Compressive Data Aggregation on Mobile Wireless Sensor Networks for Sensing in Bike Races

TL;DR

This paper introduces a compressive sensing-based data aggregation method for mobile wireless sensor networks in bike races, leveraging spatial priors and belief propagation to efficiently track bike positions.

Contribution

It proposes a novel data aggregation approach combining compressive sensing with spatial priors modeled by a graphical model for bike race monitoring.

Findings

Effective data compression in bike race simulations

Utilizes spatial priors for improved accuracy

Validated on real bike race trajectories

Abstract

-This paper presents an efficient approach to data collection in mobile wireless sensor networks, with the specific application of sensing in bike races. Recent sensor technology permits to track GPS position of each bike. Because of the inherent correlation between bike positions in a bike race, a simple GPS log is inefficient. The idea presented in this work is to aggregate GPS data at sensors using compressive sensing techniques. We enforce, in addition to signal sparsity, a spatial prior on biker motion because of the group behaviour (peloton) in bike races. The spatial prior is modeled by a graphical model and the data aggregation problem is solved, with both the sparsity and the spatial prior, by belief propagation. We validate our approach on a bike race simulator using trajectories of motorbikes in a real bike race, the "crit{\'e}rium du dauphin{\'e} 20XX" .