Toward Failure-Resilient Service Function Chaining: Novel Formulation and Heuristic Approach.

Modern ICT infrastructures are becoming increasingly complex and tightly interdependent due to rapid technological trends such as virtualization, softwarization, large-scale data proliferation, and cloudification. The massive deployment of Beyond 5G networks and the Internet of Things (IoT) further amplifies the need for highly reliable, fault-tolerant, and automated network management. Ensuring service continuity under large-scale failures and disaster scenarios has therefore become a critical research challenge.

In Beyond 5G core networks, Service Function Chains (SFCs) are commonly protected by deploying redundant instances of Virtual Network Functions (VNFs) across geographically distributed data centers and by provisioning both primary and dedicated backup paths. However, failure-resilient SFC provisioning entails solving a set of tightly coupled optimization problems, including VNF placement, working and backup path computation, and resource allocation with intra- and inter-SFC capacity sharing. Each of these subproblems is NP-hard, and their joint optimization significantly increases computational complexity. As optimization techniques play an increasingly central role in emerging infrastructures such as data center networks and Beyond 5G systems, it is both timely and necessary to investigate scalable and efficient approaches for enhancing network resilience.

This PhD thesis aims to tackle these challenges by designing advanced Integer Linear Programming (ILP) and Mixed Integer Programming (MIP) models, together with efficient heuristic and meta-heuristic algorithms, to optimize resource allocation for resilient SFC provisioning in Beyond 5G core networks.

Research Objectives

Objective 1: Intra-SFC Sparse Capacity Sharing

Develop a novel ILP/MIP formulation for resilient SFC provisioning that leverages concurrent VNF instantiation, multi-path routing, and heterogeneous load distribution across paths.

Design an efficient constrained shortest-path heuristic for the joint VNF placement and SFC routing problem, based on multipartite graph modeling.

Objective 2: Inter-SFC Spare Capacity Sharing

Propose ILP/MIP formulations that enable efficient spare capacity sharing among multiple SFCs while preserving failure isolation and performance guarantees.

Objective 3: Intelligent Restoration of Network Functions

Develop both MIP-based and reinforcement learning–based solutions for network function restoration under large-scale failures.

Compare their performance in terms of scalability, optimality, and response time.

Investigate hybrid approaches that combine optimization-based and learning-based methods to further enhance restoration performance.

  For more information about the PhD subject, please try to refer to its page on ADUM. (it will be available shortly)

This fully funded, 3-year PhD position will be hosted by the LIA (Computer Science) Laboratory at Avignon University, located in the picturesque city of Avignon in the south of France.

The selected PhD candidate will be jointly supervised by two researchers at LIA: one specializing in resilient network optimization and the other in operations research. This interdisciplinary supervision offers a unique opportunity to work at the intersection of network resilience and advanced optimization methods.

The successful applicant will apply for the doctoral contract of ED 536, reserved for the CORNET team of the LIA laboratory. The position comes with a standard doctoral salary of at least €2,300 gross per month (from January 1, 2026).

We encourage applications from second-year master’s students or final-year engineering students with a strong background in mathematics (such as operations research, AI, machine learning, or applied mathematicss) and/or  networking.