Real-time communication (RTC) applications, such as virtual reality and video conferencing, impose stringent ultra-low latency requirements, making efficient video rate control a critical challenge in 5G networks. In scenarios where wireless link conditions fluctuate rapidly, ensuring latency performance while maintaining high bandwidth utilization poses a significant difficulty for existing congestion control algorithms. Conventional schemes primarily rely on end-host measurements and reactive adjustments, which often suffer from delayed responses to network dynamics, leading to elevated tail latency and inefficient bandwidth usage. To address these limitations, this paper proposes PIKA, a deployable link-assisted congestion control mechanism specifically designed for real-time video streaming over 5G networks. PIKA exploits bandwidth and queue state information collected at the base station to compute congestion signals in real time. These signals are embedded into packet headers within the 5G core network and subsequently parsed by the receiver, which feeds the information back to the sender via ACKs. Based on this feedback, the sender performs fine-grained bitrate adjustments, enabling timely adaptation to network dynamics. Experimental results demonstrate that PIKA outperforms both conventional congestion control protocols and emerging RTC-oriented approaches in 5G edge environments, achieving superior performance in terms of end-to-end latency and average bitrate.