Link Selection in Hybrid RF/VLC Systems under Statistical Queueing Constraints

TL;DR

This paper investigates how to optimally select between RF and VLC links in indoor environments to meet QoS constraints related to buffer overflow and delay, using a multi-mechanism approach with performance analysis.

Contribution

It introduces a link selection strategy for hybrid RF/VLC systems under QoS constraints, considering power limits and physical layer parameters, with comprehensive performance evaluation.

Findings

Optimal link selection improves QoS guarantees.

Using both RF and VLC simultaneously enhances performance.

Physical layer parameters significantly impact system performance.

Abstract

The co-deployment of radio frequency (RF) and visible light communications (VLC) technologies has been investigated in indoor environments to enhance network performances and to address specific quality-of-service (QoS) constraints. In this paper, we explore the benefits of employing both technologies when the QoS requirements are imposed as limits on the buffer overflow and delay violation probabilities, which are important metrics in designing low latency wireless networks. Particularly, we consider a multi-mechanism scenario that utilizes RF and VLC links for data transmission in an indoor environment, and then propose a link selection process through which the transmitter sends data over the link that sustains the desired QoS guarantees the most. Considering an ON-OFF data source, we employ the maximum average data arrival rate at the transmitter buffer and the non-asymptotic bounds on data buffering delay as the main performance measures. We formulate the performance measures under the assumption that both links are subject to average and peak power constraints. Furthermore, we investigate the performance levels when either one of the two links is used for data transmission, or when both are used simultaneously. Finally, we show the impacts of different physical layer parameters on the system performance through numerical analysis.