This paper addresses the issue of uplink integrated sensing and communication (ISAC) in 6G wireless networks, where the sensing echo signal and communication signal are received simultaneously at the base station (BS). To mitigate mutual interference between sensing and communication, the authors propose a joint design of sensing transmit waveform and communication receive beamforming, aiming to maximize the weighted sum of normalized sensing rate and normalized communication rate. The problem is formulated as a non-convex optimization, which is computationally complex and challenging to solve using conventional methods. To address this, the authors perform equivalent transformations to reduce the design complexity and develop a deep learning (DL)-based scheme called "ISACNN" to enhance the overall performance of ISAC. The DL-based scheme is trained using a negative objective function as the loss, and it effectively learns the optimal sensing waveform and communication beamforming. The effectiveness and robustness of the proposed DL-based scheme are validated through theoretical analysis and simulation results. The paper also discusses the impact of various parameters such as sensing SNR, communication SNR, number of BS antennas, and number of CUs on the system performance.This paper addresses the issue of uplink integrated sensing and communication (ISAC) in 6G wireless networks, where the sensing echo signal and communication signal are received simultaneously at the base station (BS). To mitigate mutual interference between sensing and communication, the authors propose a joint design of sensing transmit waveform and communication receive beamforming, aiming to maximize the weighted sum of normalized sensing rate and normalized communication rate. The problem is formulated as a non-convex optimization, which is computationally complex and challenging to solve using conventional methods. To address this, the authors perform equivalent transformations to reduce the design complexity and develop a deep learning (DL)-based scheme called "ISACNN" to enhance the overall performance of ISAC. The DL-based scheme is trained using a negative objective function as the loss, and it effectively learns the optimal sensing waveform and communication beamforming. The effectiveness and robustness of the proposed DL-based scheme are validated through theoretical analysis and simulation results. The paper also discusses the impact of various parameters such as sensing SNR, communication SNR, number of BS antennas, and number of CUs on the system performance.