Radar emitter signal sorting plays a critical role in modern electronic warfare. To address the limitations of traditional sorting methods—such as low efficiency and limited accuracy in complex electromagnetic environments—as well as the challenges posed by radar-communication spectrum coexistence brought about by the development of 5G communication technologies, this paper proposes an innovative solution based on multi-task learning and multi-domain feature fusion. The proposed method first constructs a multi-task learning framework to jointly preprocess noisy radar-communication mixed signals, including signal cleansing, denoising, and time-frequency feature enhancement. Then, a feature fusion network is designed to extract multi-scale features from both the time-frequency image domain and the time-delay-Doppler domain. An improved iterative attention mechanism is employed to achieve cross-domain feature fusion, resulting in a high-dimensional feature representation with strong discriminative power. Finally, a clustering model based on DeepCluster is used to accurately sort radar emitter signals. It is demonstrated by experiments that under low signal-to-noise ratio (SNR) conditions of -6 dB, the proposed method effectively suppresses communication signal interference and achieves a sorting accuracy of 93.75%. Compared with existing approaches, the proposed solution demonstrates significant advantages in enhancing the robustness of signal preprocessing and maintaining high sorting accuracy, underscoring its strong potential for practical engineering applications.