Sonia Heemstra de Groot

Anudeep Karnam, Kishor C. Joshi, Jobish John et al.

Future wireless networks must deliver deterministic end-to-end delays for workloads such as smart-factory control loops. On Ethernet these guarantees are delivered by the set of tools within IEEE 802.1 time sensitive networking~(TSN) standards. Credit-based shaper (CBS) is one such tool which enforces bounded latency. Directly porting CBS to 5G/6G New Radio (NR) is non-trivial because NR schedules traffic in discrete-time, modulation-dependent resource allocation, whereas CBS assumes a continuous, fixed-rate link. Existing TSN-over-5G translators map Ethernet priorities to 5G quality of service (QoS) identifiers but leave the radio scheduler unchanged, so deterministic delay is lost within the radio access network (RAN). To address this challenge, we propose a novel slot-native approach that adapts CBS to operate natively in discrete NR slots. We first propose a per-slot credit formulation for each user-equipment ({UE}) queue that debits credit by the granted transport block size~(TBS); we call this discrete-time CBS (CBS-DT). Recognizing that debiting the full {TBS} can unduly penalize transmissions that actually use only part of their grant, we then introduce and analyze {CBS} with Partial Usage ({CBS-PU}). {CBS-PU} scales the credit debit in proportion to the actual bytes dequeued from the downlink queue. The resulting CBS-PU algorithm is shown to maintain bounded credit, preserve long-term rate reservations, and guarantees worst-case delay performance no worse than {CBS-DT}. Simulation results show that slot-level credit gating--particularly CBS-PU--enables NR to export TSN class QoS while maximizing resource utilization.

Bram Van Bolderik, Vlado Menkovski, Sonia Heemstra de Groot et al.

Neural network based receivers have recently demonstrated superior system-level performance compared to traditional receivers. However, their practicality is limited by high memory and power requirements, as separate weight sets must be stored for each code rate. To address this challenge, we propose LOREN, a Low Rank-Based Code-Rate Adaptation Neural Receiver that achieves adaptability with minimal overhead. LOREN integrates lightweight low rank adaptation adapters (LOREN adapters) into convolutional layers, freezing a shared base network while training only small adapters per code rate. An end-to-end training framework over 3GPP CDL channels ensures robustness across realistic wireless environments. LOREN achieves comparable or superior performance relative to fully retrained base neural receivers. The hardware implementation of LOREN in 22nm technology shows more than 65% savings in silicon area and up to 15% power reduction when supporting three code rates.