A GenAI-Assisted Framework for Predictive Bandwidth Allocation in TWDM-PON

Time and Wavelength Division Multiplexed Passive Optical Networks (TWDM-PONs) need adaptive resource allocation, but conventional mechanisms react only after traffic demand is reported. This paper proposes a Generative Artificial Intelligence (GenAI)-assisted framework that integrates Software-Defined Networking (SDN), virtual Optical Line Termination Hardware Abstraction (vOLTHA), GenAI-based traffic forecasting, and Dynamic Bandwidth Allocation (DBA), and separates the forecasting model from the allocation rules. A Python-based simulation evaluates one realization of the framework, a predictive DBA in which forecasted Assured Forwarding (AF) demand guides Optical Network Unit (ONU) service ordering and bounded bandwidth allocation, with static ONU-to-wavelength assignments. Without any training on PON data, the TimesFM 3 foundation model gave the lowest mean absolute error in predicting the current-cycle AF grant, 57.92% and 53.51% lower than Long Short-Term Memory (LSTM) for two traffic compositions, and a 20.87% lower error than LSTM in forecasting AF arrivals. At the network level, the proposed DBA reduced the mean AF packet delay by about 17.3% relative to Interleaved Polling with Adaptive Cycle Time (IPACT), while preserving aggregate throughput and introducing no Best Effort (BE) packet loss up to 80% offered load.

Authors

Institutions

Publication Details

Journal
Photonics
Published
2026-11-19
DOI
https://doi.org/10.3390/photonics13111091
Primary Topic
Advanced Photonic Communication Systems
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A GenAI-Assisted Framework for Predictive Bandwidth Allocation in TWDM-PON

I‐Shyan Hwang, Andrew Fernando Pakpahan
Photonics
Advanced Photonic Communication Systems
article

A GenAI-Assisted Framework for Predictive Bandwidth Allocation in TWDM-PON

I‐Shyan Hwang, Andrew Fernando Pakpahan
article en

Abstract

Time and Wavelength Division Multiplexed Passive Optical Networks (TWDM-PONs) need adaptive resource allocation, but conventional mechanisms react only after traffic demand is reported. This paper proposes a Generative Artificial Intelligence (GenAI)-assisted framework that integrates Software-Defined Networking (SDN), virtual Optical Line Termination Hardware Abstraction (vOLTHA), GenAI-based traffic forecasting, and Dynamic Bandwidth Allocation (DBA), and separates the forecasting model from the allocation rules. A Python-based simulation evaluates one realization of the framework, a predictive DBA in which forecasted Assured Forwarding (AF) demand guides Optical Network Unit (ONU) service ordering and bounded bandwidth allocation, with static ONU-to-wavelength assignments. Without any training on PON data, the TimesFM 3 foundation model gave the lowest mean absolute error in predicting the current-cycle AF grant, 57.92% and 53.51% lower than Long Short-Term Memory (LSTM) for two traffic compositions, and a 20.87% lower error than LSTM in forecasting AF arrivals. At the network level, the proposed DBA reduced the mean AF packet delay by about 17.3% relative to Interleaved Polling with Adaptive Cycle Time (IPACT), while preserving aggregate throughput and introducing no Best Effort (BE) packet loss up to 80% offered load.

PhotonicsVol. 13(11)
Universitas Advent Indonesia (ID), Yuan Ze University (TW)
Advanced Photonic Communication Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon