Loading...

LiFi grid: A machine learning approach to user-centric design

Pashazanoosi, M ; Sharif University of Technology | 2020

453 Viewed
  1. Type of Document: Article
  2. DOI: 10.1364/AO.396804
  3. Publisher: OSA - The Optical Society , 2020
  4. Abstract:
  5. A novel machine learning (ML) clustering algorithm, named light-fidelity (LiFi) Grid, is proposed to design amorphous cells of LiFi access points (APs) in order to maximize the minimum signal-to-interference-plus-noise ratio (SINR) from the viewpoint of user-centric (UC) network design. The algorithm consists of two phases. Explicitly, the first phase consists of finding clusters of user densities based on the mean-shift (MS) clustering algorithm. In contrast to some other clustering algorithms, such as K-means, MS does not need to know the number of clusters in advance. Furthermore, the combined transmission scheme is assumed in each cell. In the second phase, this paper proposes a novel clustering algorithm that addresses the problem of grouping APs based on the positions of users—UC design—in optical wireless networks (OWNs). Hence, it addresses the dynamic resource allocation problem in OWNs if APs are considered as network resources. Based on the maximization of minimum SINR metric, LiFi Grid demonstrates the superior performance relative to conventional fixed-shape cell-centric network designs. Additionally, full compatibility of the LiFi Grid clustering algorithm with the Institute of Electrical and Electronics Engineers standard 802.15.7 is also shown. © 2020 Optical Society of America
  6. Keywords:
  7. Machine learning ; Signal interference ; Signal to noise ratio ; Turing machines ; User centered design ; Dynamic resource allocations ; Institute of Electrical and Electronics Engineers ; Machine learning approaches ; Number of clusters ; Optical wireless networks ; Signal to interference plus noise ratio ; Transmission schemes ; User-centric designs ; K-means clustering
  8. Source: Applied Optics ; Volume 59, Issue 28 , 2020 , Pages 8895-8901
  9. URL: https://pubmed.ncbi.nlm.nih.gov/33104575