Department of Computer Science and Technology

Technical reports

Improving commercial LiFi network feasibility through rotation invariance, motion prediction, and bandwidth aggregation at the physical layer

Daniel M. Fisher, Jon A. Crowcroft

November 2021, 79 pages

This technical report is based on a dissertation submitted May 2020 by the first author for the degree of Master of Philosophy (Advanced Computer Science) to the University of Cambridge, Downing College.

DOI: 10.48456/tr-965

Abstract

In recent years, the number of devices per household has exponentially increased. Additionally, Internet traffic surged during the 2020–2021 calendar years due to the COVID-19 pandemic. Both of these occurrences have served as reminders that the spectrum available for radio frequency (RF) based networking is running thin, as internet service providers had to find creative ways to serve this growth of data demand. Introduced in 2011 by Dr. Harald Haas, Light Fidelity, or LiFi, offers a suitable supplement to ameliorate this challenge as well as offer faster data rates to consumers. However, due to the line-of-sight nature of data transmission required by LiFi, path blockage to the photodetector and natural light obstruction have proven to be significant challenges in commercially implementing the computer network. Hybrid networks have been proposed, serving as a middle ground where the networking load is balanced between LiFi and WiFi depending on which is available. However, many challenges exist with this concept, in particular with optimizing the logic so it is worthwhile.

The goal of this thesis is to further develop this field and push LiFi closer to being commercially practical for the next generation of computer networks. Three lines of effort are pursued to do so. First, the path blockage problem is mitigated through a novel rotationally invariant photodetector configuration that allows for significantly better sensor visibility than previous solutions. Additionally, horizontal handovers from one LiFi hub to another, as a user moves through a space, is improved via leveraging probabilistic machine learning (ML) and other motion tracking algorithms to more accurately predict human movement and identify “hot spots.” Lastly, a novel bandwidth aggregation scheme at the physical layer is proposed and preliminarily evaluated through a simulation to significantly improve data rates when both LiFi and WiFi are available.

Full text

PDF (2.8 MB)

BibTeX record

@TechReport{UCAM-CL-TR-965,
  author =	 {Fisher, Daniel M. and Crowcroft, Jon A.},
  title = 	 {{Improving commercial LiFi network feasibility through
         	   rotation invariance, motion prediction, and bandwidth
         	   aggregation at the physical layer}},
  year = 	 2021,
  month = 	 nov,
  url = 	 {https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-965.pdf},
  institution =  {University of Cambridge, Computer Laboratory},
  doi = 	 {10.48456/tr-965},
  number = 	 {UCAM-CL-TR-965}
}