Helsinki

Programme

Wednesday July 27, 2016

Opening Session

Time Subject (Click for info)
09:00-09:25

Opening speech (Slides)


Sasu Tarkoma, University of Helsinki & Aaron Yi Ding, TU Munich
A welcoming speech from Head of the Department of Computer Science
09:25-09:50

HIIT and Research on “Augmented Research”


Petri Myllymäki, HIIT/ University of Helsinki
HIIT’s Focus Area - Augmented Research
09:50-10:15

Challenges and potential solutions for 5G security


Valtteri Niemi, University of Helsinki
Development of 5G technologies has now reached a point where study of security issues has been started in the standardization forum 3GPP. In the talk, we go through some of the key security and privacy issues that have been identified so far and discuss potential solutions. Viewpoints of some industry players such as mobile operators and cellular network vendors are also highlighted.

10:15-10:45

Coffee Break

Coffee, tea and snacks

Mobile Offloading

Session Chair: Aaron Yi Ding , TU Munich

10:45-11:10

Mobile Content Offloading in Database-Assisted White Space Networks


Jussi Kangasharju, University of Helsinki
Mobile data offloading leverages more affordable or even free network capacity to reduce the traffic experienced by cellular operators through their limited over-the-air resources. One way to harvest free capacity is to employ the white space, namely, frequencies that are assigned to licensed users but are not actively utilized, as long as no harmful interference is generated. In this talk, we characterize the benefits of harnessing node contacts for mobile content offloading through dynamic spectrum access assisted by a white space database (WSDB). We take a content-centric approach and model the selection of distributors among the subscribers of each content served through a base station. We formulate an optimization problem to maximize the offloading gain based on realistic settings. We show that such a problem is NP-hard and devise efficient heuristics for practical mobile data offloading. Our results show that the offloading gain allowed by white space is significant even when WSDB data are inaccurate.
11:10-11:35

Diffusing Your Mobile Apps: Extending In-Network Function Virtualisation to Mobile Function Offloading (Slides)


Liang Wang, University of Cambridge
We propose INFv – the first offloading system able to cache, migrate and dynamically execute on demand functionality from mobile devices in ISP networks. INFv is motivated by the huge disparity between the limited battery capacity of user devices and the ever-growing energy demands of mod- ern mobile apps. It aims to bridge this gap by extending the promising NFV paradigm to mobile applications in order to exploit in-network resources.
In this talk, we present the overall design, key algorithms, and state-of-the-art technologies adopted in the INFv system, a careful study of over 20K Google Play apps, as well as thorough evaluations with realistic settings. In addition to a significant improvement in battery life (i.e., up to 6.9x energy reduction) and execution time (up to 4x faster), INFv has two distinct advantages over previous systems: 1) a non-intrusive offloading mechanism transparent to exist- ing apps; 2) an inherent network subsystem to effectively balance computation load and exploit the proximity of in- network resources. Both advantages together enable a scalable and incremental deployment of computation offloading framework in practice.
11:35-12:00

Social-aware Hybrid Mobile Offloading: A Contribution for Edge and Fog Computing? (Slides )


Huber Flores, University of Oulu
The exploitation of an opportunistic infrastructure via computational offloading is a critical component towards the adoption of new paradigms such as edge and fog computing. Computational offloading is a promising technique to aid the processing of a mobile device. By offloading a computational task a device can save energy and increase the performance of the mobile applications. Unfortunately, in classical offloading systems, the opportunistic moments to offload a task are sporadic and short-term. In this talk, we explore a hybrid system that merges cloudlet, Device-to-Device (D2D) and remote cloud models in order to increase the spectrum of opportunistic offloading. We analyze and evaluate our system in the wild and found that in a realistic environment, a mobile is always co-located to a source that provides offloading support. This suggests that a device can schedule the processing of tasks in coordination with other devices, potentially more powerful, instead of handling the processing of the tasks by itself.

