Abstract : This thesis is divided into four parts. The ﬁrst part
is dedicated to the literature review of different models of traffic currently used in the domain of urban network traffic simulation. A new classification based on four criteria is presented for distinguishing the various traffic flow models.
The second part of the thesis studies the decomposition of large scale urban networks. The objective is to decompose the network into sub-networks of limited size while minimizing the interconnecting or boundary nodes among the sub-networks.
In the third part of the thesis, we present a traffic simulation software which has been used for generating the data used to design a large memory. This memory is used to provide the fastest path between an origin and a destination of the sub-network given the initial state of the arcs of the subnetwork and the origin-destination node of the vehicle.
The last part of the thesis is the most innovative one. We show how to use data analysis techniques to design the memory and thus allows real time decision making.