual will only worsen the current urban chaos. Haphazard planning based on obsolete or patchy data is no longer an option.”
This is where big data research can play a key role.
The digital age has opened the floodgates of data, but much of it still lies scattered, unprocessed or in formats beyond the use of planners, resource managers and citizens.
Big data involves analysing massive volumes of data coming in fast, often in real time. Such data is generated through our increasing use of digital devices and web-supported tools and platforms everyday. Volume, velocity and variety are the three ‘Vs’ that characterise big data. A fourth ‘V’ is the value that could be extracted from them. 
Until recently, only governments and large corporations could afford supercomputers needed for big data analysis. Now — with increased computer memory, retrieval and processing capabilities and lower costs — the tools are finally within reach of developing country researchers. The right kind of collaborations can derive many insights useful for public policy and planning.
City planners need to know where people live and congregate, when and how they move, their economic conditions, where they spend their money, and about their social networks. Currently the best big data source for these variables involves mobile phones — ubiquitous device used by the rich and poor alike.
Mobile network big data (MNBD) is produced by all phones, smart and otherwise, and include call detail records (CDRs) generated when calls and texts are sent or received, web is accessed, and prepaid values are loaded. When analysed, these can provide a good (though not perfect) proxy for people’s behaviour in an area at any given time.
The Colombo-based think tank LIRNEasia has just conducted some path-breaking research using MNBD.  They analysed pseudonymised, historical CDRs from multiple mobile operators to understand and monitor land use, congregations of people, peak and off-peak travel patterns, communities, and traffic. Correlations were validated using other datasets where available, for example, a conventional transport survey.
To illustrate findings, LIRNEasia created ‘heat-maps’ showing how Colombo city (resident population: 550,000) acts as a ‘sink’, sucking people from the surrounding suburbs on weekdays.
“This gives real-time insights on the geo-spatial distribution of the population, which can be used by urban planners to manage things like traffic flow in a city,” says Rohan Samarajiva, founding chair of LIRNEasia. 
MNBD allows tracking and mapping of daily changes in population densities relative to midnight (‘home location’). Daytime or ‘work’ locations may be identified, along with where people came from.
One insight: the northern part of Colombo, where the poor are concentrated, shows a lower density at mid-day on weekdays relative to midnight. This is due to large numbers crossing to the southern part to provide labour. Similarly, various types of land use can be discerned by analysing the daily loading patterns of base stations. This, in turn, enables a closer alignment of urban plans and actual land use. 
City planners like to know how people actually use specific urban environments. Samarajiva says mobile use patterns — interpreted with other local knowledge — can be leveraged to determine land use patterns. When people are found staying at a location for a significant time, adequate resources and services can be assigned to that area.
For now, MNBD is the only dataset that can provide comprehensive insights for urban planning and other public purposes that do not marginalise the poor. However, other datasets — such as those from Twitter — may also be used for specific purposes, with care.
Policy relevant big data research relies on trust-based collaborations between data custodians and researchers. LIRNEasia carefully negotiated access to telecom operator data to avoid any breach of user privacy or commercial confidentiality.
Samarajiva says his is the only research organisation in South Asia or developing countries that is working on mobile big data. Building on the Sri Lanka research, LIRNEasia plans to collaborate soon with partners in India and Bangladesh.  
“Good big data research also requires multidisciplinary teams made up of those willing to converse across disciplinary silos. Data scientists and domain experts who have this skill are in short supply,” he adds.
The benefits of big data analysis are not limited to planners and policy makers. When coupled with user-friendly web portals or smartphone apps, it can also help citizens make good choices, fast.
An early adopter is the Bengaluru Transport Information System (BTIS) started by Mapunity, an organisation that develops technology to tackle social problems and development challenges.  BTIS aggregates data streams from multiple sources (from mobile network base stations, CCTV footage, GPS sensor data from public transport, and crowd-sourced data from the public) to serve information needs of different users. 
Citizens can use this map-based information service to plan travel routes avoiding traffic congestion, and for car-pooling or paying parking fees. Planners get to see the ‘bigger picture’ with which they can attempt real-time trouble-shooting as well as long-term changes.
Collaboration is the glue that binds things together. Smart cities will work well only when politicians and planners find new ways of working with researchers and communities.
Nalaka Gunawardene is a Colombo-based science writer, blogger and development communication consultant. He is also a trustee of SciDev.Net. The views in this column are his own.
This article has been produced by SciDev.Net's South Asia desk.