Providing resilient and sustained service is quite challenging in the Service Based Architecture of distributed 5G Core (5GC) as multiple Network Functions (NFs) are involved to help serve the various User Service Requests (USRs) arriving in the control plane. In this regard, the continuous monitoring of individual NFs in a Closed Loop Automation (CLA) is a need of hour to keep up the robust and resilient functioning of the 5GC overall. Any unforeseen situations like the sudden failure, overload, or congestion of the NFs of the 5GC can drop the critical USRs unnecessarily. This paper proposes the proactive monitoring of the NFs of the 5GC in the control plane and utilizes it to intelligently schedule and serve the frequently arriving USRs and prioritize the critical slice service requests. Specifically, the Ford-Fulkerson algorithm popularly known as the Max-Flow problem solver is leveraged to proactively assess the NFs' performance and availability and use it effectively to serve critical service requests arriving during unexpected situations of failure and overloads. Our experiments based on the 3GPP-compliant 5G testbed show that, with the proposed solution, the native 5GC can serve 20% more predominant USRs, and the slice-supportive 5GC can serve 33% more massive Machine Type Communications (mMTC) slice USRs, and 47% more ultra Reliable Low Latency Communications (uRLLC) slice USRs while handling their respective peak traffic.

With multiple Network Functions (NFs) participating, the current Service Based Architecture (SBA) of 5G Core (5GC) is crafted stateless, scalable, robust, and reliable to cater to various service requirements. While Beyond 5G (BSG) and 6G networks aim to achieve higher resilience, studying the Consistency (C), Availability (A), and Partition (P) tolerance (CAP) theorem becomes a critical need for SBA-based SGC to evaluate its impact. In this paper, we evaluate the CAP theorem on the 3GPP-compliant non-roaming stateless SGC and show the trade-off between Consistency and Partition tolerance (CP) mode and Availability and Partition tolerance (AP) mode of functioning. Experiments prove that the SGC in CP mode serves more PDU session requests while remaining less available as compared to the SGC in AP mode.

The continuously growing demand for availing services on 5G has resulted in a massive amount of user data being stored and tracked at the 5G Core (5GC). Also, the amount of traffic arriving in the control plane demands a scalable yet highly available design of the datastore in the 5GC. This paper introduces a novel approach addressing these challenges by integrating the database sharding and replica set techniques into the Unified Data Management (UDM) and Unstructured Data Storage Function (UDSF) of 5GC. Sharding is employed to efficiently distribute and manage extensive user data, ensuring not only scalability but also resilience, and High Availability (HA). Specifically, we propose range-based and slice-specific sharding techniques, each tailored to meet specific service and slice requirements. Through a comparative analysis of these sharding methods on the 3GPP-compliant 5G testbed, the paper illuminates their distinct advantages and limitations. While the range-based sharding shows a significant improvement in UE registration completion time of approximately 29.13%, slicespecific sharding demonstrates an even higher average benefit of about 31.46%. We hope that the insights presented in this paper are quite promising in enhancing 5G infrastructure to facilitate a variety of 5G services and elevated end-user experience.

Being scalable, distributed by nature, and stateless by design, the Service Based Architecture (SBA) of 5G Core (5GC) has become the popular approach today, to provide reliable services to 5G users. However, the well-known Consistency (C), Availability (A), and Partition (P) tolerance (CAP) theorem places a limit on 5GC, as it typically does for any distributed system to choose between consistency and availability, whenever a partition occurs in the system. Hence, choosing between ensuring consistency for users’ state information and the availability of the service while being resilient against unforeseen sudden failures becomes a critical need for Beyond 5G (B5G) and 6G networks. In this paper, we present RECAP 5GC: a REsilience and CAP (RECAP)-aware 5GC that keeps up the High Availability (HA) service by being resilient and consistent during sudden failures of stateless functioning. To prove the feasibility and effectiveness of the proposed solution, we implement it entirely on the 3GPP-compliant control plane of the 5GC testbed between the Access and mobility Management Function (AMF) and Unstructured Data Storage Function (UDSF). Experiments on the minimal two nodes active/active cluster configuration prove that the proposed RECAP 5GC is not only resilient by dropping 47% fewer user requests than the CP mode and 86% fewer user requests than the AP mode but also 1.6% additionally available than the CP mode while remaining consistent on the user’s state and information.

