Sleep/doze controlled dynamic bandwidth allocation algorithms for energy-efficient passive optical networks.

Area of Science:

• Telecommunications Engineering
• Network Energy Efficiency
• Optical Network Optimization

Background:

• 10 Gbps Ethernet Passive Optical Networks (10G-EPONs) face challenges in energy consumption.
• Optical Network Units (ONUs) with Vertical-Cavity Surface-Emitting Lasers (VCSELs) can enter sleep modes to save power.
• Traditional dynamic bandwidth allocation algorithms (DBAs) may not fully leverage ONU sleep capabilities.

Purpose of the Study:

• To comparatively evaluate two novel just-in-time (JIT) dynamic bandwidth allocation algorithms (DBAs) for 10G-EPONs.
• To assess the energy-efficiency improvements offered by JIT and J-FIT algorithms using VCSEL ONUs.
• To analyze the trade-offs between energy savings and average delay for different DBA strategies.

Main Methods:

• Simulation-based performance evaluation of two JIT DBAs: JIT (varying polling cycles) and J-FIT (fixed polling cycles).
• Comparison of energy efficiency between VCSEL ONUs and always-ON Distributed Feedback (DFB) laser ONUs.
• Analysis of key performance metrics: polling cycle time, power consumption, energy savings percentage, and average delay.

Main Results:

• Both JIT and J-FIT DBA algorithms demonstrate enhanced energy efficiency in 10G-EPONs.
• The J-FIT algorithm achieves superior energy savings, particularly under low network load conditions.
• The JIT algorithm offers a lower average delay compared to J-FIT, with both remaining within acceptable Quality of Service (QoS) limits.

Conclusions:

• JIT and J-FIT DBA algorithms effectively improve the energy efficiency of 10G-EPONs by utilizing ONU sleep modes.
• J-FIT is more energy-efficient at lower loads, while JIT minimizes average delay.
• Both algorithms provide a viable path towards energy-saving next-generation access networks while maintaining QoS.