Real World Applications Using Parallel Computing Techniques in Dynamic Traffic Assignment and Shortest Path Search

Abstract

As the range of applications for Intelligent Transport Systems (ITS) grows wider, the efficiency of the underlying tools for Big Data Analytics becomes of crucial importance. Smart Cities are able to monitor, forecast and (possibly) control the pulse of collective interactions involving networks and environment (such as traffic and pollution) by means of key performance indicators. Technology-guided solutions can proactively support the sustainable development and the optimal management of infrastructures and services, improving the quality of life for both city dwellers and commuters. This requires processing huge amounts of data, continuously streaming in from a variety of fixed sensors (e.g. loops, cameras) and mobile devices (GPS trajectories). In particular, Mobility Control Centres need effective software solutions and fast algorithms to deal with two major problems: Traffic Forecasting and Route Guidance. This paper presents real world examples of large scale applications where both tasks are addressed by implementing parallel computing algorithms, achieving high performances and allowing real time management operations and end-user services. The first test case examines the performance of a routing platform covering the entire Austria region, while the second concerns large instances of dynamic traffic assignment for real-time forecasting.