Munich

Opening

Time Subject Resources
08:30-09:00

Opening Speech


Jörg Ott & Aaron Yi Ding, TU Munich
A welcoming speech from the workshop host.

Data Privacy


Session Chair: Sasu Tarkoma, University of Helsinki

Time Subject Resources
09:00-09:25

Diffix: Anonymity and Utility Too


Paul Francis, Max Planck Institutes
A 40-year old open problem in computer science is that of how to get both strong anonymity and good utility from data. In this talk, I'll report on Diffix, a new approach to data anonymization that makes a significant advance, maybe even breakthrough, in data anonymization. Diffix is the result of a joint research effort between MPI-SWS and the startup Aircloak.
09:25-09:50

Containing Personal Data Processing with the Databox


Richard Mortier, University of Cambridge
We are all increasingly the subjects of data collection and processing systems that use data generated both about and by us to provide and optimize a wide range of services. Means for others to collect and process data that concerns each of us -- often referred to possessively as ""your data"" -- are only increasing with the long-heralded advent of the Internet of Things just the latest example. As a result, means to enable personal data management is generally recognized as a pressing societal issue. In response to these challenges we have proposed the Databox, a collection of physical and cloud-hosted software components that provide for an individual data subject to manage, log and audit access to their data by other parties. The fundamental change this approach entails is from a world where personal data is copied to the cloud to be processed by third parties, to a world where data analysis computations are distributed to remote data stores and only the results released. This means that data subjects can have greater insight into how their data is processed, retain more control over how their data will be used in future, and avoids many of the systems and security issues that arise when creating very large, rich personal data honeypots. We will sketch the background to Databox, before discussing in more detail the platform design and implementation, as well as how we are addressing challenges in distributing analytics. For more information, see http://www.databoxproject.uk or join the discussions at https://forum.databoxproject.uk/
09:50-10:05

Visual Privacy Protection


Jiayu Shu, Hong Kong University of Science and Technology
Cameras with smaller size and higher resolution nowadays support a number of applications such as taking photos and mobile augmented reality systems. However, the ubiquitous presence of cameras, the ease of taking photos and recording video, along with “always on” and “non–overt act” features threaten individuals’ rights to have private or anonymous social lives, raising people’s concerns of visual privacy. To protect people’s visual privacy, in this talk, we first introduce Cardea, a context–aware visual privacy protection framework that enforces privacy protection on photos according to people’s dynamic privacy preferences. It provides fine–grained visual privacy protection service based on: i) face signatures, for automatically locating individuals who request privacy protection in the photo; and ii) personal privacy profiles, in which people can set their context–dependent privacy preferences, including location, scene, others’ presences, and hand gestures. We also propose a visual privacy protection protocol aiming at videos that are recorded with ubiquitous cameras. The protocol consists of: i) a trigger–and–notification interaction mechanism that allows bystanders to send out privacy notification messages once detecting camera-trigger signals from recorders, and ii) a complete video processing pipeline that protects an individual’s portrait privacy based on her fuzzy identity signature and temporary visual fingerprint.

Coffee Break 🍪 ☕


10:05 - 10:30

Mobile Networking and Architecture


Session Chair: Jörg Ott, TU Munich

Time Subject Resources
10:30-10:55

Recent Mobile and 5G Research


Sasu Tarkoma, University of Helsinki
In this talk, I will first give an overview of the Department of Computer Science at the University of Helsinki, and then present the current research activities in mobile computing and 5G areas. As a key example of the mobile computing research, I will present and discuss the Carat smartphone energy profiler and our lessons learned when analyzing mobile applications. The key idea of the methodology is to gather app, battery, and context data on mobile devices and then to analyze and diagnose the data in the cloud. We have studied various energy anomalies, user feedback effect, and analysis techniques for malware identification. Pertaining to 5G research, I will present recent results on mobile edge computing, wireless Software-Defined Networking (SDN) and network refactoring.
10:55-11:20

