A smart data product to monitor traffic
The city of Pully and Swisscom have engaged 18 months ago in a co-creation project that aims at building a smart data product, which provides continuous measurement of the traffic situation in downtown Pully. The project is called “Mobility Observatory”.
The innovation and development challenge is that we are not using GPS data to measure the traffic, but anonymized and aggregated network traces generated by Swisscom’s mobile network (see also my first article on this topic). We capture all the interactions created when a smartphone “talks” to a mobile communications antenna, such as phone calls, SMS, app synchronizations, etc. and transform them into traffic indicators.
We, Pully and Swisscom, had the chance on the 6th of December to present the outcome and the learnings at the 5th Smart City Conference organized by the Swiss Federal Office of Energy in St. Gallen. We shared with the audience the following two points about the project:
- What have we achieved?
- What have we learned?
Understanding the traffic dynamic in a city center
The web based platform, which we have developed during the project, enables Pully to measure the traffic in its city center. After 6 months of monitoring, we can draw the following conclusions:
- Pully has known that the traffic was high, with a weekday average of 10’000 cars per day in the area of observation. As a new insight, we have been able to categorize this traffic and measure precisely the transit part, which is when a car drives through the area of observation without stopping. In September, this part was equal to 54% of the total traffic.
More than half of the traffic downtown Pully is generated by cars passing through the city center without making a stop.
- By monitoring the origin and destination of this transit traffic, the second insight is that a large share of it is local.
More than 35% of the traffic going through the center has its origin within Pully or the two adjacent areas, which are downtown Lausanne and Lutry.
Since a big part of the traffic originates locally, the city is in a position to act on it accordingly and is able to take the right measures to reduce the transits. With these new insights, Pully will be in a position to justify upcoming infrastructure changes.
As a result, the city might remodel its urban setup in the city center and adapt its public transport offering, working towards having a better balance between cars, public transportation and soft mobility. The modifications should improve the attractiveness and quality of life in the city center.
Furthermore, this newly developed tool will allow to observe if city attractiveness increases over time.
The outcome of our project is a new tool that supports data-driven urban infrastructure decision making and measuring the impact of those decisions.
The project learnings
Under the concept of “Smart City”, municipalities are going through a profound digital transformation. In the case of Pully, the digital technology being used is “Big Data” and the objective is to increase city attractiveness.
We believe that to further accelerate and increase the success of other Smart City projects, sharing project experience is key. This point was also highlighted at the Smart City Conference in St. Gallen by Rogier Havelaar from PostNL, who came presenting his experience with engaging in the Smart City program of the city of Amsterdam.
We have structured our project learnings by using the Smart City Piano, the analytical tool Swisscom has developed together with the Global Center for Digital Business Transformation at IMD Business School.Let’s summarize the key findings along the three chords “assessing the potential”, “preparing the groundwork” and “managing the execution”.
Assessing the potential
Our project has two technology drivers
- Swisscom’s 4G mobile network, which covers 98% of the Swiss population. Further, smartphone penetration is high in Switzerland. As a result, more than 3 terabytes of data are generated every day, enabling us to reconstruct precise mobility patterns.
- Big data languages and frameworks such as Scala and Spark allow us to manage a flow of more than 600’000 data points per second.
Since it is a brand new way of monitoring mobility, finding the appropriate numbers for the business case is difficult. However, we could demonstrate that data acquisition costs were not higher than traditional traffic measurement methods, while adding substantial new benefits.
For example, creating origin-destination matrixes are currently based on surveys. However, such surveys cost a lot of money and their study samples have a limited population representability, due to the typically low reach out to about 2% to 5% of the total population, while Swisscom’s network reaches up to 60% of the population, which equals our market share.
Preparing the ground work
The project was validated by the entire executive team of Pully. Furthermore, a city councilor is part of the steering committee.
We work with anonymized and aggregated data. Swisscom does not share raw data with Pully, only traffic indicators are presented on the application.
Managing the execution
The city of Pully has a dedicated and transversal coordinator for special projects. The project was also structured in a way that a two level communication and exchange is ensured: Project team meetings and steering committees have been set up. Project meetings were organized according to the Scrum methodology, which was as well selected for the software development of the tool.
We have formed a team that combines two required skills, which are urban planning and data science. This skill combination was necessary, since it has enabled the data scientists to develop the right algorithms that match the needs of the urban planner. It was especially necessary that they agreed together on the basic assumptions and definitions of traffic measurement, such as the definition of a commuter, the definition of origin etc.
Since we are dealing with sensitive data, it was key to have a proper communication plan and get the acceptance of the citizens for the project. The purpose of the whole undertaking, which is improving Pully’s city center attractiveness, was well communicated and the specific concerns about data privacy have been addressed.
The development of this digital tool was also complemented with a participative approach that included the population to understand how it wants the city center to evolve.
Our next destination
Moving forward, we have formed an academic collaboration with EPFL, which supports us in calibrating our new tool and measuring its accuracy level. We aim reaching this with data fusion, where we want to use additional data from road sensors or from cameras and map it with our network trace data.
To summarize, we have started a project that doesn’t require an additional infrastructure deployment and which uses a single source of data – a sensor that everyone has in his pocket – the smartphone. In addition, we have successfully set up the project in a co-creation mode, where Pully and Swisscom both actively participated and developed the digital tool. Finally, it is important to mention that our project hasn’t ended yet, since a digital transformation is a journey and not a sprint. We not only continue to enrich our dataset, but also have teamed up with a new academic partner, the EPFL.
You can find below the slides that Mark Zolliker, Alexandre Bosshard and I have presented at the conference.