An Online Convex Optimization Approach for Network Slicing

Abstract

The emergence of 5G and beyond networks introduces a significant shift in resource allocation across a wide array of services and applications, all supported by a singular physical infrastructure. This advancement necessitates an efficient mechanism design, termed Network Slicing. This concept provides a cohesive abstraction of network and computing resources, allowing for the efficient handling of diverse service requirements. For accelerated processing, computation-intensive tasks can be offloaded to radio access networks (RANs) by utilizing multi-access edge computing (MEC) technology. However, developing a RAN slicing policy encounters challenges. Decisions regarding the allocation of radio spectrum and computational resources impact the availability of these resources at base stations, thus affecting workload distribution. Furthermore, the unpredictable nature and variability of Quality of Service (QoS) requirements further complicate the RAN slicing policy. This work explores the application of online convex optimization (OCO) in network slicing, leveraging its adaptive nature for real-time decision-making in time-variant costs and dynamic network conditions. Emphasis is placed on optimizing resource allocation for communication and computation while satisfying QoS requirements. The first work addresses a network-slicing scenario with dynamic allocation of radio spectrum and computing resources, distributing workloads across multiple base stations. The primary QoS constraint addressed is delay, catering to both delay-sensitive and delay-tolerant services. Adaptability of OCO methods effectively manages the challenges of varying costs and QoS constraints. In the second work, the focus shifts to delay-sensitive services, considering an edge-cloud orchestrated network, leveraging cloud's capacity for efficient task processing. This phase employs a novel OCO method utilizing predictions for cost functions and constraints to distribute workload and allocate computing and communication resources while satisfying QoS. This innovation enhances the algorithm's adaptability to dynamic, time-varying costs within the network, enabling more informed decision-making. Rigorous experimentation and evaluation of the proposed OCO methods demonstrate their effectiveness in achieving optimal solutions for workload distribution and resource allocation. The results highlight the approach's adaptability and scalability in wireless communication networks, contributing valuable insights and practical solutions to network slicing optimization in contexts characterized by varying costs and QoS constraints.