O-RAN fronthaul-linkkien linkkiagregaatioryhmien ohjelmistototeutus linkkien kestävyyden parantamiseksi
Leo Talsta · Aaltodoc (Aalto University) · 2026
High reliability is a key requirement for modern 5G mobile networks. Link Aggregation Groups (LAG) provide a method to increase connection reliability by utilizing redundant parallel links. Existing open-source LAG features, such as the Linux bonding driver, allow users to flexibly set up LAGs between Linux-operated machines, such as the System-on-Chip (SoC) devices used in O-RAN-based 5G mobile networks. Although the use of Linux bonding LAGs would improve the reliability of O-RAN fronthaul links, they are currently not widely used in mobile networks. This is because Linux bonding can only access the software-processed M-plane traffic, leaving the hardware-processed U/C-planes outside of the scope of the driver. The objective of this thesis is to enable the use of Linux Bonding LAG implementation within the Nokia O-RAN fronthaul for aggregating hardware-processed eCPRI traffic in the User and Control planes. For this purpose, a LAG compatibility layer was designed and implemented for the Nokia Ethernet Subsystem (ETHSS) software, allowing the ETHSS to replicate the bonding adjustment behavior of the driver for the User and Control planes. A System Component Test (SCT) was created for engineering samples of Nokia 5G SoCs to determine the effectiveness of the implementation. The developed LAG feature reduced the packet loss rate of a two-link LAG setup from 50\% to 0\% when one of the bonded links was brought down. The feature was compatible with both active-active and active-passive LAG configurations, and performed bond adjustment actions in 7.7 ms (active-active) and 8.0 ms (active-passive). The test results show that the implemented LAG compatibility feature accurately replicates the behavior of the Linux bonding driver. Using the feature, O-RAN fronthaul traffic can be routed through a LAG interface for increased connection reliability. Future work includes evaluating dynamic Link Aggregation Control Protocol (LACP) interfaces and advanced load sharing algorithms to further enhance adaptability and performance.