Mobile Intelligence Lab

We are a multi-disciplinary team at the University of Edinburgh, whose research crosses the boundary between network engineering and data science. We strive to make scientific advances in artificial intelligence, to improve the analysis, resilience, and management of next generation mobile systems. These efforts build on the intersection between the team's expertise in the theory and practice of networked systems, and our recent research on purpose-built neural models that can tackle complex problems in the mobile networking domain.

News

19 April, 2022 – Our work on detecting incipient large-scale network attacks using deep learning accepted for publication in Computer Communications.

3 January, 2022 – Work with Salih Ergut (OREDATA) on mobile traffic forecasting using a handover-aware spatiotemporal graph neural network accepted for publication in IEEE Communications Letters.

15 December, 2021 – Paul Patras wrote a contribution in AIhub about the research my team has been pursuing to elucidate mobile traffic consumption at city scale using AI.

12 December, 2021 – Work with Xavier Costa (NEC Labs) on Adversarial Attacks Against Deep Learning-based Network Intrusion Detection Systems and Defense Mechanisms was accepted for publication in IEEE/ACM Transactions on Networking.

7 December, 2021 – Work with collaborators at CSU on predictive audio/video bitrate adaptation was accepted for publication in IEEE Wireless Communications Letters.

10 November, 2021 – Paul Patras gave a tech talk on harnessing the power of AI for mobile traffic analysis at OREDATA.

20 October, 2021 – Paul Patras spoke about the team's research on AI-driven mobile traffic analytics at the MIT CINCS / Hamilton Institute Seminar series.

15 October, 2021 – Research with Doug Leith (Trinity College Dublin) on Android OS privacy has been featured in a number of press articles including in EuroNews, the Irish Times, and The Register.

Members

Paul Patras

Paul Patras leads the Mobile Intelligence Lab. He is a Reader (Associate Professor) and Chancellor's Fellow in the School of Informatics at the University of Edinburgh and a member of the Institute for Computing Systems Architecture (ICSA). He holds a Ph.D. in Telematics Engineering from University Carlos III of Madrid and is the first alumnus of the IMDEA Networks Institute. He held visiting research positions at the University of Brescia, Northeastern University, Technical University Darmstadt, and Rice University. His research bridges the gap between fundamental mathematical models and real-world applications of networked systems, focusing on problems related to artificial intelligence in mobile networks, performance optimisation, security and privacy, prototyping and test beds.

Haoyu Liu

Haoyu Liu is a third-year PhD student in the School of Informatics at the University of Edinburgh. He received B.Sc. degrees from the University of Edinburgh and the South China University of Technology in May 2019. His research focuses on developing machine learning tools for IoT and Network Security.

Alec Diallo

Alec Diallo is a third-year PhD student at the University of Edinburgh. Previously, he worked in Paris as a machine learning research engineer. His current research seeks to bridge the gap between the ever-evolving nature of cyber threats and the security and privacy of users' data on networked systems, by using Artificial Intelligence to build automatic network threat detection and counteraction mechanisms.

Weihe Li

Weihe Li is a first-year PhD student in the School of Informatics at the University of Edinburgh. He received a M.E. in Computer Science and Technology from the Central South University China. His research focuses on developing new techniques for accurate detection of specific types of traffic flows in high-speed networks.

Rupen Mitra

Rupen Mitra is a final-year MPhil student in the School of Informatics at the University of Edinburgh. He holds degrees in computer science and engineering from Ohio State University and University of Cincinnati. Previously, he worked with Nokia Siemens Networks and Ericsson. His research interests are at the intersection of networked systems, Internet measurements, and security.

Rui Li

Rui Li is currently a researcher at Samsung AI Cambridge, UK and an affiliated member of the Mobile Intelligence Lab. She graduated with a PhD in Informatics from the University of Edinburgh in 2019, and her thesis was advised by Dr Paul Patras. Rui is interested in domain customisations of machine learning in mobile networks and wireless communications.

Past members

Yini Fang, MPhil, 2019–2022 (now at Chinese University of Hong Kong)

Luyang Xu, Research Visitor, 2020 – 2021 (now with Computer Network Information Center, Chinese Academy of Sciences)

Alexis Duque, Post-doctoral Research Associate, 2020–2021 (now at Net AI)

Chaoyun Zhang, PhD, 2016–2020 (now at Tencent)

Media Coverage

Our survey of deep learning in mobile and wireless networks has been picked up by Chinese media and was covered in The Heart of the Machine, Trencent, NetEase, SINA, and Sohu.com.

Our security and privacy investigation of the Belkin WeMo ecosystem featured in The National, The Herald, Phys.org, Technology.org, ECN Mag, tekk.tv, qianjia.com, The Edinburgh Reporter, Edinburgh Live, Scottish Construction Now, Scottish Housing News. Forth Radio also ran a short interview with me about this work.

Research

Deep Learning Powered Mobile Network Analytics

In-depth city-scale mobile network traffic analysis is essential for a range of network engineering, such as network slicing and anticipatory resources provisioning. Unraveling in real time the geographic distribution of data consumption and predicting future demands in real-time remains challenging, due to substantial measurements, storage and post-processing overheads. This project tackles these challenges by developing dedicated deep learning techniques that extract abstract features characteristic to mobile traffic, in order to serve intelligent resource management tools in future networks.

Automatic Threat Detection and Anomaly Counteraction in Home IoT

The number of Internet-connected devices is expected to reach 1 trillion by 2035 and a large fraction of these devices will become an integral part of households. This can improve the productivity and quality of life of their users, but also exposes them to new cyber security and privacy risks. This project focuses on designing intelligent algorithms that can detect and counteract cyber threats originating from or targeting Internet of Things (IoT) devices in user homes, without requiring manual intervention. We are also investigating novel mechanisms that can allow users to operate safely with potentially compromised gadgets, while ensuring such devices will not damage the networking infrastructure or the operation of other equipment.

Recent Publications

H. Liu and P. Patras, "NetSentry: A Deep Learning Approach to Detecting Incipient Large-scale Network Attacks", to appear in Computer Communications, 2022.
Y. Fang, S. Ergut, and P. Patras, "SDGNet: A Handover-aware Spatiotemporal Graph Neural Network for Mobile Traffic Forecasting", IEEE Communications Letters, 2022.
C. Zhang, X. Costa-Perez, and P. Patras, "Adversarial Attacks Against Deep Learning-based Network Intrusion Detection Systems and Defense Mechanisms", IEEE/ACM Transactions on Networking, 2022.
W. Li, J. Huang, W. Lyu, J. Wang, and P. Patras, "VAMP: A Predictive Approach to Audio/Video Bitrate Adaptation Over Wireless Networks", IEEE Wireless Communications Letters, 2022.
Y. Fang, A. Diallo, C. Zhang, and P. Patras, "Spider: Deep Learning-driven Sparse Mobile Traffic Measurement Collection and Reconstruction" to appear in Proc. IEEE Globecom 2021.
Narengerile, J. Thompson, P. Patras, and T. Ratnarajah, "Deep Reinforcement Learning-Based Beam Training for Spatially Consistent Millimeter Wave Channels" to appear in Proc. IEEE PIMRC 2021.
A. Diallo and P. Patras, "Adaptive Clustering-based Malicious Traffic Classification at the Network Edge", in Proc. IEEE INFOCOM 2021.
C. Zhang, M. Fiore, I. Murray, and P. Patras, "CloudLSTM: A Recurrent Neural Model for Spatiotemporal Point-cloud Stream Forecasting", in Proc. AAAI 2021.