Section: New Results

Belief propagation inference for traffic prediction

Participants : Cyril Furtlehner, Yufei Han, Jean-Marc Lasgouttes, Victorin Martin.

This work [57] deals with real-time prediction of traffic conditions in a setting where the only available information is floating car data (FCD) sent by probe vehicles. The main focus is on finding a good way to encode some coarse information (typically whether traffic on a segment is fluid or congested), and to decode it in the form of real-time traffic reconstruction and prediction. Our approach relies in particular on the belief propagation algorithm.

These studies are done in particular in the framework of the projects Travesti and Pumas.

This year's highlights are

• A paper describing a new sufficient condition for local stability of the Belief Propagation algorithm has been published and presented in an international conference [38] .

• The work about the theoretical aspects of encoding real valued variables into a binary Ising model has been summarized in a publication currently under reviewing process.

• Ideas about adding macroscopic variables within the Ising model are currently being tested using the software BPstruction developed last year.

• Victorin Martin has given a talk at the “Colloque Jeunes Probabilistes et Statisticiens” at CIRM, where he presented his work one the design of a latent Ising model for real valued inference.

• Cyril Furtlehner, Yufei Han and Victorin Martin presented the work done in the Travesti project at the workshop on inference organized by Inria and Mines ParisTech (see  9.1 ).