The concept of the Smart City is driven by data. By sensing and connecting and analyzing every element of city life, new insights can be derived that automate processes, improve the life quality for citizens and help elected leaders make better decisions. It’s a compelling vision but implementation is not so simple. Here we look at the example of traffic analytics.

Edge_processing

The economics of data
Generally speaking, citizen and business taxes pay for everything in a city, and there are a lot of important services to pay for. So any technology investment, no matter how transformative, must also deliver cost efficiencies for tax payers.

When it comes to smart city projects, most people think of the one-off cost of sensors and network equipment.  But it is the unseen recurring costs of moving huge amounts of raw data into the big data platform, and the ongoing system integration costs of combining, analyzing and visualizing massive amounts of diverse data that are the real cost culprits. The good news is that a flexible approach to when and where to process and transport data can make a big difference.

Take the example of streetlight-mounted cameras for analyzing traffic flow:

  • Raw data: the cameras produce a constant stream of video data that demand a high bandwidth (and high cost) connection to the cloud.
  • Small data: by placing processing power with the camera (known as edge-processing), software can extract the useful data from the mass of raw video. Traffic count per lane, the mix of different types of vehicles are two examples.  This distilled “small data” is simple and affordable to transmit to the cloud.  Streetlight management systems can use the traffic count information to adaptively dim and brighten lights according to weight traffic. Another benefit is that all vehicle and personal identifiable data is discarded.
  • Big data: the big data platform can add value to the small data by combining it with information from many other sources. The cost of integrating those sources is proportional to the amount of data and how it is structured.  By only supplying small amounts of distilled data, these costs can be dramatically reduced.

Existing systems reflect the way cities are organized
Typically, there are two main layers.  First, there are departments with specialist expertise in a particular function – such as street lighting.  This is where day-to-day operational decisions are made and where established expert systems not only monitor but also take action, often automated actions.

Overseeing these departments is the municipal executive leadership.  This is where big data analytics platforms tend be deployed.  By painstakingly combining data from multiple departmental systems and direct sensor feeds, trends can be identified and dashboards can connect leaders with what’s really happening on the ground.

But this 2-layer organizational arrangement, which is remarkably similar all over the world, makes it very difficult to impose a one-size-fits-all big data platform that does more than monitor and analyze, because it doesn’t tend to be operated by the departments with expert domain knowledge.