Low-latency Communications and Architectural Strain


Jari Arkko, Ericsson
Modern networking tools enable us to build low-latency applications, with highly optimized individual protocol components and software controlled, virtualized and tailored network functions. This talk looks at the strain that the strive for low-latency applications brings for the Internet architecture. The talk takes a system viewpoint, considering the construction applications and networks as a whole rather than viewed from individual links or specific protocol point of view.
11:20-11:45

LAMPS: A Loss Aware Scheduler for Multipath TCP over Highly Lossy Networks


Xiaoming Fu, University of Goettingen
A variety of wireless communication links today, such as HSPA+ access in high speed trains, balloon-based aerial wireless networks and satellite Internet connections have high loss rates. In such environments, Multipath TCP (MPTCP) offers a robust solution compared to regular TCP. However, MPTCP and existing schedulers suffer from performance degradation for both CBR and bulk traffic. To tackle this challenge, we develop LAMPS, a novel scheduler for MPTCP, which considers both the loss and delay when selecting subflows and chooses segments based on subflows' state. The design goal of LAMPS is to achieve a steady performance for different traffic and significantly reduce the unnecessary bandwidth consumption, especially in case of bursty losses. We have implemented LAMPS and evaluated its performance for Dynamic Adaptive Streaming over HTTP (DASH), CBR, and bulk traffic. Our experiment results show that LAMPS preserves application latency, keeps low memory consumption, and significantly reduces extra bandwidth consumption in the presence of high packet loss rate.
11:45-12:00

Requirements-driven Design of Next-Generation Mobile Networks


Matteo Pozza, University of Helsinki
5G networks come with the promise of satisfying simultaneously the requirements coming from different verticals, such as e-Health and automotive. To reach this promise, two things are required: a NFV-based flexible network architecture, and an intelligence to deploy the network functions so that the requirements in input are satisfied. In this talk, we discuss potential avenues for the creation of AI-assisted networking, as well as the deployment through the composition of network functions and elements with the modular Network-In-a-Box concept for maximizing flexibility and fault-tolerance.

Lunch 🥗 🍖 🍺


12:00 - 14:00

Novel Edge Communications


Session Chair: Paul Francis, Max Planck Institutes

Time Subject Resources
14:00-14:25

SmartVLC: When Smart Lighting Meets VLC


Qing Wang, KU Leuven
Most of the research efforts in the Visible Light Communication (VLC) area assume that the intensity of LED light is constant. This is not true when Smart Lighting is introduced to VLC, which requires the LEDs to adapt their brightness according to the intensity of the natural ambient light. The intensity adaptation severely affects the throughput performance of the data communication. In this talk, I will present SmartVLC, a system that can maximize the throughput (benefit communication) while still maintaining the LEDs' illumination function (benefit smart lighting). SmartVLC is implemented on low-cost commodity hardware and several real-life challenges in both hardware and software are addressed to make SmartVLC a robust real-time system. Comprehensive experiments demonstrate that SmartVLC supports a communication distance up to 3.6m, and improves the throughput performance of state-of-the-art solutions by up to 170%, without bringing any flickering to users.
14:25-14:50

PassiveVLC: Enabling Practical Visible Light Backscatter Communication for Battery-free IoT Applications


Chenren Xu, Peking University
This talk presents a novel and practical backscatter communication using visible light for battery-free IoT applications. Based on the idea of modulating the light retroreflection with a commercial LCD shutter, we effectively synthesize these off-the-shelf optical components into a sub-mW low power visible light passive transmitter along with a retroreflecting uplink design dedicated for power constrained mobile/IoT devices. On top of that, we design, implement and evaluate PassiveVLC, a novel visible light backscatter communication system. PassiveVLC system enables a battery-free tag device to perform passive communication with the illuminating LEDs over the same light carrier and thus offers several favorable features including battery-free, sniff-proof, and biologically friendly for human-centric use cases. Experimental results from our prototyped system show that PassiveVLC is flexible with tag orientation, robust to ambient lighting conditions, and can achieve up to 1 kbps uplink speed. Link budget analysis and two proof-of-concept applications are developed to demonstrate PassiveVLC's efficacy and practicality.
14:50-15:05

