Section: New Results

Self-Organization

Participants : Nathalie Mitton, Valeria Loscri, Farouk Mezghani, Simon Duquennoy, Anjalalaina Jean Cristanel Razafimandimby.

Bayesian communications

In the last few years, Internet has become a very important vector of information sharing. Beyond the interconnection of computers and devices, there is still an important expansion capability, thanks to the capacity to interconnect heterogeneous devices. This extension of Internet known as Internet of Things (IoT) leads to (inter)connection of billions of objects. Nevertheless, IoT paradigm raises many challenges, such as the need to manage a massive amount of data generated by sensing devices. It was observed that, with the increase of sensors density, the redundancy of data increases. Thus, uploading raw data to the cloud can become extremely inefficient.

In order to address this issue, we proposed a Bayesian Inference Approach (BIA), able to remove a great amount of spatio-temporal correlated data [46], [10], [35].

In order to validate these approaches it was considered that experiments in real-world scenarios were needed. More specifically, we considered indoor tests in [46] and agricultural/outdoor experiments in [47]

Alert diffusion

Opportunistic communications present a promising solution as a disaster network recovery in emergency situations such as hurricanes, earthquakes and floods where infrastructure might be damaged. Recent works have proposed opportunistic-based alert diffusion approaches useful for trapped survivors. However, two main features were left behind. On the one hand, these works do not consider the assortment of networks integrated in mobile devices (e.g. WiFi-Direct, WiFi ad-hoc, cellular, bluetooth) and the choice of the network interface is left to the user who has no idea what is best or might be in a physical or psychological distress that impede the efficient selection. On the other hand, most of these works are based on selfish diffusion which might drain quickly the battery power. Moreover, they do not consider various energy level, which obviously influences the alert diffusion scheme. [17], [27], [28], [44] propose COPE and its demo, a cooperative opportunistic alert diffusion approach for disaster scenario that considers mobile devices that come with multiple network interfaces and with various battery power level. In order to maintain mobile devices alive longer, survivors form cliques and zones in which they diffuse alternately and periodically alert messages until reaching a potential rescuers team. Simulation results show that COPE largely outperforms the selfish diffusion scheme in terms of energy consumption while guaranteeing an important alert delivery success.

Consensus-based Leader election

In low-power wireless networking, new applications such as cooperative robots or industrial closed-loop control demand for network-wide consensus at low-latency and high reliability. Distributed consensus protocols is a mature eld of research in a wired context, but has received little attention in low-power wireless settings. In [21], [55], we present $A^2$: Agreement in the Air, a system that brings distributed consensus to low-power multi-hop networks. $A^2$ introduces Synchrotron, a synchronous transmissions kernel that builds a robust mesh by exploiting the capture effect, frequency hopping with parallel channels, and link-layer security. $A^2$ builds on top of this reliable base layer and enables the two-and three-phase commit protocols, as well as network services such as group membership, hopping sequence distribution and re-keying. We evaluate $A^2$ on four public testbeds with different deployment densities and sizes. $A^2$ requires only 475 ms to complete a two-phase commit over 180 nodes. The resulting duty cycle is 0.5% for 1-minute intervals. We show that $A^2$ achieves zero losses end-to-end over long experiments, representing millions of data points. When adding controlled failures, we show that two-phase commit ensures transaction consistency in $A^2$ while three-phase commit provides liveness at the expense of inconsistency under specific failure scenarios.