The increasing demand for real-time applications such as cloud gaming, augmented/virtual reality (AR/VR), remote control, and industrial automation, has placed stringent requirements on mobile networks to deliver ultra-low latency and high reliability. As 5G networks evolve, ensuring consistently low delays, even during congestion periods, is critical for these real-time applications.

This thesis investigates two network-assisted rate adaptation mechanisms: Low Latency Low Loss Scalable Throughput (L4S) and Access Network Bitrate Recommendation (ANBR). Both mechanisms aim to reduce latency and packet loss while maximizing throughput during periods of congestion. L4S, standardized by 3GPP and IETF, uses Explicit Congestion Notification (ECN) marking in the IP header of the packets, where the base station marks packets to signal early signs of congestion. This allows the sender to react promptly and adjust its transmission rate using a scalable congestion control algorithm. ANBR, also standardized by 3GPP, takes a different approach by providing rate recommendations from the base station to the user equipment (UE) using MAC layer messages.

While both technologies share similar goals, L4S has seen significant industry interest in recent times, whereas ANBR remains relatively underexplored. Despite their potential and similarities, the coexistence of these two technologies and suitability for different scenarios have not been thoroughly investigated.

This research done in collaboration with KPN, addresses this gap by evaluating the comparative performance, suitability, and coexistence of L4S and ANBR for different network scenarios. The research combines theoretical and practical analysis. The units of research include literature and standards reviews, simulations using ns-3, and practical experiments conducted at KPN's test lab. Latency, packet loss, and throughput are analyzed for each experiment.

The findings provide insights into the advantages and disadvantages of L4S and ANBR, and highlight the applications for which they are most suitable. Based on the findings, recommendations are proposed to guide the effective adoption and integration of L4S and/or ANBR in KPN.
A key finding from the research is that L4S is better suited for applications requiring ultra-low latency, while ANBR is more appropriate for applications with higher throughput sensitivity. With L4S, telecom operators can have better control over latency and define queueing thresholds at which rate adaptation should begin for the applications, enabling them to better ensure that the Quality of Service (QoS) requirements of each application are met. In contrast, ANBR does not directly target queueing delay; instead, it uses a window mechanism to send rate recommendations to the UE, which limits its ability to control latency.