Herakles: A system for sensor-based live sport analysis using private peer-to-peer networks

Abstract

Tactical decisions characterize team sports like soccer or basketball profoundly. Analyses of training sessions and matches (e.g., mileage or pass completion rate of a player) form more and more a crucial base for those tactical decisions. Most of the analyses are video-based, resulting in high operating expenses. Additionally, a highly specialized system with a huge amount of system resources like processors and memory is needed. Typically, analysts present the results of the video data in time-outs (e.g., in the half-time break of a soccer match). Therefore, coaches are not able to view statistics during the match. In this paper we propose the concepts and current state of Herakles, a system for live sport analysis which uses streaming sensor data and a Peer-to-Peer network of conventional and low-cost private machines. Since sensor data is typically of high volume and velocity, we use a distributed data stream management system (DSMS). The results of the data stream processing are intended for coaches. Therefore, the front-end of Herakles is an application for mobile devices (like smartphones or tablets). Each device is connected with the distributed DSMS, retrieves updates of the results and presents them in real-time. Therefore, Herakles enables the coach to analyze his team during the match and to react immediately with tactical decisions.