Robust Network Compressive Sensing

This book investigates compressive sensing techniques to provide a robust and general framework for network data analytics. It demonstrates the feasibility of compressive sensing in network analytics, discusses the challenges in applying it to real-world data, and proposes a robust compressive sensing technique to address the problem of noise, errors, and anomalies. It also proposes a data-driven synchronization algorithm to address the problem of lack of synchronization. The combination of the two techniques can reduce the ranks.

Format: Paperback / softback
Length: 90 pages
Publication date: 23 October 2022
Publisher: Springer International Publishing AG

This book delves into the realm of compressive sensing techniques, offering a robust and versatile framework for network data analytics. Its primary objective is to introduce a comprehensive compressive sensing framework that encompasses tasks such as missing data interpolation, anomaly detection, data segmentation, and activity recognition. By showcasing the benefits of this approach, the book aims to provide a valuable tool for analyzing network data.

Chapter 1 provides an introductory overview of compressive sensing, encompassing its definition, limitations, and the diverse range of network analysis applications it supports. It sheds light on the fundamental principles that underpin compressive sensing and its ability to efficiently handle large datasets.

Chapter 2 delves into the practicality of compressive sensing in network analytics. The authors employ it to detect anomalies in a customer care call dataset obtained from a Tier 1 ISP in the United States. They employ a regression-based model to establish a relationship between calls and events. Through their demonstration, the authors highlight the effectiveness of compressive sensing in identifying significant factors and leveraging the low-rank structure and temporal stability to enhance detection accuracy.

Chapter 3 delves into the challenges associated with applying compressive sensing to real-world data. Understanding the underlying reasons behind these challenges is crucial for developing effective methods to mitigate their impact. The authors analyze a wide range of real-world traces, shedding light on various factors that contribute to the violation of the low-rank property in practical scenarios. They particularly emphasize the role of noise, errors, and anomalies in compromising the low-rank nature of matrices.

In Chapter 4, the authors propose a robust compressive sensing technique specifically designed to address the issue of noise, errors, and anomalies in real-world data. By decomposing real-world data into a low-rank matrix, the technique explicitly accounts for anomalies, ensuring a more accurate representation of the underlying data structure.

Overall, this book serves as a valuable resource for researchers and practitioners in the field of network data analytics. It provides a comprehensive introduction to compressive sensing techniques, showcases their practical applications, and addresses the challenges associated with real-world data analysis. By leveraging the power of compressive sensing, the book enables efficient and effective processing of large datasets, leading to insightful discoveries and informed decision-making in network-related domains.

Weight: 168g
Dimension: 235 x 155 (mm)
ISBN-13: 9783031168284
Edition number: 1st ed. 2022