Recognizing Traffic Jams with Hovering Data Clouds

Many complex structures in our modern world exist independent of the individual entities they are composed of, giving them an "organic" quality. Important examples include traffic phenomena, e.g., traffic jams; despite of strong efforts over many years, centralized computing has been unable to deal with the resulting problems in a satisfactory manner. With the growing power of sensing devices and wireless communication, participants in traffic are no longer restricted to display passive, particle-like behavior; instead, local data exchange makes it technically feasible to aim for decentralized coordination between cars. One fundamental concept for making use of these possibilities comprises Hovering Data Clouds, which consist of relevant information that is kept by ever-changing carriers; a prototypical scenario arises in a traffic jam, where data is maintained by passing it on to newly arriving cars. In this study, we present algorithmic methods for this concept. This paper is part of project AutoNomos 1 (www.auto- nomos.de), which aims at traffic control in a decentralized manner.