Conventional communication simulation models, such as iterative-based channel estimation algorithms, struggle to meet the complex and dynamic requirements of Low-Altitude Economy Networking, where real-time responsiveness, adaptability, and efficiency are critical. These limitations become particularly pronounced in scenarios involving unmanned aerial vehicles (UAVs), urban air mobility (UAM), and distributed sensor infrastructures, which demand agile and intelligent networking solutions. As the Low-Altitude Economy Networking evolves toward high-density, low-latency, and mobility-driven applications, existing Transformer-based communication models face barriers to deployment due to their excessive computational load and parameter volume—especially on resource-constrained edge devices. To address the aforementioned challenges, this paper proposes a Lightweight Transformer-based communication simulation model(AutoLTS), featuring a domain-specific multi-layer large language model (LLM) architecture and a Transformer-driven lightweight semantic analysis module. By dynamically leveraging advanced techniques in deep representation learning, attention mechanisms, and adaptive compression, the proposed model effectively captures the characteristics of time-varying wireless channels, simulates complex air-to-ground propagation environments, and infers latent variables for blind channel equalization. This approach significantly reduces computational overhead while maintaining critical performance metrics, thereby offering a scalable and efficient solution for intelligent communication system design in resource-constrained and dynamic networking scenarios. This method tailored for Low-Altitude Economy Networking scenarios, which aims to significantly reduce computational overhead while ensuring performance, thereby providing support for the design and verification of Low-Altitude Economy Networking.