Recognizing birds in Lauzelle with DTN strategies

Details

Supervisors

Bonaventure, Olivier ; Ingenzi, Vany

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en informatique, à finalité spécialisée

Abstract

enThe project at hand aims to capture vocalization from the many bird’s species present in a wood not covered by network signal with the use of a microphone in a Raspberry Pi, and with a study on green implementations. The data should be processed by BirdNet, previously installed in the board, and be ready to be transferred to the data-mules that pass nearby with a phone application, specifically crafted to Android systems. The Raspberry, however, has a limited memory capacity, and will be forced to optimize its storage as much as possible, which contains valuable data. Since the data carriers perform a very uncertain tour, it must rely on an optimistic strategy to pass the data forward. This study uses a Rust implementation of Bundle Protocol version 7, called DTN-RS, running inside the Raspberry Pis and Android application. Once the app users arrive to the other side of the network, where the information can be uploaded to the cloud, they can alert the capturing station that this data can be freed from memory, by making the reverse path.