12:00-13:00

Lunch

Lunch served in the Main Building Unicafé restaurant

Edge Networking

Session Chair: Sasu Tarkoma , University of Helsinki

13:00-13:25

Pervasive Computing at the network edge and beyond (Slides)


Stephan Sigg, Aalto University
We discuss the implementation of pervasive services at the network edge and beyond. In particular, we consider the calculation of mathematical functions through simultaneous superimposition in Pervasive environments. We trade computational load for communication load for parasitically, reader-powered, or potentially backscatter smart devices. Specifically, we present a communication scheme by which mathematical computations can be executed at the time of wireless transmission."
13:25-13:50

Applications at the Edge (Slides)


Hannu Flinck, Nokia Bell Labs
"This talk describes how mobile edge computing can bring value to applications by using up-to-date information available only at the edge of the radio network. Mobile Edge Computing (MEC) platform provides additional value to all parties – developers, operators and end users through. In the context of Cellular IoT applications MEC offers fast reaction times, better reliability and optimizations for congestion and overload control of potentially massive amount signaling caused by these applications. MEC can also proxy and optimize application level data transport protocols such as HTTP, TCP and Constrained Application Protocol, etc. Edge video orchestration that was deployed at Shanghai International Circuit demonstrated its capability clearly. Throughput guidance is a tool to improve content delivery. It has been tested extensively in large scale field trials. Summary of the field trial results will be provided."
13:50-14:15

Off-the-Shelf Wi-Fi and SDN: How to Create Virtual Wires in the Wi-Fi Networks


Seppo Hätönen, University of Helsinki
The Software-Defined Networking mainly revolves around the concepts of wires and ports. In general, these do not exist in wireless networks. Currently, the best proposals to bring SDN and flow control to the wireless networks require a dedicated virtual access point per client. This requires extensive software modifications to the APs and the APs may have hardware limitations how many virtual APs they can support. In this talk, I present two simple ways to bring SDN to the wireless networks using off-the-shelf consumer and enterprise equipment with minimal changes to the APs and their controlling software."

14:15-14:45

Coffee Break

Coffee, tea and snacks

Networking

Session Chair: Yu Xiao, Aalto University

14:45-15:10

Open Connect Everywhere: A Glimpse at the Internet Ecosystem through the Lens of the Netflix CDN (Slides)


Steve Uhlig, Queen Mary University of London
Netflix has become a worldwide video-streaming platform and the source of a large amount of the Internet traffic. It has achieved this without Building its own datacentres, by controlling a network of servers deployed at Internet Echange Points (IXPs) and within Internet Service Providers (ISPs). Despite its wide success and novel approach to infrastructure deployment, we have little understanding of its deployments and operations. Through extensive measurements, we reveal Netflix's footprint, its traffic patterns and volumes. Our analysis of Netflix's server deployment exposes the diversity of the Internet ecosystem world-wide. Our findings also suggest that the specifics of each regional ecosystem, coupled with the demand of each local market, explain the different deployment strategies.
15:10-15:35

A Refactoring Approach for Optimizing Mobile Networks (Slides)


Ashwin Rao, University of Helsinki
Mobile networks are expected to serve a wide range of verticals such as Internet of Things, however the Long Term Evolution (LTE) network is optimized for basic mobile operator services and offers limited services to other verticals. Furthermore, the network functions serving an LTE network are largely implemented as dedicated single function devices that exchange a large number of signaling messages for replicating the state of the mobile devices using the LTE network. Modularizing these network functions would enable a refactoring of the LTE network allowing operators to compose networks that adapt and evolve with the influx of verticals. In this article, we present a new approach for refactoring the network functions serving LTE networks. In particular, we show that it is possible to separate and modularize the control and data planes of LTE networks without modifying the mobile devices. With the help of three use-cases, we show that our refactoring approach can be leveraged to compose a modular mobile network optimized for the verticals using its services. We also show that deploying network functions at the edge significantly reduces the signals exchanged within a mobile network.
15:35-16:00

