Online Admission Control and Resource Allocation in Network Slicing under Demand Uncertainties

TL;DR

This paper addresses the challenge of efficiently managing network slicing in 5G/6G networks under demand uncertainties by formulating a robust online admission control and resource allocation problem, proposing an optimal and a heuristic solution.

Contribution

It introduces a novel mathematical formulation using Binary Linear Programming with $b3$-Robustness for uncertainties and develops a fast heuristic algorithm for large-scale networks.

Findings

Heuristic achieves about 7-10% optimality gaps.

Heuristic is approximately 30 times faster than the optimal algorithm.

Heuristic improves accepted requests and reduces power consumption.

Abstract

One of the most important aspects of moving forward to the next generation networks like 5G/6G, is to enable network slicing in an efficient manner. The most challenging issues are the uncertainties in computation and communication demand. Because the slices' arrive to the network in different times and their lifespans vary, the solution should dynamically react to online slice requests. The joint problem of online admission control and resource allocation considering the energy consumption is formulated mathematically. It is based on Binary Linear Programming (BLP), where, the $\Gamma$-Robustness concept is exploited to overcome Virtual Links (VL) bandwidths' and Virtual Network Functions (VNF) workloads' uncertainties. Then, an optimal algorithm is proposed. This optimal algorithm cannot be solved in a reasonable amount of time for real-world and large-scale networks. To find near-optimal solution efficiently, a new heuristic algorithm is developed. The assessments' results indicate that the efficiency of heuristic is vital in increasing the accepted requests' count, decreasing power consumption and providing adjustable tolerance vs. the VNFs workloads' and VLs traffics' uncertainties, separately. Considering the acceptance ratio and power consumption that constitute the two important components of the objective function, heuristic has about 7% and 10% optimality gaps, respectively, while being about 30X faster than that of optimal algorithm.