Slicing a FANET for heterogeneous delay-constrained applications

Abstract

Today’s networks use a “one size fits all” approach that makes them unsuitable to satisfy the performance requirements of heterogeneous applications in terms of latency, scalability, availability and reliability. To this purpose, the concept of network slices has been recently introduced to meet heterogeneous requirements of vertical applications deployed on top of common network infrastructures.The objective of this paper is to define a 6G zero-touch management framework based on a Flying Aerial Network (FANET) made by a set of UAVs, which is able to provide ground devices with network slices that are able to guarantee edge computing to remote areas according to heterogeneous requirements in terms of mean delay and jitter. An inter-slice orchestrator is introduced to split the computation power of the Computing Element of each UAV between the different slices, while an intra-slice orchestrator is in charge of managing horizontal offload among UAVs to obtain load balancing, so decreasing mean delay and delay jitter. An optimization problem based on Reinforcement Learning (RL) is defined to optimize the horizontal-offload decision process of the Intra-slice Orchestrators at run-time. An extensive simulation campaign is used to evaluate the performance of the system and to demonstrate the ability of the proposed framework to achieve energy efficiency, therefore increasing the flight autonomy of the FANET.