Machine Learning-based Service Function Chain over UAVs: Resource Profiling and Framework

Abstract

Unmanned Aerial Vehicles (UAVs) are expected to align with Machine Learning (ML) to deliver smart Internet of Things (IoT) services onboard such as human detection or medicine delivery. In the context of massive IoT service deployment over UAVs, a platform such as Edge-Cloud model integrated synergistically with Network Function Virtualization (NFV) can be used for system management and control Quality of Service (QoS) via Service Function Chain (SFC) embedding strategy. However, implementing such framework should face many challenges due to limited energy and computing resources on UAVs. This issue is addressed in our article, which subsequently proposes a novel NFV-enabled Edge-Cloud architecture dedicating to implement ML-based services to UAVs. The system is implemented in our testbed that involves drone serving object detection service as a means to investigate the effects of resource, energy utilization of SFC embedding on smart service performance and drone flight duration. Results show that the proposed Edge-Cloud framework is able to deploy smart IoT services to limited resource devices such as UAVs and deliver QoS awareness via SFC embedding.