MobiUK 2021 Programme

MobiUK took place online in 2021. Recordings of the talks and abstracts can be found below.

Monday 5th July 2021

10:00

Welcome to MobiUK (Nigel Davies)

Zoom Chair: Dong Ma

10:15

Paper session 1: Federated Learning and Mobile Systems

Session Chair: Paul Patras; Zoom Chair: Hong Jia

Hongrui Shi and Valentin Radu. Federated Learning with Student Models and Attention Transfer across Heterogeneous Devices

Gautham Krishna Gudur and Satheesh Kumar Perepu. Heterogeneous Zero-Shot Federated Learning with New Classes for On-Device Audio Classification

Daniel Beutel, Taner Topal, Akhil Mathur, Xinchi Qiu, Titouan Parcollet, Pedro Porto Buarque de Gusmao and Nicholas Lane. Flower: An Open-source Federated Learning Framework for both Industry and Research

11:15

Break

11:45

Paper session 2: Sensing and Machine Learning

Session Chair: Katayoun Farrahi; Zoom Chair: Pedro Porto Buarque De Gusmao

Edgar Liberis, Lukasz Dudziak and Nicholas Lane. μNAS: Constrained Neural Architecture Search for Microcontrollers

Mohammad Malekzadeh, Richard G. Clegg, Andrea Cavallaro and Hamed Haddadi.

Making Neural Networks Adaptive to Changes in the Dimensions of Sensor Data

Shyam Tailor, Felix Opolka, Pietro Lio and Nicholas D. Lane. Adaptive Filters and Aggregator Fusion for Efficient Graph Convolutions

12:45

Lunch

14:00

Keynote 1: Is the Mobile Web Experience Improving?

Aruna Balasubramanian

Session Chair: Nic Lane; Zoom Chair: Dimitris Spathis

There is no question that mobile browsers are critical to how users access Internet content today. The problem however is that we do not have a good understanding of whether mobile browsing performance is improving. In fact, in this talk I argue that we don’t even know how to measure mobile Web performance.

There are a number of metrics used to objectively measure Web page load times, but these metrics do not measure user perception. To make matters worse, mobile users start to scroll the page even before the page is loaded, but most metrics only measure the latency of loading the first viewport. In the first part of the talk, I will discuss a new user-perception metric we define called uPLT. We conduct crowdsourcing studies to measure this uPLT metric both for desktop and mobile users; I will discuss some differences we find in these users. Based on our measurement, we developed a new model to estimate uPLT using existing metrics (that can be easily obtained from browsers).

In the second part of the talk I will switch gears to discuss performance. I will describe a tool we developed called WProfM that lets us measure the main bottlenecks during the page load process. Using this tool, we find that, on mobile browsers, the performance bottleneck is the computational tasks. However, on desktop browsers, the performance bottleneck is the network. This finding has important implications because much of the optimizations for Web focus on improving the network. Our work further shows that the compute bottleneck on mobile browsers is exacerbated in lower-end mobile devices that are popular in the developing regions of the world. In effect, our study shows that web developers rely on the presence of high-end mobile devices to mask their decreased performance. But users who do not wish (or cannot afford) to constantly upgrade their devices are experiencing a slower web.

 

Bio: Aruna Balasubramanian is an Associate Professor at Stony Brook University.  She received her Ph.D from the University of Massachusetts Amherst, where her dissertation won the UMass outstanding dissertation award and was the SIGCOMM dissertation award runner up. She works in the area of networked systems. Her current work consists of two threads: (1) significantly improving Quality of Experience of Internet applications, and (2) improving the usability, accessibility, and privacy of mobile systems. She is the recipient of the SIGMobile Rockstar award, a Ubicomp best paper award, a VMWare Early Career award, several Google research awards, and the Applied Networking Research Prize. She is passionate about improving the diversity in Computer Science and broadening participation. She leads the diversity committee at Stony Brook and is an active member of the N2Women group.

