Energy Efficient VLC based Fog Architectures

Sixth generation (6G) networks are envisioned to provide multi-terabit per second (Tbps) communication, very low latency, high spectrum and energy efficiency, and high connectivity density. Visible light communication (VLC) has gained significant attention as one of the technologies to transition the envisaged features of 6G into actuality. With the increasing availability of fog resources in proximity to end users, and the potential for dense deployment of VLC access points in indoor environments, it is essential to design energy and cost-efficient backhaul network architectures to establish connectivity between the indoor VLC access points connecting fog resources to the rest of the network. This thesis adopts several technologies to deliver end-to-end energy-efficient communication for a fog-cloud system with indoor fog resources.
Firstly, we leverage the advantages of passive optical networks (PONs) and propose an array waveguide grating router (AWGR) based PON for a VLC backhaul network. A mixed integer linear program (MILP) model is formulated to place processing demands in the cloud-fog resources over the proposed architecture in a manner that maximises energy efficiency. The model minimises the processing and networking power consumption under preconfigured VLC resources (i.e., access points and wavelength selection). Relative to a backhaul network based on the traditional spine-and-leaf architecture, total power consumption savings of up to 95% are achieved.
Secondly, we investigate several techniques to improve the performance of the AWGR PON-based backhaul architecture. We enhance the utilisation of available networking and processing resources and improve the energy efficiency of the architecture by allowing processing splitting among multiple nodes and introduce dynamic allocation of bandwidth to access points. The adoption of these enhancements improves the total power savings by 71% and 59%, respectively. We also show that expanding the fog architecture to exploit fog resources in neighbouring buildings can reduce the total power consumption by up to 24%. Furthermore, we enhance the architecture by adding a layer of AWGRs to establish passive connections to the fog resources in all rooms resulting in total power consumption savings of up to 59%.
Thirdly, we investigate the influence of jointly optimising the fog and VLC resource allocation over the proposed architecture. The MILP model is extended to optimise the selection of the access points and wavelengths along with the placement of the processing demands while minimising the total power consumption. The joint optimisation results in total power savings of up to 13% compared to the case where access points and wavelengths are preconfigured.
Finally, we propose a point-to-point (P2P) PON-based architecture as a VLC backhaul network. The results indicate that the adoption of the P2P PON architecture results in lower power consumption, better resource utilisation, and lower latency compared to the AWGR PON architecture. Moreover, the P2P PON architecture can potentially reduce the deployment cost of the indoor backhaul network.