Proximity Services for Smart Communities: Enabling Technologies and Platforms


Michael Haus, TU Munich
A smart community integrates Internet of Things (IoT) and modern information and communication technologies (ICT) to manage community assets, such as Transportation systems and power plants. To enable new services such as Proximity-Based Services (PBS) for smart communities, Device-to-Device (D2D) communication is a promising technique that can leverage ICT infrastructure, especially for distributed IoT devices deployed at large scale. As a driving force for PBS, D2D enables communication among physically nearby devices and can be supported by a variety of communication technologies and context information. In this talk, we derive a proximity taxonomy that categorizes different communication Technologies and context information to serve as development guidelines for PBS in smart communities. We evaluate the communication technologies and system platforms regarding their suitability for PBS. Moreover, we developed two practical PBS systems, as use case, to demonstrate how to utilize D2D for IoT device management and data sharing among nearby users in different environments.

Coffee Break 🍪 ☕


15:05 - 15:30

Mobile Edge Applications


Session Chair: Richard Mortier, University of Cambridge

Time Subject Resources
15:30-15:55

Crowdsourcing Network and Traffic Measurements to Illuminate the Mobile Ecosystem with Lumen


Narseo Vallina-Rodriguez, IMDEA Networks - ICSI
As a society we have come to rely upon our mobile phones for myriad daily tasks. It is striking how little insight we, as mobile users and researchers, have into the operation and performance of our devices and network, into how (or whether) mobile apps protect the information we entrust to them, and with whom they share it. The research community (including the speaker) have energetically used a variety of approaches to gain empirical understanding of the mobile device/network ecosystem; however, these techniques have had to make trade-offs that affect either the scale, scope or granularity of measurements. In this talk, we will describe how we designed Lumen --a user-friendly mobile traffic monitor-- to capture mobile traffic traces locally on the device, at scale and with real user stimuli by crowd-sourcing means. We will describe our ongoing research efforts to better illuminate privacy and security aspects of mobile traffic but also for network characterisation and performance enhancement.
15:55-16:20

Edge-powered Mobile Augmented Reality Applications


Yu Xiao, Aalto University
16:20-16:35

Edge Computing for Context-aware Applications


Le Nguyen, Aalto University

Evening Activities

Social event at Schneider Bräuhaus im Tal. Details here.

Edge Services and IoT Networking


Session Chair: Xiaoming Fu, University of Goettingen

Time Subject Resources
09:00-09:25

Mobile Augmented Reality


Pan Hui, University of Helsinki & HKUST
09:25-09:50

Parallel Networks: From SDN to CPSS-Oriented Smart Networks for IoT


Feiyue Wang, Chinese Academy of Science
09:50-10:05

Authenticated Authorization


Carsten Bormann, Universität Bremen

Coffee Break 🍪 ☕


10:05 - 10:30

Mobile and Edge Offloading


Session Chair: Aaron Yi Ding, TU Munich

Time Subject Resources
10:30-10:55

Enabling GPU Offloading on Android Devices


Sokol Kosta, Aalborg University
10:55-11:20

Evidence-aware Mobile Computational Offloading


Huber Flores, University of Helsinki
Computational offloading can improve user experience of mobile apps through improved responsiveness and reduced energy footprint. A fundamental challenge in offloading is to distinguish situations where offloading is beneficial from those where it is counterproductive. Currently, offloading decisions are predominantly based on profiling performed on individual devices. While significant gains have been shown in benchmarks, these gains rarely translate to real-world use due to the complexity of contexts and parameters that affect offloading. We contribute by proposing crowdsensed evidence traces as a novel mechanism for improving the performance of offloading systems. Instead of limiting to profiling individual devices, crowdsensing enables characterizing execution contexts across a community of users, providing better generalisation and coverage of contexts. We demonstrate the feasibility of using crowdsensing to characterize offloading contexts through an analysis of two crowdsensing datasets. Our results demonstrated that crowdsensed evidence can be used to characterize the diversity of contexts, which suggests that depending on users ’mobility, opportunistic infrastructure to assist mobile devices can be dynamically allocated either in the cloud or at the edge.
11:20-11:35

