Ricerc@Sapienza

This project is highly innovative with respect to the current state of the art, and the results are expected to bring advancement both from a theoretical and practical point of view. To the best of our knowledge, there are no research projects targeting the same objectives of Sky-Edge up to date, see Table 1. To be more specific, resource allocation for MEC is a pivotal topic in several research communities [9], [12]-[27]. Resource allocation strategies for MEC are either static or dynamic. The static formulation deals with short time applications, in which users request for a computation with a well-defined computational constraint [9], [12]-[20]. On the other hand, in dynamic scenarios, the application continuously generates data to be processed, possibly with an unknown process [21]-[27]. Providing on-demand radio coverage exploiting base stations mounted on aerial platforms is an idea that has received considerable attention since a few years, see, e.g., [28]. However, all the aforementioned strategies did not consider mobile edge computing with UAVs. The distinctive feature of Sky-Edge is to exploit aerial and terrestrial platforms not only to improve radio access capability, but to move the edge cloud to the sky by incorporating MEC functionalities on aerial platforms. This idea was originally suggested in [29], and later investigated in a few recent works [30], [31]. In [30], the authors considered a single UAV with onboard a radio access point and a mobile edge host (MEH), serving a set of mobile users on ground, and proposed an algorithm to jointly optimize: the data rate in the uplink and downlink channels, using either orthogonal or non-orthogonal multiple access (NOMA), the computation rate on the UAV, and the flight trajectory. The formulation proposed in [30] was static and all variables were considered deterministic, i.e., perfectly known. Very recently, the joint optimization of resource allocation and UAVs trajectory was reformulated in a dynamic framework, incorporating channel random models [31]. Also, in [31], a single UAV was considered, and its altitude was taken fixed and given. In Sky-Edge we plan to go well beyond these preliminary works and provide advancements in the following directions. First of all, our main goal is to provide seamless service continuity, i.e., C3 services on demand anytime and everywhere in 3D space. Provisioning of C3 services in 3D is enabled via a combination of conventional ground (terrestrial) base stations and flying (non-terrestrial) nodes including UAVs, HAPSs, or satellites, that are dynamically organized and their C3 resources are jointly orchestrated via dynamic optimization and learning algorithms. For this reason, we plan to investigate a scenario composed of a fleet of heterogeneous UAVs, each having its own specific features in terms of payload, flight autonomy, typical altitude, revisit time, computation, storage, connectivity, etc. To ensure service continuity we will derive new mobility management algorithms incorporating proactive allocation of resources and dynamic handover between radio access points and MEHs. Differently from [30] and [31], where the UAV altitude was considered fixed and the flight time was also given a priori, we will optimize the selection of the flight altitude, taking into account coverage needs. Furthermore, rather than assuming flight duration as a priori fixed, we will maximize the flight duration under constraints on end-to-end delay.
Enlarging our perspective, the project might provide a substantial evolution of the 5G ecosystem, thus opening several opportunities for developing new cloud services to be deployed in scarcely covered areas. From the academic point of view, this will provide unique opportunities to develop new applications, enrich students' curriculum and encourage collaboration with industrial companies. At the societal level, UAV-empowered mobile edge computing will help to extend 5G services in remote/rural areas and hot spots, thus enabling (non-exhaustive list):
1) Efficient handling of emergency situations (e.g., natural disasters or terroristic threats), enabling fast responses especially in areas with high risks and for services requiring a fast reaction time;
2) Robustifying current wireless cellular networks and sensor networks in case of node failures;
3) Enabling smart agriculture, by providing on demand analysis of soil, drainage, crop health assessment, etc.;
4) Improving entertainment, by providing contents on demands or running context-aware application running on data sent by mobile users and incorporating images from high altitude for augmented reality applications;
5) Improving the environment exploration capabilities, e.g., by devising distributed online mining mechanisms.
In conclusion, Sky-Edge might pave the way to several developments that may have a deep impact at the societal level.