D3QN-Based TT&C Mission Planning Method for Mega-Constellations under Sparse TT&C Resources

Telemetry, Tracking, and Control (TT&C) refer to the process by which ground TT&C stations track and measure satellites while issuing control commands to monitor satellite status, adjust orbital and attitude parameters, and transmit data. To date, Chinese TT&C network has evolved into a comprehensive aerospace TT&C system characterized by high precision, high reliability, and a high degree of automation. This network is structured around a central TT&C hub that connects tracking stations and relay satellites. TT&C mission planning is a key technology to ensure the stable in-orbit operation of satellites and maximize their application effectiveness. However, in the context of large-scale low and medium Earth orbit constellations, the TT&C coverage is sparse, TT&C arcs are short, and the large number of concurrent tasks leads to frequent conflicts, making the coordination of limited ground TT&C resources challenging and resulting in low mission completion rates. To address these challenges, this paper proposes a Fusion Channel Attention-D3QN-Based TT&C Mission Planning Method for mega-constellations under sparse TT&C resources. First, we systematically analyse the TT&C mission planning problem under sparse resources and construct an integer programming model with hybrid optimization objectives, including TT&C arc length and task priority. A state-action space is then designed for mega-constellations operating under sparse TT&C resources. To enhance the planning performance, we introduce a Dueling Network structure to refine the definition of TT&C resources and available time slots, thereby decoupling state values from action advantages and reducing the complexity of high-dimensional TT&C mission planning. Furthermore, to address the limitation of the Dueling Double Deep Q-Network (D3QN) in distinguishing the importance of different feature information, we integrate an attention mechanism into the D3QN framework. Specifically, a local attention mechanism is introduced to amplify the available value of sparse TT&C resources, promoting balanced resource utilization. Finally, simulation results demonstrate that the Channel AttentionD3QN TT&C mission planning network achieves increasing training rewards and gradually converges. Independent test set evaluations confirm the ability to rapidly generate effective planning solutions, validating its effectiveness. This research provides theoretical and technical support for the operation of Chinese TT&C system.