Experimental study of mobility in the soccer field with application to real-time athlete monitoring

Abstract

Abstract-Live monitoring of athletes during sporting events can help maximise performance while preventing injury, and enable new applications such as referee-assist and enhanced television broadcast services. A major challenge is the extraction of athlete physiological data in real-time, since the radio range of body-worn sensor devices is limited, necessitating multi-hop routing mechanisms. However, little is known about the highly dynamic operating conditions on a soccer field under which communication protocols need to operate.In this work we conduct field experiments in which we outfit first-division soccer players with sensor devices and record their inter-connectivity during a real game. Our first contribution profiles the key properties of the dynamic wireless topologies arising in the soccer field, and highlights the consequences for routing mechanisms. We show that the topology is in general sparse, with short encounters and power-law distributed interencounters. Importantly, the co-ordinated movement of players in the field gives rise to significant correlations amongst links, an aspect that can potentially be exploited by routing. Our second contribution develops a model for generating synthetic topologies that mirror connectivity in a real soccer game, and can be used for simulation studies of routing mechanisms. Its novelty lies in explicitly modelling the underlying auto-correlation and cross-correlation properties of the links, from which derived measures such as inter-encounter times and neighbourhood distributions follow. Our study is an important first step towards understanding and modelling dynamic topologies associated with sports monitoring, and paves the way for the design of real-time routing algorithms for such environments.

Experiments with wireless sensor networks for real-time athlete monitoring

“…The long inter-meeting times between players are a fundamental limitation that routing protocols must contend with. Our experiments, profiling and modelling of the connectivity are described in detail in [4]. Further, as we show in [5], the attenuation caused by the human body results in significant asymmetry in wireless range on the two sides of the body (with the Mote mounted on the right arm we see a range of up to 16m on the right, as opposed to only 3m on the left).…”

“…Other protocols such as Seek and Focus protocol [6], PROPHET [11] and MAXPROP [12] assume that the interaction between nodes is sufficiently long to allow the exchange of utility metrics which determine forwarding decisions. However, our analysis of the interconnectivity between soccer players in [4] shows connectivity periods to be too short to allow exchange of such metrics. The only routing schemes proposed for highly dynamic networks, as arise in the soccer monitoring application, include Direct Transmission [13], Randomized Routing [14] and Spray and Wait [15].…”

“…In [6] the authors show that, for mobility models where the expected meeting time is a concave function of distance, randomized forwarding results in a reduction of the expected delivery time of the message. The assumptions for this result call for every node to move independently according to a mobility model which has this property; this does not happen in a soccer game, which exhibits correlation between player movements [4]. However, for players who hold more central positions in the field, random forwarding results in the sample reaching more peripherally located players, who have smaller expected times to hit the base.…”

Modeling signal strength of body-worn devices

“…As we shall show, when the object in question is a human being, both distance and orientation are important. From our experiments with first division soccer players, we were able to show that link correlations exist between soccer players due to their co-ordinated motion [6]. We have also created a multi-hop protocol which balances resource consumption and delay to deliver data in an optimal manner [7].…”

Application specific study, analysis and classification of body area wireless sensor network applications

“…Due to the limited wireless range of worn sensors (BASN), it is difficult to take the physiological data of athlete in real time and it needs multi-hop routing mechanism to implement such type of system. In [57] the authors proposed a model that can produce synthetic dynamic topologies using stochastic attributes. The model is useful in simulating the performance of different routing strategies for monitoring soccer player's movements during a game.…”