Street congestion is a problem every commuter faces. Unfortunately, getting live congestion data requires internet access and you might already be stuck when you get the data. The question that comes to mind is whether we can predict congestion in advance. Most traffic jams in large cities happen every day at certain locations at some predictable hour. Using live data to predict congestion hours in real time or use the network structure itself to predict problematic sections before they even happen might provide us with a solution to this problem.

This work explores congestion patterns on Chicago and Anaheim road networks using the congestion barcode algorithm, which it then uses to characterize congested areas. At the end it presents a new approach to detouring vehicles in case a road becomes unusable, which unfortunately does not provide any computational benefits.

Our paper is available on https://goo.gl/2sPDPZ