DCFNet: Doppler Correction Filter Network for Integrated Sensing and Communication in Multi-User MIMO–OFDM Systems

Integrated sensing and communication (ISAC) is a headline feature for the forthcoming IMT-2030 and 6G releases, yet a concrete solution that fits within the established orthogonal frequency division multiplexing (OFDM) family remains an open problem. Specifically, Doppler-induced inter-carrier interference (ICI) destroys subcarrier orthogonality of OFDM sensing signals, blurring range–velocity maps and severely degrading sensing accuracy. Building on multi-user multi-input-multi-output (MIMO) OFDM systems, this paper develops a model-driven ISAC framework that jointly optimizes transmit and receive beamforming to maximize multi-user communication sum-rate under sensing-performance constraints, while mitigating Doppler ICI. To this end, we propose a Doppler-correction filter network (DCFNet), an AI-native ISAC model that achieves fine-grained range-velocity estimation precision with minimal complexity and without altering the legacy frame structure. A bank of DCFs derived from the Doppler physics first shifts and suppresses dominant ICI components, and a subsequent deep neural network cancels the residual interference to yield a clean radar sensing image. To further enhance the range and velocity estimation precision, we propose DCFNet with local refinement (DCFNet-LR), which applies a generalized likelihood ratio test (GLRT) to refine target estimates of DCFNet to sub-cell precision. Simulation results show that DCFNet-LR runs 143 times faster than a maximumlike-lihood search and achieves significantly superior performance, reducing the range and velocity RMSE by factors of 2.7 × 10−4and 6.7 × 10−4compared to conventional detection methods.