Jian Wu

The Earth Observation Satellite Task Scheduling Problem (EOSTSP) is an NP-Hard challenge critical for space-based information acquisition. Conventional optimization methods are hampered by high development costs, limited adaptability to dynamic scenarios, and poor interpretability. We propose a training-free solution that leverages the reasoning capabilities of Large Language Models (LLMs) through a multi-agent framework. Our approach introduces an order-driven scheduling strategy that fundamentally reconceptualizes the EOSTSP as a real-time order fulfillment process, drawing inspiration from dynamic resource allocation paradigms in complex service systems. This original methodology transforms the traditionally monolithic scheduling problem into a structured sequential decision flow, where each observation task is processed as an independent order through specialized decision phases. Implemented via the Tri- Stage LLM-Coordinated (TSLC) framework, this novel approach enables autonomous scheduling through coordinated multi-agent reasoning while maintaining full operational feasibility via external validation. Empirical results demonstrate performance improvements of 4.58% and 13.38% over state-of-the-art algorithms in routine and emergency scenarios respectively, while providing transparent interpretability through its structured decision-making process, effectively overcoming the black-box limitations of traditional optimization approaches.

The agile earth observation satellite scheduling problem with time-dependent transition constraints is a complex combinatorial optimization problem that has been proven NP-hard. To address this challenge, a Deep Q-network guided genetic algorithm with a local intensification strategy (DQN-LIGA) is proposed. First, we adopt a hybrid initialization method that combines greedy and random strategies to generate a high-quality and diverse initial population. Next, multiple crossover–mutation operator pairs are designed. A Deep Q-network is then employed as a policy model to adaptively select operator combinations based on individual and evolutionary states. Moreover, a simulated physical oscillation strategy (SPOS) is proposed to periodically perturb elite individuals, which is designed to refine solutions and strengthen local intensification throughout the evolutionary process. Finally, extensive experiments are conducted on 20 benchmark instances under both China and World scenarios. The results demonstrate that DQN-LIGA consistently achieves higher observation profits compared to state-of-the-art algorithms, and the integration of SPOS further improves solution quality, especially under complex scheduling conditions.

This paper focuses on the research of anti-jamming issues for Low Earth Orbit (LEO) satellite constellations. Initially, the anti-jamming problem is modeled as a Local Interaction Markov Game (LIMG) and proven to be an Exact Potential Game (EPG), with at least one pure strategy Nash Equilibrium (NE) existing. Secondly, based on the theoretical analysis and the ”offline training and online execution” architecture, a Distributed Multi-Agent Deep Reinforcement Learning-based anti-jamming (DMDRLA) scheme is proposed, and its convergence and asymptotic optimality are theoretically analyzed. Finally, simulations validate that the proposed DM-DRLA scheme can effectively balance the training costs and performance optimization of the anti-jamming model, making it suitable for anti-jamming issues in resource-constrained LEO satellite networks.

A mega hybrid constellation comprising low Earth orbit (LEO) and geostationary orbit (GEO) satellites represents a prevalent architecture for future space-based networks. However, the emergence of mega constellations has exacerbated the shortage of spectrum resources. To address this issue, this paper investigates a method for LEO satellites within such constellations to expand their available spectrum by sharing the downlink spectrum of GEO satellites. Firstly, to avoid interference with GEO satellites and optimize the beam coverage for LEO user (LU), a coalition formation game model for LU based on cooperation criteria is constructed, and the existence of a stable coalition structure is proven. Secondly, to determine this stable coalition structure, a coalition formation game algorithm based on the best response (BR) is proposed, and its convergence is theoretically validated. Additionally, to more efficiently determine the beam radius and center covering the LU in the coalition, an improved algorithm for solving the outer circle of LU in the coalition using a K-dimensional tree is presented. Simulation results demonstrate that the proposed method effectively balances convergence time and accuracy. Without affecting GEO satellite communications, LEO satellites can share the downlink spectrum of GEO satellites, thereby enhancing the utilization of spectrum resources within the hybrid constellation.