Research in CARISSMA and Vehicle2X-Communication


Christian Facchi, Technische Hochschule Ingolstadt
The research centre CARISSMA has been established to develope and test a global safety system for cars. In this talk a presentation of the capabillities will be given, focussing on Vehicle2X-Communication.
11:35-11:50

FADES: Fine-grained Edge Offloading with Unikernels


Vittorio Cozzolino, TU Munich
FADES is an edge offloading architecture that empowers us to run compact, single purpose tasks at the edge of the network to support a variety of IoT and cloud services. The design principle behind FADES is to efficiently exploit the resources of constrained edge devices through fine-grained computation offloading. FADES takes advantage of MirageOS unikernels to isolate and embed application logic in concise Xen-bootable images.We have implemented FADES and evaluated the system performance under various hardware and network conditions. Our results show that FADES can effectively strike a balance between running complex applications in the cloud and simple operations at the edge. As a solid step to enable fine-grained edge offloading, our experiments also reveal the limitation of existing IoT hardware and virtualization platforms, which shed light on future research to bring unikernel into IoT domain.

Lunch 🥗 🍖 🍺


12:00 - 14:00

Edge Modelling and Architecture


Session Chair: Yu Xiao, Aalto University

Time Subject Resources
14:00-14:25

Distributed Learning in Mobile Edge Computing: An Analytical Approach


Andrea Passarella, IIT-CNR, Italy
The most widely adopted approach for knowledge extraction from raw data generated at the edges of the Internet (e.g., by IoT or personal mobile devices) is through global cloud platforms, where data is collected from devices, and analysed. However, with the increasing number of devices spread in the physical environment, this approach rises several concerns. The data gravity concept, one of the basis of Fog and Mobile Edge Computing, points towards a decentralisation of computation for data analysis, whereby the latter is performed closer to where data is generated, for both scalability and privacy reasons. Hence, data produced by devices might be processed according to one of the following approaches: (i) directly on devices that collected it (ii) in the cloud, or (iii) through fog/mobile edge computing techniques, i.e., at intermediate nodes in the network, running distributed analytics after collecting subsets of the data. Clearly, (i) and (ii) are the two extreme cases of (iii). It is worth noting that the same analytics task executed at different collection points in the network, comes at different costs in terms of traffic generated over the network. Precisely, these costs refer to the traffic generated to move data towards the collection point selected (e.g. the Edge or the Cloud) and the one induced by the distributed analytics process. Until now, deciding if to use intermediate collection points, and which one they should be in order to both obtain a target accuracy and minimise the network traffic, is an open question. In this talk, we propose an analytical framework able to cope with this problem. Precisely, we consider learning tasks, and define a model linking the accuracy of the learning task performed with a certain set of collection points, with the corresponding network traffic. The model can be used to identify, given the specification of the learning problem (e.g. binary classification, regression, etc.), and its target accuracy, what is the optimal level for collecting data in order to minimise the total network cost. We validate our model through simulations in order to show that setting, in simulation, the level of intermediate collection indicated by our model, leads to the minimum cost for the target accuracy.
14:25-14:50

Prism: A Proxy Architecture for Datacenter Networks


Lars Eggert, NetApp
In datacenters, workload throughput is often constrained by the attachment bandwidth of proxy servers, despite the much higher aggregate bandwidth of backend servers. We introduce a novel architecture that addresses this problem by combining programmable network switches with a controller that together act as a network ``Prism'' that can transparently redirect individual client transactions to different backend servers. Unlike traditional proxy approaches, with Prism, transaction payload data is exchanged directly between clients and backend servers, which eliminates the proxy bottleneck. Because the controller only handles transactional metadata, it should scale to much higher transaction rates than traditional proxies. An experimental evaluation with a prototype implementation demonstrates correctness of operation, improved bandwidth utilization and low packet transformation overheads even in software. Datacenter, TCP, proxying, rewriting, load-balancing
14:50-15:05

