A5045967404- Cloud Computing and Resource Management
- Advanced Queuing Theory Analysis
- Software System Performance and Reliability
- IoT and Edge/Fog Computing
- Network Traffic and Congestion Control
- Distributed systems and fault tolerance
- Age of Information Optimization
- Simulation Techniques and Applications
- Anomaly Detection Techniques and Applications
- Distributed and Parallel Computing Systems
- Caching and Content Delivery
- Service-Oriented Architecture and Web Services
- Software Reliability and Analysis Research
- Advanced Software Engineering Methodologies
- Advanced Wireless Network Optimization
- Data Stream Mining Techniques
- Peer-to-Peer Network Technologies
- Network Security and Intrusion Detection
- Parallel Computing and Optimization Techniques
- Transportation Planning and Optimization
- Advanced Data Storage Technologies
- Advanced Statistical Process Monitoring
- Optimization and Search Problems
- Petri Nets in System Modeling
- Scientific Measurement and Uncertainty Evaluation
Imperial College London
2016-2025
Eindhoven University of Technology
2018
Imperial Valley College
2017
Institute e-Austria Timisoara
2016
William & Mary
2007-2014
Chinese University of Hong Kong
2014
Columbia University
2014
Microsoft (United States)
2014
Télécom Paris
2014
IBM (United States)
2014
Efficient anomaly detection and diagnosis in multivariate time-series data is of great importance for modern industrial applications. However, building a system that able to quickly accurately pinpoint anomalous observations challenging problem. This due the lack labels, high volatility demands ultra-low inference times Despite recent developments deep learning approaches detection, only few them can address all these challenges. In this paper, we propose TranAD, transformer network based...
Abstract Recent years have seen the massive migration of enterprise applications to cloud. One challenges posed by cloud is Quality-of-Service (QoS) management, which problem allocating resources application guarantee a service level along dimensions such as performance, availability and reliability. This paper aims at supporting research in this area providing survey state art QoS modeling approaches suitable for systems. We also review classify their early some decision-making problems...
Efficient anomaly detection and diagnosis in multivariate time-series data is of great importance for modern industrial applications. However, building a system that able to quickly accurately pinpoint anomalous observations challenging problem. This due the lack labels, high volatility demands ultra-low inference times Despite recent developments deep learning approaches detection, only few them can address all these challenges. In this paper, we propose TranAD, transformer network based...
We present the Java Modelling Tools (JMT) suite, an integrated framework of tools for performance evaluation computer systems using queueing models. The suite offers a rich user interface that simplifies definition models by means wizard dialogs and graphical design workspace. features JMT span wide range state-of-the-art methodologies including discrete-event simulation, mean value analysis product-form networks, analytical identification bottleneck resources in multiclass environments,...
Cloud computing is emerging as a major trend in the ICT industry. While most of attention research community focused on considering perspective providers, offering mechanisms to support scaling resources and interoperability federation between Clouds, developers operators willing choose without being strictly bound specific solution mostly neglected. We argue that Model-Driven Development can be helpful this context it would allow design software systems cloud-agnostic way supported by model...
Cloud computing is emerging as a major trend in the ICT industry. While most of attention research community focused on considering perspective providers, offering mechanisms to support scaling resources and interoperability federation between Clouds, developers operators willing choose without being strictly bound specific solution mostly neglected. We argue that Model-Driven Development can be helpful this context it would allow design software systems cloud-agnostic way supported by model...
Finding optimal configurations for Stream Processing Systems (SPS) is a challenging problem due to the large number of parameters that can influence their performance and lack analytical models anticipate effect change. To tackle this issue, we consider tuning methods where an experimenter given limited budget experiments needs carefully allocate find configurations. We propose in setting Bayesian Optimization Configuration (BO4CO), auto-tuning algorithm leverages Gaussian Processes (GPs)...
Microservices based architectures are increasingly widespread in the cloud software industry. Still, there is a shortage of auto-scaling methods designed to leverage unique features these architectures, such as ability independently scale subset microservices, well ease monitoring their state and reciprocal calls. We propose address this with ATOM, model-driven autoscaling controller for microservices. ATOM instantiates solves at run-time layered queueing network model application....