StackMap: Low-Latency Networking with the OS Stack and Dedicated NICs (Slides)


Lars Eggert , NetAPP
StackMap leverages the best aspects of kernel-bypass networking into a new low-latency OS network service based on the full-featured TCP kernel implementation, by dedicating network interfaces to applications and offering an extended version of the netmap API for zero-copy, low-overhead data path alongside control path based on socket API. For small-message, transactional workloads, StackMap outperforms baseline Linux by 4 to 78 % in latency and 42 to 133 % in throughput. It also achieves comparable performance with Seastar, a highly-optimized user-level TCP/IP stack that runs on top of DPDK.

16:00-

Evening activities


Click for details
For interested participants: We will go and see a jazz concert at Esplanadi (Link: JazzEspa) and after that attend a craft beer festival at the Central Railway Station (Link: SOPP). The participants are required to pay for themselves. The concert is free and drinks can be purchased from the bar nearby. The beer festival entrance fee is 10€ + 3 € for the glass which you can keep or return for 2 € remimbursement. Drink price varies, but is served in small tasting portions. For additional details, feel free to contact Lauri Suomalainen by email or face to face at the venue.

Thursday July 28, 2016

Edge Applications

Session Chair: Eiko Yoneki, University of Cambridge

Time Subject (Click for info)
09:00-09:25

Off the Hook: Real-Time Client-Side Phishing Prevention System (Slides)


Samuel Marchel, Aalto University
Phishing is a major problem on the Web. Despite the significant attention it has received over the years, there has been no definitive solution . Existing solutions for steering users away from phishing websites are typically server-based and have several drawbacks: they compromise user privacy, are not robust against adaptive attackers who serve different content at different times, and do not provide any guidance to users after flagging a website as a phish.
To address these limitations, we present a new phishing prevention system implemented as a client-side application and a browser add-on. It uses information extracted from website visited by the user to detect if it is a phish and warn the user. It also determines the target of the phish and offers to redirect the user there. The underlying technique for phishing detection and target identification relies on two core observations:
  1. Although phishers try to make a phishing webpage look similar to its target, they do not have unlimited freedom in structuring the phishing webpage
  2. A webpage can be characterized by a small set of key terms; how these key terms are used in different parts of a webpage is different in the case of legitimate and phishing webpages.
Based on these observations, we developed a machine learning based phishing detection system with several notable properties: it requires very little training data, scales well to much larger test data, is language-independent, fast, resilient to adaptive attacks and implemented entirely on client-side. In addition, we developed a target identification component that can identify the target website that a phishing webpage is attempting to mimic.
09:25-09:50

Enabling Fine-Grained Access Control for Android Apps


Yao Guo, Peking University
Many Android apps access sensitive personal data such as location and contacts. The current permission-based access control cannot control how sensitive might be used within the same app. For example, if an app obtains the permission to access location, it can use it for both navigation and advertisement. In my talk, I will present out recent work on how to infer the purpose of permission uses within an app, as well as how to provide fine-grained access control based on purposes and UI components.
09:50-10:15

Mobile Augmented Reality


Pan Hui, HKUST
Mobile Augmented Reality (MAR) is widely regarded as one of the most promising technologies in the next ten years. With MAR, we are able to blend information from our senses and mobile devices in myriad ways that were not possible before. The way to supplement the real world other than to replace real world with an artificial environment makes it especially preferable for applications such as tourism, navigation, entertainment, advertisement, and education. In this talk, I will first give an overview of the MAR research in our HKUST-DT System and Media Lab and then I will use two systems that we have developed as examples to illustrate the research challenges that we have to face for our research. These two MAR systems are Ubii – Ubiquitous Interface for Seamless Interaction between Digital and Physical Worlds, and Hyperion – A Wearable Augmented Reality System for Text Extraction and Manipulation in the Air. I will go through them in details.