15:00

Community building session

Session Chair: Cecilia Mascolo; Zoom Chair: Dong Ma

15:45

Break

16:00

Keynote 2: Protecting User Privacy in a World Filled with Connected Sensors

Heather Zheng

Session Chair: Alastair Beresford; Zoom Chair: Jing Han

Thanks to near-ubiquitous deployment of smart connected devices (e.g. security cameras, voice assistants, smart appliances), our homes, offices, and many public spaces are filled with sensors that constantly monitor and capture our behaviors. While initially designed to improve efficiency and quality of our life,  these devices also pose a real security and privacy risk to all of us.   In this talk, we will discuss two types of security and privacy risks related to smart devices:  an explicit one (e.g., cameras and microphones that commonly exist in many devices) and an implicit one (e.g., WiFi devices that can silently monitor our movements behind the wall using WiFi  signals).   We will then discuss practical defense mechanisms against those risks.    

Bio: Heather Zheng received her PhD degree from the University of Maryland, College Park in 1999. After spending six years as researchers in industry labs (Bell-Labs, Crawford Hill, NJ, and Microsoft Research Asia), she joined the UC Santa Barbara faculty in 2005, and became Associate and Full professor in 2009 and 2013, respectively. In July 2017, she joined the University of Chicago as the Neubauer Professor in Computer Science. Some of her awards include the IEEE Fellow (2015), the MIT Technology Review’s TR-35 (Young Innovators Under 35) and the World Technology Network Fellow.  Heather has been actively working on security and privacy issues for both machine learning models and mobile computing systems. Her research work has been frequently featured by media outlets, such as New York Times, Boston Globe, LA Times, MIT Technology Review, and Computer World.   She was the TPC co-chair of MobiCom’15 and DySPAN’11 conferences, and the general co-chair of HotNets 2020.  Currently she serves on the steering committee of MobiCom, the steering committee of ACM/IEEE TON, and as the chair of the SIGMOBILE Research Highlights committee.

17:00

Close

Tuesday 6th July 2021

10:00

Paper session 3: Health

Session Chair: Sarah Clinch; Zoom Chair: Catherine Tong

Haotian Wang, Abhirup Ghosh, Jiaxin Ding, Rik Sarkar and Jie Gao. Heterogeneous Interventions Reduce the Spread of COVID-19 in Simulations on Real Mobility Data

Dimitris Spathis, Ignacio Perez-Pozuelo, Soren Brage, Nicholas Wareham and Cecilia Mascolo. Towards unsupervised wearable representations for longitudinal cardio-fitness prediction

Tong Xia, Lorena Qendro and Cecilia Mascolo. Uncertainty-Aware Digital Diagnosis from Sounds

Ahmed Ibrahim, Sarah Clinch and Simon Harper. Motivation-based Interest Recognition from Digital Phenotyping

11:00

Newcomers’ talks

Session Chair: Christos Efstratiou; Zoom Chair: Daniel Hugenroth

Jagmohan Chauhan (Southampton)

Title: On Device Learning and Audio Sensing for Healthcare

Bio: Jagmohan Chauhan is a newly minted lecturer at the department of Electronics and Computer Science, University of Southampton. Prior, he was a postdoctoral researcher in Mobile Systems Group working with Professor Cecilia Mascolo at the University of Cambridge. He completed his PhD from University of New South Wales in Australia in the area of usable security. His ongoing research interests include on-device deep learning, creating novel sensing solutions and applying principles from mobile systems, deep learning, and data science to solve some of the important issues in the area of healthcare, marine, and environmental sciences.

 

Valentin Radu (Sheffield)

Title: Training and Inference across Ubiquitous Devices

Bio: Valentin is a Lecturer in Ubiquitous Computing at the University of Sheffield since August 2020. Before that, he was a researcher in the Bonseyes project, exploring optimisation techniques for efficient inference of deep neural networks on resource constrained devices. He received his PhD from the University of Edinburgh for his work in mobile sensing.