Towards a Hyperlocal Exascale Computational Environment


Teemu Kärkkäinen, TU Munich

Coffee Break 🍪 ☕


15:05 - 15:30

Big Data Analytics


Session Chair: Pan Hui, University of Helsinki / HKUST

Time Subject Resources
15:30-15:55

Big Data for Adaptive, Connected Mobility Services


Christian Prehofer, Fortiss & TU Munich
15:55-16:20

Carat: Collaborative Smartphone Energy Diagnosis


Eemil Lagerspetz, University of Helsinki
16:20-16:35

Pervasive Games and Human Mobility


Leonardo Tonetto, TU Munich

Evening Activities

Spontaneous Oktoberfest visit

Cloud and Energy Efficient Computing


Session Chair: Falko Dressler, Paderborn University

Time Subject Resources
09:00-09:25

GreenIoT and Other Current Work at RISE SICS


Anders Lindgren, RISE SICS, Sweden
In this talk, I will present some current work ongoing at RISE SICS, including the GreenIoT project in which a testbed for open public IoT data is being built and direct ICN access to sensor data is being tested as well as other current initiatives at the institute.
09:25-09:50

LRZ and the Munich Research Network as a Science Benchmark


Dieter Kranzlmüller, LMU & LRZ
09:50-10:05

OS, Virtualization, and Edge Computing


Pekka Enberg, University of Helsinki
Edge devices come in many flavors. You have high-end edge computing systems like HPE Edgeline, and low-end single-board, multicore systems like Raspberry Pi and ASUS Tinker Board. The edge devices are expected to offer services to other devices in their network, which highlights the importance of their networking performance. In this presentation, we discuss how current POSIX-based OS architecture limits performance of low-end edge devices and detail virtualisation overheads on the high-end edge devices. We then discuss the design of an OS which is aimed at minimizing the observed overheads. Our proposed OS builds on the insights from existing shared-nothing approaches, push event model, and kernel bypass techniques.

Coffee Break 🍪 ☕


10:05 - 10:30

Vehicular Computing and Edge Security


Session Chair: Dieter Kranzlmüller, LMU / LRZ

Time Subject Resources
10:30-10:55

Cars as Enablers for Future Information and Communication Technologies


Falko Dressler, Paderborn University
10:55-11:20

Establishing Secure Keys in Body Area Networks from Correlated Acceleration


Stephan Sigg, Aalto University
11:20-11:45

Case Studies of System Security Failures


Tuomas Aura, Aalto University
11:45-12:00

IoT-Keeper


Ibbad Hafeez, University of Helsinki

Lunch 🥗🍖🍺


12:00 - 14:00

Mobile Analytics and Blockchain


Session Chair: Stephan Sigg, Aalto University

Time Subject Resources
14:00-14:25

From Social Network Data to Mobility Patterns


Constantinos Antoniou, TU Munich
14:25-14:50

Massive Spatial and Temporal Data Computation


Weixiong Rao, Tongji University
14:50-15:15

Security, Privacy, and Performance in Next-Generation Blockchains


Bryan Ford, EPFL
Computer scientists have long known ways to build secure systems from independent, mutually distrustful parties, via tools such as Byzantine consensus and threshold cryptography - but today's “blockchain bandwagon” has finally brought to mainstream society both a realization that decentralized security is possible and a practical appreciation for its value. Currently-deployed blockchains, however, are slow, unscalable, weakly consistent, profligate in energy use, and have effectively re-centralized due to market pressures. This talk will summarize the EPFL DEDIS lab’s ongoing work to rethink blockchain architecture to improve scalability, efficiency, functionality, and decentralization. DEDIS’s ByzCoin architecture is the first blockchain system to scale and adapt Castro-Liskov PBFT consensus to a permissionless blockchain environment, demonstrably achieving two orders of magnitude higher throughput and lower transaction latency with immediate finality. Incorporating DEDIS’s new decentralized randomness protocols enable further “scale-out” scalability through secure sharding. SkipChains, a new cryptographically traversable blockchain structure, enable mobile low-power devices to follow and securely verify blockchain-based updates efficiently, even while offline or via peer-to-peer communication. Finally, combining decentralized randomness with verifiable SkipChains enables secure implementation of more efficient decentralized “mining” foundations such as investment-based Proof-of-Stake and democratic Proof-of-Personhood.

Coffee Break 🍪 ☕


15:15 - 15:30

Discussion and Closing


15:30 - 17:00