This paper considers the exploitation of unmanned aerial vehicles (UAVs) inwireless networking, with which communication-enabled robots operate as flyingwireless relays to provide connectivity or a capacity boost to a ground user.We focus on the particular problem of (automatic) UAV positioning, whichgreatly affects the end-to-end throughput performance. While existing methodsrely on propagation distance minimiza- tion and statistical models for thepresence or absence of a line-of-sight (LOS), we propose an approach capable ofleveraging local topological information so as to offer better performanceguarantees. The proposed method allows to strike a trade-off between minimizingdistance path loss and discovering (near) LOS opportunities at locations awayfrom the base station (BS)-user axis. Furthermore, the algorithm is shown tofind the global optimal UAV position, although it only requires a localexploration of a signal strength map and the length of search trajectory isonly linear to the geographical scale. Hence, it lends itself to onlineimplementation. Significant throughput gains are found when compared to otherpositioning approaches based on LOS statistical models.
展开▼