Current advancements in the design of the service-based architecture of 5G Core (5GC) are flexible to be deployed with Network Function Virtualization (NFV) in the form of network slices. While some slices tolerate the delay and some do not, this versatility of slice services imposes challenges to Mobile Network Operators in designing a robust, scalable, and yet resilient 5GC. Therefore, in this work, we build a self-resilient 5GC that not only addresses the scheduling requirements per slice type but also yields service satisfaction. Precisely, we build it as a combination of large Integer Linear Programming based Column Generation (CG) and Artificial Intelligence based Deep Learning in a Closed Loop Automation with a Self Organizing Network paradigm. During unexpected situations of overload and failures of serving instances, the proposed CG technique in the work improves the scheduling and serving of control plane user requests by dropping 15% lesser user requests in delay-tolerant slices and 5% lesser in delay-sensitive slices, than the existing state-of-the-art solutions. With 95% learning accuracy, the Deep Learning-based Long Short Term Memory (LSTM) method predicts the availability of serving instances of the Network Functions (NFs) of the 5GC. Overall, by deriving the optimal number of serving instances for each NF of the 5GC, this blend of CG and AI in the 5GC not only contributes to efficient scaling and reconfiguration of the slices but also improves the resilience and High Availability of the 5GC. We believe that this proposed work is a highly promising solution for building Cognitive Autonomous Networks (CAN) of core side slices for upcoming Beyond 5G (B5G) and 6G in both cloud and edge deployments.

The control plane of 5G Core (5GC) is typically shared among multiple dependent network slices of the data plane. But as the number of dependent slices and services on the common and shared control plane increases, its resilience threat also increases. This paper proposes FlexCore: a 5GC that is not only flexible and scalable but also resilient to cater to various service requirements on both stateful and stateless architectures of 5GC. FlexCore is built with an eXpress Data Path (XDP) and extended Berkeley Packet Filter (eBPF) based SCTP load balancer hooked at the entry point of the 3GPP compliant 5GC control plane, and a set of micro-AMF instances to serve the user requests. Precisely, the FlexCore is fabricated to honor the variety of incoming user requests on the control plane as per the service requirements, like, per slice, per user, or per control procedure of users too. Experiments on a 3GPP compliant 5G testbed show that FlexCore can provide average latency reduction of up to 14% and 79% on stateful and stateless architectures, respectively, and up to 63% latency reduction for latency-critical slices on the slice-aware architecture.

As network slicing is the chief enabler for future Beyond 5G(B5G) and 6G networks, multiple tenants interoperate cost-effectively to provide a variety of slice services on a common physical infrastructure. However, this opens the doors to cross-slice disruptions with Man-in-the-Middle (MITM) attack which ultimately disrupts the slice services in the data plane. In this paper, we address such possible cross-network slice disruptions in a zero-trust and multi-tenant based 5G network by proposing different design techniques namely, secure communication and Artificial Intelligence (AI)-based anomaly detection to prevent them. Our experiments on a 5G testbed prototype show that in the secure communication method, Attribute-Based Encryption (ABE) provides higher security benefits in confidentiality and implicit authorization. However, symmetric encryption and integrity protection prevent cross-slice disruptions with less communication overhead, but with a weaker security level. On the other hand, with online learning and noise tolerance capabilities, AI-based Hierarchical Temporal Memory (HTM) can proactively detect the occurrences of the identified cross-slice disruptions.

The presence of a Load Balancer (LB)s is much significant to keep up the High Availability (HA) and resilience of the scalable 5G Core (5GC). The whole system may collapse just because of inefficient LB at any NF, resulting in total disruption to the High Availability (HA) service. In this paper, we present the LOCOMOTIVE 5GC which outperforms the traditional hot standby in both HA and resilience during various dynamic conditions. LOCOMOTIVE serves 16% (at least) more user requests compared to hot standby in the control plane while handling unexpected overloaded conditions (without the failure of LB). During the failures of LB, it drops 22% lesser user requests than hot standby. With this outstanding resilience, LOCOMOTIVE even achieves 4% better availability than the hot standby in an active-active cluster configuration. To prove the feasibility of LOCOMOTIVE and to encourage further research works in the world of LBs, we developed its entire framework in a 3GPP compliant 5G test-bed system along with eXpress Data Path (XDP) and extended Berkeley Packet Filter (eBPF) framework.

Advancements in 5G and edge computing infrastructure increase the need to deploy location-based services for mission-critical applications like Vehicle to Everything (V2X) and Intelligent Transport System (ITS). In this demonstration, we showcase the location service capabilities of our 5G Core (5GC), coupled with Multi-access Edge Computing (MEC) for delay-sensitive ultra-Reliable Low Latency Communication (uRLLC) service types like V2X and ITS. We believe that this work will guide Mobile Network Operators in building a location assistance service system for emergencies and delay-critical applications.

