IMDEA Networks Institute Publications Repository

Coverage Optimization with a Dynamic Network of Drone Relays

Arribas, Edgar and Mancuso, Vincenzo and Cholvi, Vicent (2019) Coverage Optimization with a Dynamic Network of Drone Relays. [Journal Articles] (Accepted for publication)

[img] PDF (Coverage Optimization with a Dynamic Network of Drone Relays) - Accepted Version
Download (21Mb)

Abstract

The integration of aerial base stations carried by drones in cellular networks offers promising opportunities to enhance the connectivity enjoyed by ground users. In this paper, we propose an optimization framework for the 3–D placement and repositioning of a fleet of drones with a realistic inter-drone interference model and drone connectivity constraints. We show how to maximize network coverage by means of an extremal-optimization algorithm. The design of our algorithm is based on a mixed-integer non-convex program formulation for a coverage problem that is NP-Complete, as we prove in the paper. We not only optimize drone positions in a 3–D space in polynomial time, but also assign flight routes solving an assignment problem and using a strong geometrical tool, namely Bézier curves, which are extremely useful for non-uniform and realistic topologies. Specifically, we propose to fly drones following Bézier curves to seek the chance of approaching to clusters of ground users. This enhances coverage over time while users and drones move. We assess the performance of our proposal for synthetic scenarios as well as realistic maps extracted from the topology of a capital city. We demonstrate that our framework is near-optimal and using Bézier curves increases coverage up to 47% while drones move.

Item Type: Journal Articles
Uncontrolled Keywords: Aerial networks, Relay, UAV, Mobile networks, Optimization, Bézier curves.
Subjects: UNSPECIFIED
Divisions: UNSPECIFIED
Depositing User: Edgar Arribas
Date Deposited: 15 Jul 2019 09:58
Last Modified: 15 Jul 2019 09:58
URI: http://eprints.networks.imdea.org/id/eprint/2027

Actions (login required)

View Item View Item