Toward a Dynamic QoS Management over LTE/EPC Network using ANFIS

Abstract

The promise of high bandwidth and low latency for real-time applications are still open issues and challenges in Next Generation Networks (NGN). Thus, sophisticated management is required to improve the Quality of Service (QoS) in these networks and to respect the available network resources. In this paper, we propose a new approach of hybrid criticality to manage dynamically multimedia traffic and requests customers through Differentiated Service (Diffserv) policies using Adaptive Neuro-fuzzy Inference System (ANFIS) and implement them to the Policy Control Rule Function (PCRF) for selecting the appropriate Code Point (CP) to users. We extend the proposed approach in IMS networks with PCRF and PCEF interaction. This approach is based on internal and external factors for making-decision. The internal factors depict the QoS parameters and the external factors characterize the user information. This paper presents a new architecture ANFIS-based for Long Term Evolution (LTE) packet core to enhance QoS management. Simulations show that by adopting hybrid criticality using ANFIS, an improved QoS parameter and time response are gained compared to the classical cases where all flows exhibit the same level of criticality and compared also to Fuzzy Logic technique.