Low-Latency Upstream Scheduling in Multi-Tenant, SLA Compliant TWDM PON

TL;DR

The paper addresses the challenge of delivering low-latency, SLA-driven upstream scheduling in multi-tenant TWDM-PONs. It extends the vDBA concept to a multi-channel TWDM-PON by introducing a stateful, SLA-aware bandwidth-map merging engine and a MILP-based formulation (involving a 4D binary matrix $X$) to minimize SLA breaches, supported by flow-breach likelihood and channel-freetime data. Key findings show that multi-channel scheduling yields better SLA adherence at low ONU tuning times, but gains diminish as tuning times grow, with tuning-time regimes around $250\,\mathrm{ns}$ to $1\,\mathrm{\mu s}$ identified as important for maintaining performance. The work enables SLA-compliant, multi-tenant PONs for 5G/6G fronthaul and beyond, providing a path toward flexible, channel-aware upstream sharing under strict latency guarantees.

Abstract

We present a multi-tenant multi-wavelength upstream transmission scheme for virtualised PONs, enabling compliance with latency-oriented Service Level Agreements (SLAs). Our analysis highlights an important trade-off between single-channel vs. multi-channel PONs, depending on ONUs tuning time.

Figures (7)

• Figure 1: TWDM multi-tenant upstream scheduling algorithm
• Figure 2: MILP Notations and Equations
• Figure 3: Pseudocode for TWDM
• Figure 4: Tuning time: 0s
• Figure 5: Tuning time: 250ns