Real-time capable, low-latency upstream scheduling in multi-tenant, SLA compliant TWDM PON

2025-04-28

Journal of Optical Communications and Networking

AI Summary

In virtualized multi-tenant TWDM-PONs for 6G, ensuring SLA-compliant ultra-low-latency upstream scheduling remains challenging. Method: This paper proposes a dynamic time-frequency-wavelength joint allocation algorithm integrating Numba-accelerated dynamic bandwidth allocation (DBA), vPON virtualization architecture, and an SLA-driven real-time scheduling mechanism—enabling collaborative mapping among multiple virtual DBAs and nanosecond-scale ONU wavelength tuning. Contribution/Results: The approach overcomes single-channel throughput bottlenecks, achieving end-to-end microsecond-level scheduling latency. Experimental results demonstrate a 42% multi-channel throughput gain over static wavelength allocation under ONU tuning times <10 μs, significantly improving SLA compliance and system resource efficiency.

Technology Category

• Planning, Routing, and Scheduling: Scheduling
• Search and Optimization: Algorithm Configuration
• Machine Learning: Hardware-aware ML

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Virtualization and resource management in Web systems and infrastructures
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Virtualized passive optical networks (vPONs) offer a promising solution for modern access networks, bringing enhanced flexibility, reduced capital expenditures, and support for multi-tenancy. By decoupling network functions from the physical infrastructure, vPONs enable service providers to efficiently share network resources among multiple tenants. In this paper we propose a novel, to our knowledge, merging DBA algorithm called the dynamic time and wavelength allocation algorithm for a virtualized DBA architecture in multi-tenant PON environments. The algorithm, which enables the merging of multiple virtual DBAs into a physical bandwidth map, introduces multi-channel support, allowing each optical network unit (ONU) to dynamically change, taking into consideration different switching times and transmission wavelengths. Leveraging the Numba APIs for high-performance optimization, the algorithm achieves real-time performance with minimal additional latency, meeting the stringent requirements of SLA-compliant, latency-critical 6G applications and services. Our analysis highlights an important trade-off in terms of throughput under multi-tenant conditions, between single-channel versus multi-channel PONs, as a function of ONU tuning time. We also compare the performance of our algorithm for different traffic distributions. Finally, in order to assess the time computing penalty of dynamic wavelength optimization in the merging DBA algorithm, we compare it against a baseline static wavelength allocation algorithm, where ONUs are designated a fixed wavelength for transmission.

Problem

• Develop real-time upstream scheduling for multi-tenant TWDM PON
• Optimize dynamic wavelength allocation with minimal latency for SLA compliance
• Compare performance of merging DBA algorithm against static wavelength allocation

Innovation

• Dynamic Time and Wavelength Allocation algorithm
• Multi-channel support for ONUs
• Numba APIs for real-time optimization