[Paper Review] Energy Cost Minimization by Joint Radio and NFV Resource Allocation: E2E QoS Framework.

Abolfazl Zakeri, Narges Gholipoor+3 more|arXiv (Cornell University)|Jul 14, 2019

Software-Defined Networks and 5G44 references4 citations

TL;DR

This paper proposes an end-to-end joint radio and NFV resource allocation framework to minimize energy and server costs while meeting diverse QoS requirements. By using an alternative search method and a novel heuristic algorithm with admission control, the approach reduces network cost by approximately 8% compared to greedy methods.

ABSTRACT

In this paper, we propose an end to end joint radio and network function virtualization (NFV) resource allocation for next generation networks providing different types of services with different requirements in terms of latency and data rate. We consider both the access and core parts of the network, and formulate a novel optimization problem whose aim is to perform the radio resource allocation jointly with virtual network function (VNF) embedding, scheduling, and resource allocation such that the network cost, defined as the consumed energy and the number of utilized network servers, is minimized. The proposed optimization problem is non-convex, NP-hard, and mathematically intractable, and hence, we adopt the alternative search method (ASM) to decouple the main problem into some sub-problems of lower complexity. Moreover, w propose a novel heuristic algorithm for embedding and scheduling of VNFs by proposing a novel admission control (AC) algorithm. Then, we compare the performance of the proposed algorithm with a greedy-based solution in terms of the acceptance ratio and the number of active servers. Our simulation results show that the proposed heuristic algorithm outperforms the conventional ones by approximately 8%.

Motivation & Objective

• Address the challenge of minimizing network energy and server utilization costs in next-generation heterogeneous networks.
• Jointly optimize radio resource allocation, VNF embedding, scheduling, and resource allocation across access and core networks.
• Ensure end-to-end QoS for diverse services with varying latency and data rate requirements.
• Develop a scalable solution for NP-hard, non-convex optimization problems in integrated radio-NFV resource management.

Proposed method

• Formulate a non-convex, NP-hard optimization problem that jointly minimizes energy consumption and number of active servers.
• Apply the alternative search method (ASM) to decompose the main problem into lower-complexity sub-problems.
• Design a novel heuristic algorithm for VNF embedding and scheduling with an integrated admission control (AC) mechanism.
• Decouple radio and NFV resource allocation through iterative optimization using ASM to improve tractability.
• Use simulation-based evaluation to compare the proposed heuristic with a greedy baseline in terms of acceptance ratio and active server count.

Experimental results

Research questions

• RQ1 How can joint radio and NFV resource allocation minimize total network cost while satisfying diverse QoS requirements?
• RQ2 What is the impact of a novel admission control mechanism on VNF scheduling and resource utilization?
• RQ3 To what extent does the proposed heuristic outperform greedy-based approaches in terms of cost and acceptance ratio?
• RQ4 How effective is the alternative search method in solving the non-convex, NP-hard optimization problem?

Key findings

• The proposed heuristic algorithm achieves approximately 8% better performance in cost reduction compared to the greedy-based solution.
• The admission control mechanism improves acceptance ratio by effectively managing VNF requests based on resource availability and QoS constraints.
• The joint optimization framework successfully reduces the number of active servers while maintaining required service quality.
• Simulation results confirm that the ASM-based decomposition enables feasible and scalable solutions to the otherwise intractable optimization problem.