Compressed Sensing based Protocol for Interfering Data Recovery in Wireless Multi-Hop Sensor Networks
Megumi Kaneko
20 March 2014, 14h30 Salle/Bat : 2013/DIG-Moulon
Contact :
Activités de recherche : Optimisation dans les réseaux
Résumé :
We consider a multi-hop wireless sensor network that measures sparse events and propose a novel protocol based on Compressed Sensing (CS) as an alternative to traditional Media Access Control (MAC) scheduling and routing protocol. Instead of avoiding collisions, our protocol based on CS and flooding exploits interferences by superimposing the data measurements "over-the-air", simultaneously received at any node. Thanks to our protocol design, each node is able to recover and forward only new data towards the sink. Numerical results show that our protocol achieves near zero reconstruction errors at the sink, while greatly reducing overhead and delays compared to conventional protocols. These results reveal a new and promising approach to protocol design through CS-based optimization.
