Novel Adaptive Multi-User Multi-Services Scheduling to Enhance Throughput in 5G-Advanced and Beyond

Optimum network slice resource allocation for multiple User Equipment (UE) requesting several services concurrently in a scalable manner is a complex open research problem. Each network slice is designed to support distinct applications with corresponding Service Level Agreements (SLAs) across multiple UEs. The distribution of network resources across slices has been a challenge to achieve the SLAs under limited bandwidth and power budget constraints. This is referred to as inter-slice resource allocation, where a UE may simultaneously request multiple services hosted across slices. When a UE is assigned a slice, the slice shares the available resources across new and existing UEs without violating the SLA. This is referred to as intra-slice scalability. While inter-slice resource allocation has been addressed in the literature, further throughput gains can be achieved by improving intra-slice scalability. This paper presents a novel Probabilistic Intra-slice Resource Service Scheduling (PRSS) method. PRSS algorithm works in two stages. In the first stage, the service throughput is estimated using a multinomial probabilistic model, followed by dynamic conditional resource estimation in the second stage. For newly requested services, the resources are re-estimated and reconfigured across prior inter and intra-slice services. The proposed two-stage method allows the system to be highly adaptive with respect to the number of UEs and their requested services hosted across slices. Analytical and simulation results show that by using PRSS, the number of services served with a better experience in terms of service quality, i.e., throughput is enhanced by 20% to 75% compared to state-of-the-art schedulers.