Network Slicing in 5G Edge Networks with Controlled Slice Redistributions

Abstract

Edge computing with network slicing enables 5G networks to meet the diverse and stringent requirements of new services and applications. We study the problem of admitting network slice requests and serving currently active slices in the 5G edge network, while avoiding active slice redistributions in the network. We pose it as a constrained optimization problem that seeks to maximize the total reward or revenue to the network operator from serving the new and active slices minus a term that penalizes slice redistributions in the network. We propose a three-phase polynomial-time, greedy, heuristic approach called RESET to solve this NP-hard problem. The first phase employs a cost function that determines the order in which new slice requests and a fraction of the active slices are served. It takes into account the bandwidth, storage, and computing requirements of a slice request. The second phase selects an edge cloud (EC) to assign an admitted request based on the residual bandwidth of its incoming links, and storage and computing resources at the EC. The last phase determines the forwarding path to route the traffic associated with the slice request to the selected EC. Extensive simulation results show that RESET achieves a total reward that is competitive with the optimal solution and is higher than benchmark schemes, while requiring far fewer slice redistributions.