Abstract :
[en] The rapid growth of mobile traffic and emerging service requirements have accelerated the evolution of wireless networks, particularly through the integration of satellite communication (SatCom) systems and terrestrial networks (TNs). This convergence has led to Integrated Satellite-Terrestrial Networks (ISTNs), envisioned as a promising enabler of seamless, ubiquitous, and high-capacity connectivity in future 6G networks. By leveraging the strengths of both TNs and satellite networks (SatNets), ISTNs are able to enhance coverage, support direct-to-device (D2D) services, provide satellite-based backhaul, and improve spectrum utilization. However, several key challenges arise due to the dynamic behavior of non-geostationary satellites-especially low Earth orbit (LEO) satellites (LSats), heterogeneous terrestrial environments, uneven traffic demands, spectrum coexistence, and traditional fixed spectrum regulation.
Addressing these challenges necessitates a careful study of various technical issues. To efficiently deliver traffic demands under system constraints, critical technique issues must be addressed, including: (i) bandwidth (BW) allocation across services, systems, and satellite-based backhaul links, (ii) link association among users (UEs), base stations (BSs), LSats, (iii) ensuring seamless connectivity, (iv) comprehensive resource management, encompassing resource block (RB) assignment and power control; and (v) effective traffic management. These issues introduce various problems which must be addressed.
Driven by ISTN scenarios identified by standardization bodies and the aforementioned technique issues, within the scope of this thesis, three critical ISTN problems are considered:
(1) backhaul-link BW allocation and two-tier user association,
(2) throughput enhancement and seamless handover in C-band 5G-ISTNs, and
(3) resource management in dynamic spectrum sharing ISTNs.
The first problem investigates the ISTN scenarios wherein a LEO constellation provides backhaul links for TNs. To efficiently deliver data from UEs to BSs and from BSs to LSats, BW allocation backhaul links, a time-window-based optimization framework is proposed to minimize transmission time through jointly managing two-tier UE-BS-LSat association, backhaul BW allocation, subchannel assignment, and power control. To efficiently solve this challenging non-convex mixed-integer non-linear programming (MINLP) problem, a centralized algorithm is proposed by leveraging successive-convex-approximation (SCA) and compressed-sensing techniques. Furthermore, a decentralized algorithm is developed to support scalable and efficient deployment as well as to offload computations from central node.
The second problem explores ISTNs supporting the co-primary TN-SatNet spectrum use in urban environments, specifically for automotive UEs with both TN and SatNet direct-to-vehicle access. Accordingly, a joint UE association (UA) and power control mechanism is developed to maximize system throughput and minimize connection state changes. To deal with this problem efficiently, two algorithms are developed: an SCA-based algorithm addresses the full time-window problem while a prediction-based one enables the optimization in sequential sub-time-windows.
The third problem introduces a digital-twin (DT)-aided dynamic spectrum sharing (DSS) framework for D2D-enabled ISTNs, where the TN and SatNet share the same 5G-NR band.
The goal is to minimize system congestion via joint optimization of BW slicing, traffic steering, and radio resource management (RRM). This framework comprises two optimization problems: (i) a joint-RA problem based on DT-predicted data for network and resource decisions, and (ii) a refinement problem that updates decisions using real-time feedback. Corresponding algorithms are designed to align with the proposed system architecture.
Extensive simulations are conducted under practical settings, especially ray-tracing (RayT)-based evaluations using realistic 3D maps of London for the second and third problems. Numerical results demonstrate the superior performance of the proposed algorithms for each of the three problems compared to conventional benchmarks in terms of transmission time reduction, throughput improvement, minimized connection transition events, and congestion alleviation, respectively. Overall, this thesis contributes efficient RRM designs for ISTNs across diverse scenarios and highlights promising directions for future ISTN research.