Mobile Network Operator (MNO)s avail the benefit of providing an isolated service with the network slice in a variety of forms like enhanced Mobile Broadband (eMBB), ultra Reliable Low Latency Communications (uRLLC), and massive Machine Type Communications (mMTC). However, they face challenges in handling situations of overload, congestion, and scaling with the arrival of unexpected control plane User Service Requests (USRs) leading to the dropping of USRs, and hence disturbing the slice’s High Availability (HA). This paper presents a novel High Availability supportive self Reliant NEtwork Slicing System (HARNESS) for 5G Core, powered by intelligent and autonomous Self Organizing Network (SON) paradigm. In HARNESS, we propose algorithms to intelligently schedule and serve the significant portion of control plane USRs for both delay tolerant and delay sensitive slices, to ensure their uninterrupted HA service provision. Along with efficient resource utilization, the HARNESS shows improvements over the traditional HA provisioning methods, in preventing the occluding of crucial control plane USRs by 60% and reducing the average response time for these USRs by 50%. We developed the HARNESS framework in a 5G test-bed system using eXpress Data Path (XDP) and extended Berkeley Packet Filter (eBPF) mechanism for coherent and optimized end-to-end working of it. With this additional gear, HARNESS productively contributes to achieving 3.2% better slice service HA in the minimal two nodes active/active cluster configuration.

5G network supports end-to-end logically isolated networks in the form of network slices, catering to the needs of users of various primary network services, namely enhanced Mobile Broadband (eMBB), ultra Reliable Low Latency Communications (uRLLC), and massive Machine Type Communication (mMTC). Mobile Virtual Network Operators (MVNO)s often face challenges in achieving strong slice isolation and High Availability per slice during overload and scaling situations as the 5G network uses a shared environment for slices with multiple domains, especially considering a variety of services and devices. In this paper, we propose a novel Self Optimizing Network Slicing framework (SONS) leveraging Self Organizing Network by building it as an autonomous slice system in 5G network slicing management for efficient slice sharing and isolation. Precisely, we formulate a system model with Probabilistic Graphical Model (PGM) based Markov Network, building it as an Artificial Intelligence based learning framework. We propose Slice Belief Propagation based algorithms and Deep Learning based Long Short Term Memory (LSTM) methods to aid in serving user requests and reconfiguration of self optimizing slice. Our experiments on the proposed SONS framework shows improvement in serving higher number of users with uninterrupted connectivity by 80% in eMBB, 35% in uRLLC, and 52% in mMTC when compared to standard slice deployments, while handling the worst case of peak traffic in the control plane of 5G Core network.

Network slicing is one of the core features of the 5G network to meet the requirements of various network services, namely, enhanced Mobile Broadband (eMBB), ultra Reliable Low Latency Communications (uRLLC), and massive Machine Type Communication (mMTC) by building an isolated virtual network of resources typically in a Network Function Virtualization (NFV) environment. We have built a 3GPP compliant 5G Core (5GC) testbed and enabled network slicing on it to provide eMBB, uRLLC, and mMTC services to end users. In this demonstration, we showcase the capabilities of our zero-touch framework of 5GC network slicing management and orchestration in an NFV environment with the network functions of 5GC and the applications on eMBB and uRLLC slice services. We believe that this work will guide Mobile Network Operator (MNO)s in building zero-touch autonomous network slices in 5GC along with their smart management and orchestration in closed loop automation.

In today's moving world to ambitious 5G, 3GPP has defined three fundamental and promising services, namely enhanced Mobile Broadband (eMBB), ultra Reliable Low Latency Communication (uRLLC), and massive Machine Type Communication (mMTC) by tuning user's needs on these services to network slicing. While the Network Function Virtualization (NFV) and Software Defined Networking (SDN) are used to enable the network slicing in the mobile network, an effective end-to-end slice management in 5G system is still a challenge to improve the network performance in terms of throughput, latency, and connectivity for each of these envisioned services. In this paper, we focus on the end-to-end network slice life cycle management of network slices on different sites using a single management and orchestration entity with a coherent proof of concept. We propose algorithms for efficiently activating, deactivating, and decommissioning the network slices, using real time status information of network slices from Network Slice Management Function (NSMF). Our results show that by adopting better strategy in these algorithms and considering learned user traffic from the past, in controlling various phases of slice life cycle, we can reduce the response time for a user request by 50%.