Self-Adaptive Resource-Allocation Scheme Combining Traffic Prediction With User Satisfaction in Multi-Wavelength VPON

Abstract

In this paper, a self-adaptive resource-allocation scheme combining traffic prediction with user satisfaction is firstly established in multi-wavelength VPON. Among VPONs, by proposing a (wavelength utility)-based dynamic wavelength allocation algorithm, the optimal allocation of wavelengths can be achieved. In a VPON: under condition that the total requested bandwidth is larger than the total bandwidth, by introducing a bandwidth constraint factor ε, the maximum amount of bandwidth granted for high-priority service can be limited, and the "starvation" problem of low-priority service can be prevented. By solving the user satisfaction model, not only the optimal allocation of low-priority service can be obtained, but also the comprehensive user satisfaction is maximized. Under condition that the total requested bandwidth is smaller than the total bandwidth, by combining an ANN prediction model with a user satisfaction model, a bandwidth allocation model based on traffic prediction and user satisfaction is created. By this model, not only the resource utilization is improved but also the transmission delay is reduced. At the same time, user satisfaction is maximized. By adjudicating on the set judgment condition, the self-adaptive function of control system for the above two conditions is implemented. By simulation, the effectiveness of the proposed scheme is demonstrated.

Energy-Efficient Sleep-Aware Slicing-Based Scheduler (SA-SBS) for Multi-Operators Virtualized Passive Optical Networks

“…The study investigated the performance of both the SBS scheme and the proposed SA-SBS scheme, with two different 6) to (24).…”

“…a) MO-VPON: In the MO-vPON approach of NFV, a centralized physical OLT (main server OLT) virtualizes multiple OLT processes or manages multiple client OLTs. Each client OLT is responsible for a subset of respective ONUs in vPON and has its bandwidth allocation policies and upstream bandwidth maps [24]. The client OLT processes can cooperate with the help of the centralized physical OLT to provide better services to their ONUs.…”

“…The study includes comparative graphs in Figs. (22) to (24) illustrate the overall mean delay for sleep modes (DR and QR) and two algorithms (SA-SBS and SBS) for three types of traffic (Assured, Non-Assured, and Best Effort) at different offered traffic loads. The evidence of the arguments for the US delay can be witnessed in these graphs, which demonstrate that the SA-SBS scheme with DR and QR outperforms the original SBS scheme in reducing upstream delays for efficient bandwidth utilization of the ONUs in AS state.…”