10:15-10:45

Coffee Break

Coffee, tea and snacks

Mobile and Sensing Applications

Session Chair: Pan Hui , HKUST

10:45-11:10

Understanding the Cyber-Physical Space by Mobile Operator Big Data


Yong Li, Tsinghua University
By focusing on characterizing the mobile traffic, web and information usage traces based on large-scale and long-time mobile big data, which is collected from the commercial mobile operator with more than 10 thousand base stations and 6.5 million users spanning over some months, we qualitatively visualize and quantitatively characterize the patio-temporal human behaviors in the physical-cyber system including mobility regularity, traffic consumption patterns, social friendship activity, online information and commodity consumption, etc. Based on these fundamental findings and credible models, we further investigate how to utilize these important insights on how to deal with the problems encountered with the current mobile networks, urban management, and robust physical-cyber systems.
11:10-11:35

Contextual factors in mobile security and privacy policy enforcement (Slides)


Markus Miettinen, TU Darmstadt
With the increasing popularity of smartphone-based applications and mobile computing, the role of context as a factor in security and privacy enforcement is gaining in importance, both as an asset to be protected and as a factor to base security policy decisions on.
With context we mean any observable features of a mobile device's ambient environment, which the device is able to observe using its on-board sensors. Examples of possible context modalities include, e.g., ambient light, noise level, temperature as well as WiFi and Bluetooth devices in the user's RF environment. While the use of context information opens great opportunities for developing apps, it also involves threats against the user's privacy, as many apps and device features require access to information about the user's context. Contextual data can potentially reveal a lot of private information about the user, if it is revealed to unauthorised parties. However, context can also help in adapting the behaviour of a mobile device and adjusting security and privacy policies according to specific situations. In this talk I will give an introduction to some recent work related to context-based policy enforcement and the use of context in security protocols, in particular as an enabler for proofs-of-presence that are not dependent on a particular infrastructure.
11:35-12:00

Context-aware Real-time Population Estimation for Metropolis (Slides)


Fengli Xu, Tsinghua University
Achieving accurate, real-time, and spatially fine-grained population estimation for a metropolitan city is extremely valuable for a variety of applications. Previous solutions look at data generated by human activities, such as night time lights and phone calls, for population estimation. However, these mechanisms cannot achieve both real-time and fine-grained population estimation because the data sampling rate is low and spatial granularity chosen is improper. We address these two problems by leveraging a key insight — people frequently use data plan on cellphones and leave mobility signatures on cellular networks. Therefore, we are able to exploit these cellular signatures for real-time population estimation. Extracting population information from cellular data records is not easy because the number of users recorded by a cellular tower is not equal to the population covered by the tower, and mobile users’ behavior is spatially and temporally different, where static estimating model does not work. We exploit context-aware city segmentation and dynamic population estimation model to address these challenges. We show that the population estimation error is reduced by 22.5% on a cellular dataset that includes 1 million users.

12:00-13:00

Lunch

Lunch served in the Main Building Unicafé restaurant

Internet of Things

Session Chair: Stephan Sigg , Aalto University

13:00-13:25

Securebox: A Platform for Smarter and Safer Networks (Slides)


Ibbad Hafeez, University of Helsinki
The number of connected devices is increasing exponentially, which has made the job of managing and securing networks more complex and demanding than ever before. In this paper, we present a novel service-based solution for securing edge networks that are poorly managed and do not offer adequate security and management features. Our proposed system includes a smart gateway Securebox offering advanced security and network management features at device level granularity and a Security and Management Service (SMS) which provides services including traffic analysis services, management services for remote device, network and security policy etc. Instead of tight coupling with hardware, our system enables flexible and on-demand deployment of security services to detect and block malicious activities in the network. Our proposed system is easy to deploy, manage and operate different networks and resolves a number of challenges in network security management domain.
13:25-13:50

