Lopofly: técnica de otimização de localização e posicionamento para redes aéreas

Abstract

In areas that require short-term and sporadic connectivity, such as events and mobile offices, it is impossible to maintain a permanent network infrastructure to provide broadband Internet access to temporary customers. The use of flying nodes to build flyingnetworks has aroused great interest from both the scientific community and the industry in general. However, these networks need to be carefully managed, mainly due to the limited power capacity of the flying nodes. Despite the different existing solutions, until the present moment, we found no research that proposes models that jointly consider all the restrictive aspects of using flying nodes: communication, mobility, energy restriction, and turnover. Considering these problems, we present a new technique Location and Positioning Optimization Technique for Flying Networks (LoPoFly). It consists of two modules: (i) location and (ii) positioning. The first (location) aims to find a place where a flying node is needed, based on the clients’ distribution. The second (positioning) is responsible for managing the relocation and exchange of flying nodes, considering energy consumption. These modules use the Deterministic Annealing meta-heuristic (DA). DA emulates a physical process (annealing), in which a solid is heated up to its melting point and cooled to reach its minimum energy configuration. It is used for clustering, compression, and classification problems. To the best of our knowledge, this is the first approach to manage flying networks covering constraints related to energy, replacement, communication, and mobility. Through simulations, we analyzed the performance of LoPoFly in two scenarios. The ability of the location module to identify new locations allowed an increase of more than 214 % in the number of connected customers, in both scenarios, compared to a random solution. The results also show that in both scenarios, LoPoFly reduces the number of nodes needed to supply the event, reducing 40 % for the first scenario and 60 % for the second.