To address the challenges posed by the diverse demands of mobile users and vertical industries, it is crucial for 5G-advanced (5G-A) networks to exhibit flexibility in system design across both the physical and higher layers. A single-frame structure alone cannot provide the necessary flexibility to accommodate multiple traffic modes. Therefore, we investigate a heterogeneous frame structure that meets the requirements. However, this structure will cause severe cross-link interference (CLI) and degrade the performance. To solve this problem, four CLI mitigation methods are proposed: slot level adaptive modulation and coding (AMC) scheduling (method 1), interference mitigation algorithm (method 2), adaptive slot turn-off (method 3), and edge user adaptive scheduling (method 4). The experimental results show that the CLI problem can be completely mitigated by deployed method 3 in a macro-base station (BS) with a low load scenario, the micro-BS uplink (UL) throughput was improved by 75.8%. Otherwise, with macro-BS in the high load scenario, when method 4 was implemented in macro-BS and methods 1 and 2 were deployed in micro-BS, a nearly complete CLI mitigation effect was obtained, reflecting that the micro-BS UL throughput increased to 57.4%. These methods are elaborated upon and evaluated within practical 5G-A networks, and the experimental results not only demonstrate the viability and efficacy of the above methods but also highlight their potential applications in future usage scenarios.