978-1-7281-6264-5/20/$31.00 ©2020 IEEE Neuro-Fuzzy Based Handover Authentication Protocol for Ultra Dense 5G Networks Vincent Omollo Nyangaresi Faculty of Biological & Physical Sciences Tom Mboya University College Homabay, Kenya. vnyangaresi@tmuc.ac.ke Anthony Joachim Rodrigues School of Informatics & Innovative Systems JOOUST Kisumu, Kenya tonyr@jooust.ac.ke Silvance Onyango Abeka School of Informatics & Innovative Systems JOOUST Kisumu, Kenya silvancea@gmail.com Abstract— Smart grids improve the management of various power network components as well as the monitoring of power flows. Their deployments require very fast communication channels, which can be facilitated by 5G networks. However, 5G’s ultra-densification boost system capacity and spectrum efficiency but introduce vulnerabilities due to the incorporation of many wireless network technologies and devices. The conventional 5G handover authentication protocols have security challenges that expose them to attacks such as message replay. As such, many authentication protocols have been proposed to secure smart grid data transmission. However, these protocols are unreliable in attacks detection and prevention, energy inefficient, have high latencies, heavy communication as well as computational overheads. In this paper, a 5G key management and handover protocol is developed to address some of these security and performance issues. The simulation results show that the proposed protocol preserves perfect forward key secrecy, is robust against de-synchronization, replay, man-in-the-middle (MITM), denial of services (DoS) and jamming attacks. Performance analysis showed that it has low communication overheads, space complexity and handover authentication latencies. Compared with 3GPP R16, the proposed protocol showed a 25% and 42.9% improvement in communication overheads and space complexity respectively. Keywords—Smart grid, handover authentication, latency, security I. INTRODUCTION The incorporation of new technologies such as software defined network (SDN), network function virtualization (NFV) and network slicing, together with network densification introduce many security issues in 5G networks. The small sized cells imply frequent handovers [1], making efficient mobility management [2], seamless and secure handovers a necessity [3]. In addition, the integration of new wireless components such as hotspots that are vulnerable to attacks renders the entire 5G network susceptible. As pointed out in [2] cloud computing and SDN increase risks of distributed denial of services (DDoS) attacks. As such, compared with 2G, 3G and long term evolution (LTE), security requirements in 5G networks are higher [4]. In 5G networks, 3rd Generation Partnership Project (3GPP) committee has specified three types of handovers, which include intra new radio (NR), inter-3GPP and that between 3GPP and untrusted non-3GPP [5]. For handover authentication, 5G’s improved Authentication and Key Agreement (5G-AKA’) is employed. Unfortunately, 5G-AKA’ protocol has many security flaws rendering it susceptible to attacks such as jamming, de-synchronization, replay attacks and lack of perfect forward secrecy[6][7]. In spite of this, it is revealed in [1] that security and privacy issues in 5G have not received much attention. As such, authors in [8] point out security and privacy issues such as authentication, encryption, key and identity management, secure storage and mobility are open challenges in 5G. To address some of these challenges, new handover authentication schemes based on techniques such as chameleon hashing [9], pairing-based cryptography [10], pairing-free identity cryptographic methods [11], and anonymous handover authentication with conditional privacy preservation [12] have been proposed. However, these schemes have their own architectures that are different from 5G or LTE network [13], making their implementation in these networks uneconomic and unfeasible due to the required changes to the underlying network. In addition, the bilinear pairing operations in [10] and [11] are computationally intensive, hence cannot satisfy 5G networks’ stringent delay requirements [2]. A research by [4] pointed out that a faster, safer and effective key management and handover protocol that ensures privacy protection and validation is crucial for 5G advancements. This calls for the development of efficient protocols to ensure faster end devices authentication in 5G networks [2]. As such, in this paper, an Artificial Neuro-Network-Fuzzy Logic (ANN-FL) based handover authentication protocol for ultra dense 5G networks is proposed. Our contributions include the following: I. We develop a multi-criteria handover protocol that is shown to reduce the number of executed handovers in 5G networks hence a reduction in associated attacks. II. The proposed protocol uses ANN-FL technique to optimize the selection of the target cell prior to the actual handover, reducing handover delays. III. We introduce the shared keys ψ and ƺ to secure the link between source 5G NR NodeB (gNB) and target gNB, and the link between Access and Mobility management Function (AMF) and User Equipment (UE) respectively. IV. We encapsulate and encrypt the session keys used during handover to thwart eavesdropping and other associated attacks. The rest of this paper is organized as follows: Section II discusses related work while Section III outlines the system model of our protocol. Section IV presents results and 2020 2nd Global Power, Energy and Communication Conference (IEEE GPECOM2020), October 20-23, 2020, Online Conference 339 Authorized licensed use limited to: Jaramogi Oginga Odinga University of Science & Technology. Downloaded on October 15,2021 at 13:16:32 UTC from IEEE Xplore. Restrictions apply.