Article

Deciding Who Counts in the Smart City

Summary of the presentation by Rachel Franklin

On the 1st of February 2023, Rachel Franklin came to AMS Institute to talk about sensors in the city. Her research's initial goal was to monitor New Castle's air quality. However, during this session, she dove into the use of sensors as a whole.

Ubiquitous

Sensors are ubiquitous. These days, many sensors are present in New Castle alone. They range from Air Quality Sensors to Linux Node Controllers (Image Processing Computer & System Health Manager). The concept of sensors is often associated with surveillance technology and the loss of privacy. However, the truth is more complex than that. Depending on their purpose, object and location, sensors can indicate inequality in the city.

Sensor deserts 

The purpose of sensors in a city can stem from the desire for a city to be relevant. In this case, the sensor would serve a marketing purpose. In addition, what is being measured is equally as essential to consider as the purpose of sensor placement. When the sensor's purpose is to measure air quality, for example, producers and hazards are often measured. Not the exposure of people to different qualities of air.

In the case of New Castle, you can find sensors in more wealthy areas, which entails more knowledge about air quality in those places. For the areas that lack sensors, data is limited or unavailable. This leads to more uncertainty and a gap between areas of knowledge about air quality (sensor deserts). The more sensors there are, the more precise the gathered data will be. (In the presentation in this collection, you can find the Index of Deprivation to see which places contain sensors for air quality and which do not.)

Trade-offs

The reason for the gap in knowledge between areas is the use of a single objective greedy algorithm. This means that sensors are placed one by one and only one subgroup is accurately measured.

Similarly, there is the multi-objective genetic algorithm. This algorithm generates a spectrum of networks representing coverage and trade-offs between different sub-groups.  

The first decision you need to make when working with the multi-objective genetic algorithm is what kind of people you want to measure. The reason for this is that if you want to measure most of the population, you will need a lot of sensors. When you measure, for instance, people who work at an office, you will lose data about others in the population because part of the population that goes to an office uses a specific route that many others (like the elderly or children) will not follow.

Aside from the sub-group you want to measure, you also need to make trade-offs regarding the sensor's quality. When the quality of the sensor is low, the price will also be low. Hence you can install more sensors and accordingly gather more data. However, the results will be of lesser quality.

Closing the knowledge gap

We can find a compromise in the genetic algorithm. This algorithm exists out of 200 algorithms, selecting the best networks that fit the criteria. Covering half of the older population, half of the child population and half of the work population, the algorithm is a more suitable alternative for the current network of sensors in New Castle.

More insight is needed to help all people benefit from the presence of sensors. Researchers developed the Air Quality Placement Decision Support Tool as part of the Spatial Inequality and the Smart City project. This tool shows a map with which the visitor interacts. On the map, you can see where in New Castle air quality sensors are currently placed. As importantly, you can see what parts of the population are now covered by the sensors. It is also possible to see what the coverage would be if you moved the sensor to a different location.

Takeaways

Overall, the key points from this session were:

  • If you want to use sensors, first consider what their purpose is in the city. The location plays a big part because you will have to choose a different location based on if you want to look at people or hazards in the city.
  • Look at geography when considering a place for a sensor. By deciding to gather data about a specific part of the population, you exclude another.
  • Mobility is important to consider. The population is not static, so it is crucial to think about where people are during different parts of the day.

Image credits

Header image: Air Quality Sensor - Canva.png