Most-Correlated Distribution-Based Load Balancing Scheme in Hybrid LiFi/WiGig Network

Recently, a hybrid network that combines radio frequency Wireless Gigabit Alliance (WiGig) networks with light fidelity (LiFi) networks has been proposed as the foundation for a high-speed wireless communication solution. A LiFi access point provides the service through the limited coverage area, LiFi attocell. Hence, LiFi networks could efficiently apply the frequency reuse concept to enhance spatial-spectral efficiency. Unfortunately, when the number of user equipment (UE) increases, new obstacles are added to those that the LiFi networks already face, such as light path obstruction, poorly aligned connections, and handover, in addition to uplink and mobility issues. To solve these issues and raise network quality of service (QoS), the hybrid LiFi/RF network has been suggested. In such networks, simultaneously and in a totally different frequency range, WiGig access points could provide tremendous data rates (gigabits per second) using the massive bandwidth of the Millimeter-Wave (mm-Wave) spectrum. Nevertheless, such hybrid networks need an effective load balancing (LB) strategy to assign the best access point (AP) and distribute enough resources for each UE depending on the location distributions of UEs (the channel between UEs and APs). The traditional LB approaches, however, use complex iterative computing procedures for each new distribution of UEs. Therefore, the Most-Correlated Distribution (MCD) Based Load Balancing Scheme is suggested in this work. The suggested method is clever enough to exploit the history of all prior load-balancing outcomes, recorded in a Distributions-Decisions Record (DDR), in order to identify appropriate allocations for the new UEs distribution, rather than going into repeated intensive complex calculations. The DDR is a list of the most common users’ distributions and the corresponding best AP allocation decisions which are calculated via the Consecutive Assign WiGig First SOA (CAWFS) LB Algorithm. The DDR record is created once, and off-line via the center processing unit (CPU). Each row in the DDR is composed of the supposed distribution and the corresponding decisions. Given the new mobile user distribution, the subset of the DDR records, that contains the most correlated distributions, is constructed. The current decisions are chosen depending on the previous decisions in the selected subset via the majority voting technique. In comparison to previous load-balancing algorithms, the proposed approach intends to provide equivalent attainable data rates and outage probability performances at lower complexity.