A Survey of Machine Learning for Network Fault Management

Nouioua, Mourad; Fournier‐Viger, Philippe; He, Ganghuan; Nouioua, Farid; Zhou, Min

doi:10.1007/978-3-030-66288-2_1

Cited by 7 publications

(6 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, an episode rule that may be found in the alarm log from a telecommunication network could indicate that it is common for an alarm

A

to be triggered in a machine after an alarm

B

occurs in another machine. Such rules provide information that can help understand the relationships between alarms and can be used to improve network maintenance by focusing on the most important alarms (Nouioua et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

A survey of episode mining

Ouarem,

Nouioua,

Fournier‐Viger

2023

WIREs Data Min & Knowl

View full text Add to dashboard Cite

Episode mining is a research area in data mining, where the aim is to discover interesting episodes, that is, subsequences of events, in an event sequence. The most popular episode‐mining task is frequent episode mining (FEM), which consists of identifying episodes that appear frequently in an event sequence, but this task has also been extended in various ways. It was shown that episode mining can reveal insightful patterns for numerous applications such as web stream analysis, network fault management, and cybersecurity, and that episodes can be useful for prediction. Episode mining is an active research area, and there have been numerous advances in the field over the last 25 years. However, due to the rapid evolution of the pattern mining field, there is no prior study that summarizes and gives a detailed overview of this field. The contribution of this article is to fill this gap by presenting an up‐to‐date survey that provides an introduction to episode mining and an overview of recent developments and research opportunities. This advanced review first gives an introduction to the field of episode mining and the first algorithms. Then, the main concepts used in these algorithms are explained. After that, several recent studies are reviewed that have addressed some limitations of these algorithms and proposed novel solutions to overcome them. Finally, the paper lists some possible extensions of the existing frameworks to mine more meaningful patterns and presents some possible orientations for future work that may contribute to the evolution of the episode mining field.This article is categorized under: Algorithmic Development > Spatial and Temporal Data Mining Algorithmic Development > Association Rules Technologies > Association Rules

show abstract

“…For example, an episode rule that may be found in the alarm log from a telecommunication network could indicate that it is common for an alarm

A

to be triggered in a machine after an alarm

B

Section: Introductionmentioning

confidence: 99%

A survey of episode mining

Ouarem,

Nouioua,

Fournier‐Viger

2023

WIREs Data Min & Knowl

View full text Add to dashboard Cite

show abstract

“…The predictive autonomous transmission agent (ATA) based on artificial neural networks (ANN) that forecasts the right forward error correction (FEC) algorithm design for limited operation as a final real-time supervising of state of polarization (SOP) traces and in relation to pre-forward error correction and bit error rate [1]. Other authors in [2] conducted a study in exploration of various machine learning (ML) techniques based on decision tree and support vector machine (SVM) for fault management [3], ideally for soft-failure detection, identification, and localization taking the merit of optical spectrum analyzers (OSA) to supervise the optical spectrum. Due to its nature as a distributed system, knowledge utilization is placed at the device or machine level and decision-making process that is located close to the centralized software-defined networking (SDN) controller.…”

Section: Introduction 11 Knowledge Managementmentioning

confidence: 99%

Review of Knowledge Management in Optical Networks, Lambda Architecture using Database Technologies in Cloud Settings

Fofanah¹,

Koroma²,

Fofana³

2021

IJSRP

View full text Add to dashboard Cite

In this paper, we present concepts, theories, and overview of knowledge management in an autonomous optical networks and Lamba Architecture in cloud related environment. This study presents some illustrative cases that has been used to illustrate the potential application of KM architecture and to evaluate the various policies for the knowledge sharing and integration algorithm. Here the knowledge is used at the optical transponder system level, sharing, and integration implemented at the node level and supervising and data analytics (SDA) controller level. The KM process has been evaluated by integrating on a metro-network situation in terms of model error convergence time and the data shared among agents. Indeed, the propagation and reinforcement actions illustrated by similar convergence time than data-based policies at various phases of the network learning process without compromising the convergence accuracy of the model prediction. The Lambda Architecture is the new model for Big Data and database research focus, that helps in data processing with a balance on throughput, latency, and fault-tolerance. To provide a complete solution and better accuracy, low latency, and high throughput, there exists no single tool. This introduced the idea to use a set of tools and methods to build a comprehensive Big Data approach. Although this paper does not provide a developed and working tool, however, provides an outline and the methods used by researchers to overcome some of the shortcomings of Lambda Architecture. The Lambda Architecture defines a set of layers to fit in a set of tools and methods rightly for constructing a comprehensive Big Data scheme: Speed Layer, Serving Layer, Batch Layer. Each layer satisfies a set of features and builds upon the functionality delivered by the layers beneath it. The Batch Layer is the place where the master dataset is warehoused, which is an unchangeable and add-only set of raw data. Also, the batch layer computes before the results using a distributed processing system like Hadoop, Apache Spark that can manage large amounts of data. The Speed Layer encapsulates new data coming in real-time and processes it. The Serving Layer comprises a parallel processing query steam engine, that takes results from both Batch and Speed Layers and responds to questions and requests in real-time with low latency. Stack Overflow is a Question-and-Answer forum with an enormous user community, millions of posts with rapid growth over the years. This paper demonstrates the Lambda Architecture by constructing a data pipeline, to add a new "Recommended Questions" section in the Stack Overflow user profile and update the questions suggested in real-time. Additionally, various indicators such as trending tags, user performance numbers such as are shown in user dashboard by querying through batch processing layer. Finally, this paper provides a seamless search of the various methods or techniques used to help solve complex databases which are provided by Stack Overflow platform infrastructure.

show abstract

“…Some of the works related to preventive fault management are: [1], [2], [3]. In [1], the alarms' proactive management is addressed as a way to take advantage of ML algorithms to follow the evolution of mobile networks and services operations.…”

Section: Introductionmentioning

confidence: 99%

A preventive maintenance approach to optimize fault management using machine learning

Pereira,

Duarte,

Vieira

2021

URSI Radio Sci. Bull.

View full text Add to dashboard Cite

Mobile networks' fault management can take advantage of Machine Learning (ML) algorithms making its maintenance more proactive and preventive. Currently, Network Operations Centers (NOCs) still operate in reactive mode, where the troubleshoot is only done after the problem identifi cation. The network evolution to a preventive maintenance enables the problem prevention or quick resolution, leading to a greater network and services availability, to a better operational effi ciency and, above all, ensures customer satisfaction.In this paper, diff erent algorithms for Sequential Pattern Mining (SPM) and Association Rule Learning (ARL) are explored, using real Fault Management (FM) data from a live Long Term Evolution (LTE) network. A comparative analysis of the performance and effi ciency between all the algorithms was carried out, having observed a decrease of 3.31% in the total number of alarms and 70.45% in the number of alarms corresponding to the same type. There were also considerable reductions in the number of alarms per network node or zone, identifying 39 nodes that no longer had any unresolved alarm.These results demonstrate that the recognition of sequential alarm patterns allows the preventive maintenance of a mobile communications network.

show abstract

A Survey of Machine Learning for Network Fault Management

Cited by 7 publications

References 68 publications

A survey of episode mining

A survey of episode mining

Review of Knowledge Management in Optical Networks, Lambda Architecture using Database Technologies in Cloud Settings

A preventive maintenance approach to optimize fault management using machine learning

Contact Info

Product

Resources

About