In this paper, we make a case for using lightweight containers fine-grained resource flexing Virtualized Network Functions (VNFs) to meet the demands of varying workloads. We quantitatively compare VNF performance and infrastructure usage three instantiations (bare metal, virtual machine, container) selected VNFs. The VNFs experiment with are Mobility Management Entity (MME) Evolved packet core (EPC) architecture cellular networks, Suricata multi-threaded Intrusion Detection System (IDS),...

Network Functions Virtualization (NFV) has enabled operators to dynamically place and allocate resources for network services match workload requirements. However, unbounded end-to-end (e2e) latency of Service Function Chains (SFCs) resulting from distributed Virtualized (VNF) deployments can severely degrade performance. In particular, SFC instantiations with inter-data center links incur high e2e latencies Level Agreement (SLA) violations. These trigger timeouts protocol errors...

Network Functions Virtualization (NFV) has enabled operators to dynamically place and allocate resources for network services match workload requirements. However, unbounded end-to-end (e2e) latency of Service Function Chains (SFCs) resulting from distributed Virtualized (VNF) deployments can severely degrade performance. In particular, SFC instantiations with inter-data center links incur high e2e latencies Level Agreement (SLA) violations. These trigger timeouts protocol errors...

The adoption of a cloud-native architecture in 5G networks has facilitated rapid deployment and update cellular services. An important part this is the implementation network functions statelessly. However, statelessness its associated serialization de-serialization data database interaction significantly increase latency. In work, we take first steps towards quantifying cost system. We compare different state management paradigms, propose number optimizations to reduce cost. Our preliminary...

Cellular service carriers often employ reactive strategies to assist customers who experience non-outage related individual degradation issues (e.g., performance degradations that do not impact at scale and are likely caused by network provisioning for devices). Customers need contact customer care request assistance before these resolved. This paper presents our with PACE (<u>P</u>ro<u>A</u>ctive <u>C</u>ar<u>E</u>), a novel, proactive system monitors, troubleshoots resolves issues, without...

Network slicing enables communication service providers to partition physical infrastructure into logically independent networks. slices must be provisioned meet the service-level objectives (SLOs) of disparate offerings, such as enhanced mobile broadband, ultrareliable low-latency communications, and massive machine-type communications. orchestrators customize placement scaling achieve SLO each network slice. In this article, we discuss challenges encountered by in allocating resources 5G...

Caller-ID spoofing deceives the callee into believing a call is originating from another user. Spoofing has been strategically used in now-pervasive telephone fraud, causing substantial monetary loss and sensitive data leakage. Unfortunately, caller-ID feasible even when user authentication place. State-of-the-art solutions either exhibit high overhead or require extensive upgrades, thus are unlikely to be deployed near future. In this paper, we seek an effective efficient solution for 4G...

As the primary channel for users to report and resolve service issues, customer care has historically been a critical resource-intensive aspect of operating cellular networks. However, owing inherent complexity in correlating network events with performance experienced by individual users, adoption data-driven solutions leveraging data realtime troubleshooting during calls remained challenge providers (CSPs). In this work, we propose novel ML aSsisted cEllular sErvice diagNosis (SEEN)...

We propose CoRE, a 360° video streaming approach that reduces bandwidth requirements compared to transferring the entire video. CoRE uses non-linear sampling in both spatial and temporal domains achieve robustness view direction prediction error transient wireless network fluctuation. Each frame samples environment all directions, with full resolution over predicted field of gradually decreasing at periphery, so missing pixels are avoided, irrespective magnitude. A chunk has main part rate,...

We propose an optimization to reduce the latency incurred by a stateless 5G control plane. The key idea is avoid redundant database read operations. achieve this reading user's state only once and sending it successive network functions in chain. Experimental results show that can total cost 33% on average.

We consider the task of learning a parametric Continuous Time Markov Chain (CTMC) sequence model without examples sequences, where training data consists entirely aggregate steady-state statistics. Making problem harder, we assume that states wish to predict are unobserved in data. Specifically, given over transition rates CTMC and some known rates, extrapolate its steady state distribution unobserved. A technical roadblock learn from has been chain rule compute gradients will not work...

Microservices enable rapid service deployment and scaling. Integrating poorly-understood microservice components into Service Function Chains (SFCs) or graphs limits a provider's control over delivery latency, however. Orchestration frameworks currently instantiate place myriads of without knowing the impact placement decisions on latency.

In this paper, we investigate the impact of debt takeover distribution utilities along with structural reforms to enhance their efficiency, in context a scheme by Indian government improve financial and operation efficiency utilities. The is called Ujwal DISCOM Assurance Yojna (UDAY). study primarily relies on descriptive analysis Difference methodology understand various stakeholders, such as: DISCOMs, generation companies, banks state governments. We found that UDAY DISCOMs companies...

CoRE is an approach for streaming 360° videos based on a non-linear sampling of the equirectangular video cube. robust to view prediction errors.

We consider the task of learning a parametric Continuous Time Markov Chain (CTMC) sequence model without examples sequences, where training data consists entirely aggregate steady-state statistics. Making problem harder, we assume that states wish to predict are unobserved in data. Specifically, given over transition rates CTMC and some known rates, extrapolate its steady state distribution unobserved. A technical roadblock learn from has been chain rule compute gradients will not work...

One of the key challenges faced by cellular network customer care agents is identifying if service problem caused network-related issues or user-device-related issues. Some providers [5], [6], therefore, employ machine learning-based troubleshooting frameworks to aid in root cause problems experienced users. However, obtaining large-scale and comprehensive ground truth result data costly requires tremendous manual efforts from networking operators. Due this limitation, training such a...

In this work, we examine the challenges that service providers encounter in managing complex function graphs, while controlling delivery latency. Based on lessons learn, outline design of a new system, Invenio, empowers to effectively place microservices without prior knowledge functionality. Invenio correlates user actions with messages they trigger seen network traces, and computes procedural affinity for communication among each action. The values can then be used make placement decisions...