Cost-Effective Radio and NFV Resource Allocation: E2E QoS Provision

TL;DR

This paper introduces an end-to-end resource allocation framework for future wireless networks that optimizes radio and NFV resources to minimize costs while ensuring diverse QoS requirements, using novel algorithms.

Contribution

It presents a new optimization model for joint radio and NFV resource management, along with an iterative algorithm and heuristics that outperform existing solutions in cost efficiency.

Findings

Achieves 34% average cost reduction compared to existing methods.

Provides a polynomial complexity heuristic with 13.66% average optimality gap.

Validates the framework through extensive simulations under various network scenarios.

Abstract

To fend off network ossification and support high degrees of flexibility and various services, network virtualization and slicing are introduced for the next-generation wireless cellular networks. These two technologies allow diversifying attributes of the future inter-networking and time-varying workloads based resource management paradigms. In this paper, we propose an end-to-end (E2E) resource allocation framework for future networks considering radio and core network by leveraging network function virtualization (NFV). We aim to minimize the network cost defined based on the utilized energy and spectrum while providing E2E quality of service (QoS) for diverse services with stringent QoS requirements. This goal is realized by formulating a novel optimization problem which performs the power and spectrum allocation in radio, and service function chaining and scheduling in the NFV environment while guaranteeing the distinct QoS constraints of the requested services. The proposed optimization problem is mixed-integer non-linear programming, which is a non-convex and NP-hard problem. To solve it, we adopt an iterative algorithm with novel admission control and a greedy-based heuristic algorithm, which is shown to have a polynomial order of complexity with 13:66% global optimality gap on average for a small scaled network. To validate the proposed framework, simulation results are carried out by considering different values of the network parameters and topologies. Moreover, our proposed framework and solution algorithm are assessed and compared with the existing works. Simulation results demonstrate that the proposed heuristic algorithm and framework outperforms the existing ones by 34% on average in cost reduction.