19 June 2025 in IEEE Internet of Things Journal
Multiaccess edge computing (MEC) technology has been widely considered as a paradigm for offloading computation-intensive and latency-sensitive tasks from ground devices (GDs). However, the deployment of ground edge servers to provide ubiquitous computing services for GDs may incur extremely high costs, primarily due to the uneven distribution of devices and the complexity of geographical environments. As a promising solution, deploying edge servers on low-Earth orbit (LEO) satellites can offer superior coverage and reduced latency for processing remote terrestrial computational tasks. Nevertheless, considering the limitations in computing resources and energy supply on LEO satellites, it is crucial to enhance task processing efficiency while taking energy consumption into account. In this article, we consider a Multisatellite cooperative edge computing network (MSCECN) where time-sensitive tasks can be offloaded to an access satellites (AS) and multiple edge satellites (ESs) through intersatellite links (ISLs). We formulate a delay-minimization offloading problem by jointly optimizing satellite selection, computing resource allocation, task scheduling, and transmission power control. To solve this problem effectively, we propose a novel joint iterative algorithm called multisatellite cooperative resource allocation (MSCRA), which integrates the relaxation method, the Lagrangian multiplier framework, and the CVX toolbox. Simulation results demonstrate that the proposed optimization scheme achieves a 35.5% reduction in total delay compared to benchmark algorithms and effectively minimizes offloading latency across various resource constraints.
18 March 2026 in IEEE Transactions on Vehicular Technology
The evolution of 6 G and Internet of Things (IoT) technology is driving the proliferation of computation-intensive and latency-sensitive applications, which impose increasingly stringent requirements on IoT Devices (IDs) for powerful and timely data processing capabilities. Space-Air-Ground Integrated Network (SAGINs) have emerged as a promising candidate for providing ubiquitous computing services for processing various sensing data generated by IDs, particularly in remote areas. However, supporting these applications in SAGINs still faces two key challenges: IDs cannot afford intensive data transmission and computation due to the limited battery and computing capability, and the Ultra-Reliable Low-Latency Communication (URLLC) constraints are difficult to satisfy due to the network dynamics. To tackle these challenges, we propose an Intelligent Reflecting Surface (IRS) backscatter-enabled SAGIN architecture, where battery-efficient, radio-frequency-chain-free uplink transmissions from IDs to Unmanned Aerial Vehicles (UAVs) are enabled, significantly reducing the communication energy consumption of IDs. We aim to minimize the system energy consumption for balancing low-latency requirements against energy-saving demands under URLLC constraints. Considering that resource allocation decisions and URLLC constraints have different time scales, the Lyapunov optimization is employed to decompose the formulated problem into tractable subproblems, which are solved by integrating Deep Reinforcement Learning (DRL) with fractional programming, semi-definite programming, and the Lagrange method. Simulation results demonstrate that the proposed algorithm significantly reduces the total energy consumption by 66.67% compared to benchmark schemes, thereby enabling sustainable, high-performance computation for remote IDs.
7 November 2026 in IEEE Transactions on Mobile Computing
1 January 2026 in IEEE Communications Letters