Smart Cities & IoT
Today, the comprehension and the anticipation of our displacements on territories is a major stake of our relation with the environment: costs reduction (deployment and maintenance of transport services, ticketing, etc.), greenhouse gases emissions reduction, optimization of mobility, etc. Many social-economic, political and ecological issues are related to the management and understanding of our movements. In addition, more and more data representing our displacements are generated , and the simultaneous use of these heterogeneous data is a research axis in full development .
The objectives of this research project are in the field of human mobility and smart city. Two strong themes worn by the city of Marseille for several years. Our goal is to increase the momentum and influence radiated by Marseilles in the use and promotion of intelligent data and our project therefore encompasses the main processes associated with the processing of this multi-source and multi-modal data. It is about defining, studying, optimizing and adapting all the steps of the chain of treatment and analysis of large masses of data extracted at multiple levels and by multiple channels: Web, social networks and wireless, connected objects, sensors and scientific applications. The ultimate goal is to design pipelines and computer models for the structuring and analysis of these heterogeneous data, with concrete applications particularly in the field of human mobility (optimization of the travel flows, understanding human dynamics, etc.) and smart city.
We seek to use all the currently available data, either open data , as specific data from the smart city: smart-grids, road sensors, ticketing, Floating Car Data, radio networks, drones and UAVs. These data are supplemented in particular by data coming from the Internet of Things (IoT): data from sensors, data of radiotelephony (2345G), access by wireless technologies (Bluetooth, WiFi), data of networks, and data more traditional (surveys). These data can be collected by setting up various industrial partnerships, with recognized regional actors. Our final objective is to transform, clean and enrich the data, detect and correct the anomalies effectively, extract knowledge from these heterogeneous data and finally propose visual analytics artifacts for a better understanding of urban dynamics.
This research work aims to define a common pattern for the extraction and structuring of multi-source and multi-modal data. The output is to better define and control the quality of these data, but also to specify the most suitable analytical pipeline to data and users. We understand that a better use of data from a smart city allows a better understanding of mobility dynamics in urban centers. Local structures can then benefit from these advances in order to better plan, adapt, and scale the infrastructure of their smart city. The city of Marseilles, already very present in the field of the use of intelligent data, will benefit from the results of this project, particularly in the understanding of mobility in its territory.
 E. Thuillier, “Extraction of mobility information through heterogeneous data fusion: A multi-source, multi-scale, and multi-modal problem,” Ph.D. Thesis, SPIM Doctoral school, UTBM, Paris, France, 2017.
 Plan d'action ANR 2018, B.6 – Axe 1, "Mobilités et systèmes urbains durables", Plan d'action ANR 2017, DEFI 6, "Mobilité et systèmes urbains durables"
 OpenDataSoft: https://public.opendatasoft.com/explore/?sort=modified
Internet Of Things
The Internet of Things (IoT) is defined by the RFID group as “the worldwide network of interconnected objects uniquely addressable based on standard communications protocol” . The IoT represents a large network of physical devices. The amount of data can be Terabytes, petabytes and even zettabytes . These data sources are heterogeneous (structured, semi-structured and unstructured) and diverse (sensors data, social data, etc.).
Specifically, there are a lot of challenges in analyzing IoT big data: vast amount of raw data, high dynamicity, context-aware, data quality etc. These challenges require new algorithms and techniques for developing applications that are able to (i) collect data from the numerous data sources of IoT, (ii) querying over these collected data and (iii) Making descisions. These algorithms should be distributed to allow them to run on top of Big Data infrastructures.