Section: New Software and Platforms

SoundCity: Urban-scale Noise Monitoring using the Urban Civics Middleware

Participants : Valérie Issarny [contact] , Fadwa Rebhi, Animesh Pathak, Sara Hachem.

URL: http://urbancivics.com/soundcity_app.html

The interest in noise pollution and its effects, especially in terms of its influence on citizens' health, has been increasing throughout the years, as illustrated by the plethora of available research and health studies. Traditionally, cities perform periodic noise monitoring to assess the noisiness of neighborhoods and update their city planning policies accordingly. However, in its current state, noise monitoring for city planning takes place once every several years and leverages only a few static sensors (e.g., BruitParif activity, http://www.bruitparif.fr ). Additionally, the outcome is not representative of citizens' personal exposure to noise as it is coarse grained and requires preplanning to determine where and when the sensing should take place. It is worth mentioning that, even if we assume that the city administration is able to use applications running on the smart phones of the city employees, gathering data by following a pre-determined spatio-temporal schedule – although an excellent strategy in terms of the quality of data gathered – would be expensive and cumbersome to organize.

Noise sensing is also complemented by data modeling approaches to generate maps that allow citizens and governments to visualize the distribution of noise over a region of interest. Those techniques range from basic data plotting to data assimilation techniques that leverage large scale data analytics further enabling the simulation and estimation of missing noise values. However, assimilation techniques are usually restricted to static sensing with predetermined locations and sensing times. This approach limits the capacity of assimilation models to provide estimates with high accuracy.

A solution to the above issues is through mobile participatory sensing, where noise monitoring is no longer restricted to a few expensive sensors. Participatory sensed data can be then leveraged, along with data provided from static sensors, by the assimilation techniques to further reduce the errors in the simulated maps. Toward that goal, we have been building the Urban Civics middleware, which we have further customized for noise monitoring [20] . The refined Urban Civics comprises middleware solutions for noise sensing, crowd-sourcing and data assimilation with the data assimilation component being more specifically developed by the Inria CLIME team. The noise-sensing itself is performed through existing dedicated applications, integrated with our middleware. Such approach enables us to alleviate the burden of domain-specific development and exploit the knowledge of domain experts.

SoundCity and supporting Urban Civics middleware are developed in collaboration with the Inria CLIME team, Ambientic (FR) and the Civic Engine at Berkeley (USA) in the context of CityLab@Inria and Inria@SiliconValley.