Toward in-flight Wi-Fi: a neuro-fuzzy based routing approach for Civil Aeronautical Ad hoc Network

In-Flight Wi-Fi connectivity (IFC) paves way for supporting the Internet of Things over the clouds by connecting things inside a moving aircraft to the ground. Aeronautical Ad hoc Networks (AANET) is a new breed of Mobile Ad hoc networks envisioned to experience IFC over remote or oceanic regions. Due to the challenging characteristics of AANET such as limited bandwidth, intermittent connectivity, dynamic topologies, the greater velocity of aircraft, and variable geographical network sizes, it is very hard to design a realistic mobility model and network routing with remarkable Quality of Service (QoS) and requires immediate research solution. As the complexity of provisioning QoS network services significantly relies on the routing layer, this work aims at providing intelligent, reliable, and efficient data delivery with ensured QoS. In the aim of attaining a solution for QoS routing, this paper presents twofold design strategies. Firstly, the mobility of moving aircraft is modeled with International Civil Aviation Organization separations standards to avoid collisions among moving aircrafts, second is, a novel hybrid approach combining deep learning and fuzzy logic is proposed to deal with highly dynamic nature and growing air traffic. The neighbor discovery phase of existing routing protocols incurs more packet overhead and delay because of the traditional beaconing method, which is overcome in this work by using Automatic Dependent Surveillance-Broadcast as it is kept on broadcasting the neighbor’s information in the cockpit display of every aircraft. To avoid the overwhelming of nodes, this work uses queuing delay of the neighbors to identify the node’s ability for packet transmission fairly distributes the load among aircrafts. The simulation results show that the proposed work provides notable improvements in packet delivery ratio, end-to-end delay, and traffic overhead compared to the existing routing protocols in sparse and dense network scenarios.