Data Meet at Wireless Edge: Web Caching and IoT case


Zhen Cao, Huawei
The increasing computing and storage capacity at the wireless access nodes (eNodeB and Access Points) creates opportunities for content caching at the wireless edge and, therefore, further reduction of content delivery latency. However, realizing a large-scale content caching at the wireless edge entails significant challenges, due to the difficulties of managing the vast amount of cached content and handling redirections among the large number of nodes. This talk covers a collaborative content caching at wireless edge nodes by coupling a low-false-positive hash table to index the cached content with centralized content redirection. This architecture is applicable to both web caching and IoT cases.
15:50-14:15

Enabling Efficient Deep Learning-Based Inferencing on Wearable Platforms


Sourav Bhattacharya, Nokia Bell Labs
Breakthroughs from the field of deep learning are radically changing how sensor data are interpreted. It is critical that the gains in inference qualities that deep models afford become embedded in the future generation of mobile and IoT applications. However, significant requirements of memory and computational power have been the main bottlenecks in the wide-scale adoption of these novel computational techniques on resource constrained wearable and mobile platforms.
In this talk, we present novel design principles that significantly lower the device resources (viz. memory, computation, energy) required by deep learning to overcome severe challenges to their mobile adoption. The main foundations of our approach are based on resource control algorithms and runtime resource scaling of deep model architectures, which are designed for the inference stage. Experiments show that the proposed optimizations allow even large-scale deep learning models to execute efficiently on modern mobile processors and significantly outperform existing solutions, such as cloud-based offloading.
14:15-14:45

Coffee Break

Coffee, tea and snacks

Localization and Sensing

Session Chair: Yong Li , Tsinghua University

14:45-15:10

Crowdsourced Indoor Mapping and Navigation (Slides)


Yu Xiao, Aalto University
In this talk, we will present, iMoon, an image-based indoor mapping and navigation system. We will discuss the opportunities and challenges of utilizing mobile crowdsensing for creating fine-grained indoor maps with low cost, and will introduce the results of our performance tests and the progress of our user studies.
15:10-15:35

Deviations of Check-in Locations and Human Mobility Trajectory (Slides )


Xichen Wang, Tshingua University
As cluster analysis and visualization on massive check-in data can clearly reveal the population spatial distribution and other trajectory characteristics, lots of work make further efforts on finding geographic spatial distribution, population scale, spatial structure and functional area distribution. Thus, the validity of check-in locations is of important significance. This talk focuses on the investigation of discrepancy of check-in and network trajectory via a large scale Sina Weibo data. First, we classify functional area distribution through the human network accesses. Then, we provide the observation that the discrepancy of locations where people access network, Sina Weibo and check-in locations. Finally, we will compare the difference of people posted on web via check-in and their actual location.
15:35-16:00

Measurement-driven Capability Modeling for Mobile Network in Large-scale Urban Environment


Jingtao Ding, Tsinghua University
For mobile networks diverse usage scenarios have different capability requirements on connection density and user experienced data rate, and modeling such capability diversity is crucial to the strategy evaluation in addressing the problem of high traffic load and scalability of network resources. Therefore, it is necessary to build a capability model in two dimensions of connection density and user experienced data rate. This talk will be focused on addressing this challenge based on an investigation of network capability in large-scale urban environment. There are mainly three parts:
  1. The spatial distribution of these two parameters can be accurately fitted by log-normal mixture model.
  2. Six basic capability patterns exist among the 9,000 cellular base stations.
  3. We build a network capability model which can generate synthetic base stations with diverse connection density and user experienced data rate.

19:00-

Social Dinner at Restaurant Savu


Click for details
Join the three course dinner in the Restaurant Savu on the island Tervasaari. See the venue page for directions!