Chris Lu (Edinburgh)

Title: Robust Perception for Autonomous Cyber-Physical Systems

Chris Xiaoxuan Lu is an Assistant professor (Lecturer, in UK parlance) in the School of Informatics at the University of Edinburgh. He received his Ph.D degree in Computer Science from the University of Oxford and a M.Eng degree from Nanyang Technological University. His research interests lie in cyber-physical systems and their intersections with robotics and artificial intelligence. His research has constantly appeared in leading venues of applied AI (e.g., CVPR/WWW/AAAI/ICRA) and cyber-physical systems (e.g., MobiCom/MobiSys/SenSys/Ubicomp), and being featured in the popular press incl. The Hacker News, TechXplore, American Security Today, Planet Biometrics and Sputnik etc.

Tam Vu (Oxford)

12:15

Lunch

13:00

Paper session 4: Location and tracking

Session Chair: Mirco Musolesi; Zoom Chair: Abhirup Ghosh

Anastasios Noulas, Jimin Tan, Rossano Schifanella and Diego Saez-Trumper. Aurama: Local Knowledge Discovery with Augmented Reality

Vatsal Mehta, Vaibhav Gandhi and Glenford Mapp. Traffic prediction and minimising delays at junctions

Pavlos Nicolaou and Christos Efstratiou. Tracking changes in daily routines of elderly users through acoustic sensing: An unsupervised learning approach

Deemah Alqahtani, Caroline Jay and Markel Vigo. Addressing Barriers to Reflection in Physical Activity Tracking

14:00

Paper session 5: Security

Session Chair: Alastair Beresford; Zoom Chair: Filip Svoboda

Diana A. Vasile, Daniel R. Thomas and Alastair R. Beresford. Automatically ensuring that only our public keys are bound to our secure messaging account

Daniel Hugenroth. Measuring Energy Consumption of Privacy-Preserving Protocols for Fun and Profit

14:30

Break

15:00

Keynote 3: Is Edge Computing Dead Already?

Mahadev Satyanarayanan (Satya)

Session Chair: Nigel Davies; Zoom Chair: Jing Han

Today, Edge Computing is a key enabler of "Edge AI'' and the Internet of Things (IoT). Defined as a tier of small cloud-like computing resources (aka "cloudlets") placed in close network proximity to mobile users and sensors, it is predicted to reach a market size of $40 billion by 2027, with a compound annual growth rate of over 30%.   Microsoft embraced edge computing as early as 2018, and by 2021 every major IT company (including Amazon, IBM, and Facebook) and telecommunications company (including AT&T, Verizon, Deutsche Telekom/T-Mobile, Vodafone, and Telefonica) has embraced it. Yet, at the cusp of its success, a foundational  assumption of edge computing is being challenged.   This is the assumption that low-latency offloading via edge computing  is vital  if mobile and IoT devices are to gain "AI-like" functionality. The recent ermergence of fixed-function on-device hardware accelerators for AI tasks calls into question the need for offloading. Why offload if you can do it all on-device?  Between these two extremes lie approaches that leverage a wide range of programmable hardware accelerators such as GPUs and FPGAs. This talk will explore the strengths and weaknesses of these approaches and conclude that they are, in fact, complementary. Based on this insight, we advocate a software-hardware co-evolution path to combine their strengths.

Bio: Satya's multi-decade research career has focused on the challenges of performance, scalability, availability and trust in information systems that reach from the cloud to the mobile edge of the Internet.  In the course of this work, he has pioneered many advances in distributed systems, mobile computing, pervasive computing, and the Internet of Things (IoT). As described in "How we created edge computing",  Satya's seminal 2009 publication “The Case for VM-based Cloudlets in Mobile Computing” and the ensuing research has  led to the emergence of Edge Computing (also known as "Fog Computing").  Satya is the Carnegie Group University Professor of Computer Science at Carnegie Mellon University.  He received the PhD in Computer Science from Carnegie Mellon, after Bachelor's and Master's degrees from the Indian Institute of Technology, Madras. He is a Fellow of the ACM and the IEEE. More info: http://www.cs.cmu.edu/~satya

16:00

Concluding remarks (Mirco Musolesi)