Joint Service Caching and Task Offloading for Multi-UAV-Assisted Offshore Edge Computing Networks

Marine Internet of Things devices (MIoTDs) are inherently constrained by limited computation resources, while conventional mobile edge computing implementation faces significant challenges in offshore environments due to unstable shore-based communications and widespread device distribution. To this end, this paper proposes an integrated offshore-aerial edge computing and caching architecture where multiple unmanned aerial vehicles (UAVs), capable of dynamically adjusting their three-dimensional positions for optimal service coverage, serve as edge servers collaborating with a shore-based base station to provide computing and caching services for MIoTDs. The objective is to minimize the total system latency by jointly optimizing task offloading ratios, user association decisions, and service caching decisions, subject to MIoTD energy, UAV energy, and cache capacity constraints. To solve the resulting mixed-integer nonlinear programming problem, we develop an efficient iterative algorithm that decomposes it into tractable subproblems, strategically employing coalition game, linear relaxation, and Lagrangian dual methods to address their distinct mathematical structures. Extensive numerical results demonstrate that the proposed scheme reduces the total latency by 15.18%, 8.90%, and 8.38% compared to local-only computing, random caching, and fixed offloading baselines, respectively, while ensuring fair service provision across MIoTDs.