Twitter in disaster mode: smart probing for opportunistic peers

Abstract

Recent natural disasters (earthquakes, floods, etc.) have shown that people heavily rely on platforms like Twitter to communicate and organize in emergencies. To mitigate communication outages due to broken infrastructure during such events, it was proposed to make mobile Apps "disaster ready": In normal operation mode, they use cellular infrastructure to communicate, whereas in disaster mode, they rely on opportunistic communication and disseminate messages in a peer-to-peer manner, using Bluetooth, WiFi ad hoc or WiFi Direct. Such hybrid applications can use the infrastructure to prepare for a network outage. In this poster, we present preliminary work towards such a hybrid solution to solve the ad hoc connectivity problem. For security reasons, discoverability of devices (e.g., in Bluetooth) is often restricted to short time periods - a limitation which prevents us from building usable opportunistic networks for disaster relief. However, if we are able to predict which peers are within transmission range, we can circumvent this limitation: Instead of scanning for discoverable devices, we can probe for potential peers and simply connect if we find them in range. In our approach, the disaster ready App obtains a list of predicted potential peers from a central server during normal operation mode, which is then used for probing in disaster mode. We discuss the challenges and limitations of such an approach and sketch how to predict contacts based on location and/or social information.