FIGURE 7 - uploaded by Manuel Herrera
Content may be subject to copyright.
xIC network: GFT-TS (on top) and its associated matrix profile (at bottom). Top 5 discords marked in red. Top 5 motifs marked in black.

xIC network: GFT-TS (on top) and its associated matrix profile (at bottom). Top 5 discords marked in red. Top 5 motifs marked in black.

Source publication
Full-text available
This paper proposes a framework to analyse traffic-data processes on a long-haul backbone infrastructure network providing internet services at a national level. This type of network requires low latency and fast speed, which means there is a large demand for research focusing on near real-time decision-making and resilience assessment. To this aim...

Contexts in source publication

Context 1
... bottom block of Table 4 summarises the results found from the analysis shown in Fig. 8. The analysis for the xOC network repeats the process done for the xIC network. For the xOC network, the top discords mostly occur near midnight, approximately between 11pm and 12:30am. Similar to the above explanation for xIC, a possible explanation for the discords in xOC seems to be the process of disconnection from internet by the ...
Context 2
... bottom block of Table 4 summarises the results found from the analysis that Fig. 8 shows. The explanation for these results is similar to the one given for the xIC network case. For the xOC network, the top discords occur around midnight. An explanation to that may lie in the many users disconnecting from the internet at around this time. The nodes with more influence on these discord events are also identified for 3 ...


... Graphs are structures that can model a set of objects (vertices) along with their relationships/connections (edges). The use of such structures has been previously adapted for dealing with engineering problems, such as anomaly detection in internet traffic networks (Herrera et al., 2021). Other researchers use machine learning on graphs due to the ability of graphs to represent and analyze data with graph neural network (GNN) models (Wu et al., 2021). ...
Sustainable management of water resources is a key challenge for the well-being and security of current and future society worldwide. In this regard, water utilities have to ensure fresh water for all users in a demand scenario stressed by climate change along with the increase in the size of cities. Dealing with anomalies, such as leakages and pipe bursts, represents one of the major issues for efficient water distribution system (WDS) operation and management. To this end, it is crucial to count on suitable methods and technologies to provide a quick, reliable, and accurate detection of such anomalies and supply disruption events. Therefore, this work proposes a novel WDS management framework based on the development of graph convolutional neural networks (GCN) models for bursts detection in WDSs. These methods rely on a WDS graph representation for a set of pressure and flow rates measures. Such a graph is used to design two GCN-based models to identify bursts. In addition, two conventional multi-layer perceptron models are used as the benchmarks to compare the graph-based methodologies. Finally, the proposed methodology is tested on a water utility network, showing the high potential of graph convolutional networks for anomaly detection on WDSs.
... This enables the analysis to consider the overall evolution of the sensor readings as well as to deal with multi scale data-analysis. Particularly, the spectra time-series has been successfully analysed by using the matrix profile method for computing distances based on an intensive computation of sub-sequences distances (Zhu et al., 2020;Herrera et al., 2021). ...
Conference Paper
Infrastructure systems in today's increasingly interconnected world employ the capabilities of the Internet of Things (IoT) technologies for their monitoring, operational control, and asset management. IoT devices can be defined as sensors (of different types) collecting, processing, and sharing time series of data. The analysis of such data often face challenges as a consequence of the high frequency of data collection and the increasing number of sensors placed on infrastructure. Power related issues, timestamp misalignment, and heterogeneous sampling designs are among the most common issues that the IoT data collection may suffer alongside the inherent complexities of large scale databases. This paper provides an overview of time series mining techniques adapted to tackle such issues in IoT data. The aim is to have a pattern recognition tool-set for developing anomaly detection algorithms. Particularly, the paper investigates how to efficiently handle large-scale time series coming from multiple sensors in a stream and following an unevenly spaced-irregular-sampling. The analysis is demonstrated through a case study of time series data mining of sensors installed for supporting the predictive maintenance of quay-cranes at the Port of Felixstowe, the largest container port in Britain.
... In order to perform frequency analysis of graph signal, graph Fourier transform (GFT) is proposed and quickly becomes the focus of attention. At GFT have involved signal denoising [4,5], digital watermarking [6][7][8], compressed sensing [9], network traffic analysis [10], graph based clustering [11], and so on. ...
Full-text available
Graph Fourier transform (GFT) is an important theoretical tool in spectral analysis of graph signal. This paper focuses on Laplacian-based GFT on two special cases of graph data. The relationship between GFT and discrete cosine transform (DCT) is revealed and proved formally. For 1D signal, we prove that GFT is unique and is equivalent to DCT. For 2D image, GFT has more than one basis, one of which is the DCT basis. The work in this paper would help reduce the computational complexity of GFT in special cases and contribute to a deeper understanding of GFT.