Optimized Service Provisioning for Next-Generation Applications with Satellite Edge Computing

Satellite Edge Computing (SEC) extends the Mobile Edge Computing (MEC) paradigm by deploying cloud computing resource pools on in-orbit satellite nodes within a constellation. By enabling computational offloading to space, SEC empowers the on-board execution and seamless completion of service. At the same time, its in-orbit storage capacity facilitates proactive content caching and efficient content delivery to end users. The distributed and localized processing and storage capabilities of satellites in the SEC paradigm, unlike the traditional bent-pipe satellite architecture, ensure high service availability and coverage, particularly for users in remote, underserved, and disaster-stricken areas where terrestrial networks are unavailable. The ubiquitous service coverage, low latency, and high bandwidth enhance scalability and meet the stringent requirements of next-generation applications such as mission-critical services, real-time analytics, and high-quality multimedia delivery, making the SEC paradigm a promising solution for service provisioning in future networks. In this context, service provisioning refers to allocating and managing resources to deliver services to end users based on application-specific requirements.

Although SEC-enabled platforms provide tremendous advantages, service provisioning in these networks faces several significant challenges. Firstly, on-board satellite resources are limited, so they cannot process or store all the tasks and data that users request. This necessitates advanced optimization algorithms to manage and allocate these scarce resources efficiently. Furthermore, the requirements of emerging next-generation applications and services are stringent (\textit{e.g.}, low latency, high reliability, and high bandwidth) and often complex to achieve, particularly in the satellite environment where resource constraints and long propagation delays are prevalent. The dynamic nature of service requests, with their heterogeneous and sometimes intricate requirements, adds another layer of difficulty in meeting these demands. Additionally, the mobility of satellites results in dynamic coverage areas, time-varying topologies, and fluctuations in link characteristics. These factors hinder seamless service provisioning and complicate the design of effective service provisioning strategies. An advanced service provisioning scheme is required to alleviate these challenges and fully leverage the benefits of the SEC. This scheme must comprehensively accommodate heterogeneous requirements, dynamic traffic requests, and evolving topologies while meeting stringent service requirements. Such a scheme is crucial in next-generation networks like 5G/6G, where Non-Terrestrial Networks (NTN) are considered a key enabling technology. NTNs improve network resilience by integrating terrestrial and satellite networks, providing ubiquitous connectivity, and improving overall network performance.

Motivated by these challenges, in this thesis, we explore service provisioning schemes for next-generation applications by leveraging the in-orbit computational and storage power of SEC. In the first contribution, we proposed a Virtual Network Function (VNF) mapping and scheduling scheme for mission-critical applications, which enables Network Functions (NFs) execution on-board in the constellation to maximize fairness in terms of End-to-End (E2E) service delay margin and reduce the E2E service delay among competing services. In the second contribution, we propose an efficient storage resource utilization method for on-board content caching by designing a novel satellite proximity-based content popularity scheme. Furthermore, we proposed an optimization framework for cache reconfiguration overhead-aware on-board content caching. We also study on-board cache-to-cache updates to reduce reconfiguration overhead, followed by an Age of Information (AoI)-aware content caching scheme to ensure cached content freshness. Additionally, in the third contribution, we focus on service provider revenue generation through ad monetization while enhancing efficient content distribution schemes. We study the impact of excessive ad insertion, which leads to user disengagement and indirectly reduces the content provider's revenue. We model and propose a joint optimization algorithm that balances revenue and end-user delivery delay while considering the constraints. We also propose seamless content delivery by allowing distributed Server-Side Ad Insertion (SSAI) and service continuity-aware content distribution strategies to mitigate user disengagement and network fluctuations.