Optimizing fronthaul resource allocation for 6G RAN

Optimizing fronthaul resource allocation for 6G RAN BSC Severo Ochoa Doctoral Symposium · May 6-8, 2025 · Barcelona (Spain) · Poster Session Dynamic bandwidth allocation Abstract: As wireless networks move toward 6G, Radio Access Networks (RANs) face increasing demands for higher speed, greater bandwidth efficiency, and lower latency, particularly in the fronthaul segment. Passive Optical Networks (PONs) offer a cost-effective way to support this, but emerging new applications like AI and extended reality make scheduling fronthaul traffic more complex. This work explores smarter scheduling methods to optimize fronthaul network performance, enabling efficient and reliable support for future 6G applications. Xiang Lu1*, H. Shakespear-Miles1 , and Sima Barzegar2 In Radio Access Networks (RANs), the fronthaul network specifically connects the distributed unit (DU) and the radio units (RUs). Passive Optical Networks (PONs) have emerged as a popular and cost-effective solution to support fronthaul networks and have great potential for deploying edge computing nodes.As the need for high bandwidth and low latency becomes more critical, the fronthaul network becomes a vital bridge between mobile devices and data centers, where high-performance computing (HPC) systems are typically deployed. Results (1) Universitat Politècnica de Catalunya (UPC), Barcelona, Spain (2) Barcelona Super Computing Center (BSC), Barcelona, Spain *e-mail: [email protected] Passive Optical Networks (PON) A PON typically consist of one Optical Line Terminal (OLT) connected to multiple Optical Network Units (ONUs) via a point-to-multipoint topology. Carrying upstream fronthaul traffic via PONs introduces extra latency due to the status-reporting mechanisms. For this, Cooperative DBA (CO-DBA) was standardized for 4G/5G fronthaul to reduce latency. However, emerging 6G applications, such as AI and extended reality, present new and greater challenges for fronthaul traffic management. Fronthaul Network in RAN We developed a Flexible Queue Management Dynamic Bandwidth Allocation (FQM-DBA) algorithm that better supports the diverse latency requirements of time-sensitive services. This algorithm can be effectively applied to fronthaul over PON, capable of meeting the performance demands of various time-sensitive 6G applications. where deterministic latency guarantees and efficient bandwidth utilization are essential. Simulations were conducted using our Python-based simulation platform, considering a scenario of a50G-PON system with one OLT serving 64 ONUs. Both low (0.3) and high (0.9) network loads were evaluated. T-CONT T-CONT T-CONT High priority High latency tolerance Data frame Report message OLT ONU Create Remove Data frame grouping Queue monitoring 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.2 0.4 0.6 0.8 1.0 PSR (a) Load IT 1 (Benchmark 1) IT 2 (Benchmark 1) IT 1 (Benchmark 2) IT 2 (Benchmark 2) IT 1 (FQM-DBA) IT 2 (FQM-DBA) CDF (b) Latency (ms) IT 1 (Benchmark 1) IT 2 (Benchmark 1) IT 1 (Benchmark 2) IT 2 (Benchmark 2) IT 1 (FQM-DBA) IT 2 (FQM-DBA) IT 1 latency limit=0.8 ms IT 2 latency limit=1.5 ms CDF (c) Latency (ms) IT 1 (Benchmark 1) IT 2 (Benchmark 1) IT 1 (Benchmark 2) IT 2 (Benchmark 2) IT 1 (FQM-DBA) IT 2 (FQM-DBA) IT 1 latency limit=0.8 ms IT 2 latency limit=1.5 ms 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0.0 0.2 0.4 0.6 0.8 1.0 PSR (a) Load OT 1 (Benchmark 1) OT 2 (Benchmark 1) OT 1 (Benchmark 2) OT 2 (Benchmark 2) OT 1 (FQM-DBA) OT 2 (FQM-DBA) CDF (b) Latency (ms) OT 1 (Benchmark 1) OT 2 (Benchmark 1) OT 1 (Benchmark 2) OT 2 (Benchmark 2) OT 1 (FQM-DBA) OT 2 (FQM-DBA) OT 1 jitter limit=0.4 ms OT 1 latency limit=1 ms OT 2 latency limit=2 ms OT 2 jitter limit=0.6 ms CDF (c) Latency (ms) OT 1 (Benchmark 1) OT 2 (Benchmark 1) OT 1 (Benchmark 2) OT 2 (Benchmark 2) OT 1 (FQM-DBA) OT 2 (FQM-DBA) OT 1 latency limit=1 ms OT 1 jitter limit=0.4 ms OT 2 latency limit=2 ms OT 2 jitter limit=0.6 ms Two service types were considered: in-time services require a maximum latency guarantee, and on-time services impose limits on latency and jitter. Three schemes were evaluated: Benchmark 1 (IACG algorithm + FIFO), Benchmark 2 (IACG algorithm + Strict Priority scheduling), and our proposed FQM-DBA algorithm. FQM-DBA can effectively support both time-sensitive services and significantly outperforms traditional schemes, particularly under high network load.