With the rise of internet things (IoT) technology, it is anticipated that large-scale sensor-based systems will permeate society, calling for novel methodologies to design, test, and operate these systems. IoT relies on networked interconnected physical devices often featuring computational capabilities. The sheer number plays a key role in revolution. For example, Gartner research predicts connect up 50 100 billion by 2020. It estimated generate ~1.7 trillion US dollars value 2020 with an...
Intelligent task placement and management of tasks in large-scale fog platforms is challenging due to the highly volatile nature modern workload applications sensitive user requirements low energy consumption response time. Container orchestration have emerged alleviate this problem with prior art either using heuristics quickly reach scheduling decisions or AI driven methods like reinforcement learning evolutionary approaches adapt dynamic scenarios. The former often fail environments,...
In recent years, the landscape of computing paradigms has witnessed a gradual yet remarkable shift from monolithic to distributed and decentralized such as Internet Things (IoT), Edge, Fog, Cloud, Serverless. The frontiers these technologies have been boosted by manually encoded algorithms Artificial Intelligence (AI)-driven autonomous systems for optimum reliable management resources. Prior work focuses on improving existing using AI across wide range domains, efficient resource...
Resource allocation in the cloud is usually driven by performance predictions, such as estimates of future incoming load to servers or quality-of-service(QoS) offered applications end users. In this context, characterizing web workload fluctuations an accurate way fundamental understand how provision resources under time-varying traffic intensities. paper, we investigate Markovian Arrival Processes (MAP) and related MAP/MAP/1 queueing model a tool for prediction deployed cloud. MAPs are...
Emerging serverless computing technologies, such as function a service (FaaS), enable developers to virtualize the internal logic of an application, simplifying management cloud-native services and allowing cost savings through billing scaling at level individual functions. Serverless is therefore rapidly shifting attention software vendors challenge developing cloud applications deployable on FaaS platforms. In this vision paper, we present research agenda RADON project (...
Building a fault-tolerant edge system that can quickly react to node overloads or failures is challenging due the unreliability of devices and strict service deadlines modern applications. Moreover, unnecessary task migrations stress network, giving rise need for smart parsimonious failure recovery scheme. Prior approaches often fail adapt highly volatile workloads accurately detect diagnose faults optimal remediation. There thus robust proactive fault-tolerance mechanism meet level...
The design of autonomic systems often relies on representative benchmarks for evaluating system performance and scalability. Despite the fact that experimental observations have established burstiness is a common workload characteristic has deleterious effects user-perceived performance, existing client-server do not provide mechanisms injecting into workload. In this paper, we introduce new methodology generating workloads emulate temporal surge phenomenon in controllable way, thus...
We propose a linear regression method and maximum likelihood technique for estimating the service demands of requests based on measurement their response times instead CPU utilization. Our approach does not require server instrumentation or sampling, thus simplifying parameterizati
The Cloud computing paradigm has revolutionised the computer science horizon during past decade and enabled emergence of as fifth utility. It captured significant attention academia, industries, government bodies. Now, it emerged backbone modern economy by offering subscription-based services anytime, anywhere following a pay-as-you-go model. This instigated (1) shorter establishment times for start-ups, (2) creation scalable global enterprise applications, (3) better cost-to-value...
Weighted round robin load balancing is a common routing policy offered in cloud balancers. However, there lack of effective mechanisms to decide the weights assigned each server achieve an overall optimal revenue system. In this paper, we first experimentally explore relation between probabilistic and weighted policies. From experiment similar behavior found these two policies, which makes it possible assign according probability estimated from queueing theoretic heuristic optimization...
ICT (Information and Communication Technology) and, in particular, software is more pervasive it cannot be considered anymore as a minor element of complex systems. In domains like cloud, big data, IoT (Internet Things), CPS (Cyber-Physical Systems) the core element. We need to consolidate engineering discipline, which, despite impressive achievements area technology, probably one youngest scientific technological disciplines with about 60 years history. This paper summarizes challenges that...