3D Determination of Message Collection and Delivery Locations for UAV-Enabled Disaster Recovery Networks

Abstract

When a large-scale disaster happens, critical infrastructure is destroyed, and many people are displaced. Unmanned aerial vehicle (UAV)-enabled disaster recovery networks can be used to support people in disaster-hit areas. However, determining UAV routes is critical to communicate with refugees. In this work, we propose a method to determine UAV locations to collect and deliver messages for UAV-enabled disaster recovery networks. Our method involves two stages: the received signal strength sensing stage and the message collection and delivery stage. We theoretically analyzed the computational complexity in terms of the number of steps and consequently evaluated the performance of our method using experimental data obtained from rural, suburban, and urban areas.

Efficient 3D-RSS map estimation method based on area classification

“…R (p) is estimated from the M RSS values with TC. As in [1], we use the total variation low rank tensor completion method [3]. It is worth mentioning that the length of Ω is reduced when more RSS maps are classified as PL-type RSS maps.…”

“…Unmanned aerial vehicles (UAVs) are promising for supporting existing terrestrial information networks since they can equipped with wireless communication tools and fly anywhere. In [1], we proposed a message collection and delivery system for UAV-enabled disaster-recovery networks based on three-dimensional maps of received signal strength (3D-RSS maps). The 3D-RSS maps are constructed from sparsely sensed RSS using tensor completion.…”

“…The 3D-RSS maps are constructed from sparsely sensed RSS using tensor completion. In [1], we found that RSS maps without any missing stationary points are essential for complete message delivery, while the sensing obviously requires the overhead of measuring many RSSs prior to message delivery. Note that this paper does not target adaptive line-of-sight (LOS) and non-line-ofsight (NLOS) environments selection.…”

“…In this paper, we propose an efficient new 3D-RSS map-estimation method by extending the existing method developed in [1]. Our proposed new method classifies 3D-RSS maps into path loss (PL)-type and tensor completion (TC)type RSS maps, according to the number of high buildings around the ground stations.…”