Friday July 29, 2016

Security and Privacy

Session Chair: Valtteri Niemi, University of Helsinki

Time Subject (Click for info)
09:00-09:25

Security in Connecting Devices to the Cloud (Slides )


Tuomas Aura, Aalto University
New Internet of Things applications typically require smart devices to be connected to the cloud. I'll discuss some of the issues in creating a secure connection where one side may be a device that does not yet know its owner and the other side is a virtual cloud service with no physical presence. The configuration of smart cloud-managed displays and the EAP-NOOB protocol are used as a case study.
09:25-09:50

Notes on Mobile Malware


Alexey Kirichenko, F-Secure Corporation
F-Secure was one of the very first companies that seriously invested into anti-malware protection for mobile devices in the beginning of 2000-s. We will briefly review the developments in that domain at large and at F-Secure and then talk about threat landscape and challenges of the recent past.
09:50-10:15

Cryptocurrency for a Device-to-Device Ecosystem (Slides)


Dimitris Chatzopoulos, HKUST
The popularity of digital currencies, especially crypto-currencies, has been continuously growing since the appearance of Bitcoin. Bitcoin is a peer-to-peer (P2P) cryptocurrency protocol enabling transactions between individuals without the need of a trusted authority. Its network is formed from resources contributed by individuals known as miners. Users of Bitcoin currency create transactions that are stored in a specialised data structure called a block chain. Bitcoin's security lies in a  proof-of-work scheme, which requires high computational resources at the miners. These miners have to be synchronised with any update in the network, which produces high data traffic rates. Despite advances in mobile technology, no cryptocurrencies have been proposed for mobile devices. This is largely due to the lower processing capabilities of mobile devices when compared with conventional computers and the poorer Internet connectivity to that of the wired networking. In this work, we propose  LocalCoin, an alternative cryptocurrency that requires minimal computational resources, produces low data traffic and works with off-the-shelf mobile devices. LocalCoin replaces the computational hardness that is at the root of Bitcoin's security with the social hardness of ensuring that all witnesses to a transaction are colluders. It is based on opportunistic networking rather than relying on infrastructure and incorporates characteristics of mobile networks such as users' locations and their coverage radius in order to employ an alternative proof-of-work scheme. Local coin features (i) a lightweight proof-of-work scheme and (ii) a distributed block chain.

10:15-10:45

Coffee Break

Coffee, tea and snacks

Big Data and Applications

Session Chair: Tuomas Aura, Aalto University

10:45-11:10

Carat: Before and After (Slides)


Eemil Lagerspetz, University of Helsinki
While Carat is a research project at the University of Helsinki, with the aim to collect a large-scale smartphone usage dataset and produce high quality research with it, the project needs its flagship smartphone applications to remain useful and attractive to their users.
For this purpose, the Carat application has been redesigned with a new look and updated visuals. This talk will cover the new application and issues we have had to overcome in order to make it suit our needs on both iOS and Android.
11:10-11:35

Efficient Large-Scale Graph Processing on Single Computer (Slides)


Eiko Yoneki, University of Cambridge
One of challenges of emerging big data research is to perform efficient and robust data processing with limited memory/computation power. I will introduce large-scale data processing from various perspectives: what and why large data, what technologies apply to big data, and what are the main challenges. Then, I would present our recent work on the graph processing that have billions of vertices and edges in a commodity single computer. Executing algorithms results in access to the external memory (e.g. SSD) and performance of I/O takes an important role, regardless of the algorithmic complexity or runtime efficiency of the actual algorithm in use.
11:35-12:00

User privacy is not Preserved with ID-removed Anonymous Cellular Data (Slides)


Zhen Tu, Tshingua University
A large number of cellular network accessing records are generated by cellular users, collected by cellular network operators and applications running on mobile phones, and publicly released for academic research or privately sold for commercial purpose. One step the mobile operators and application vendors usually do before releasing is removing the identification(ID) of each entry in the records. Many cellular operators and applications vendors believe that such anonymization is sufficient for preserving the privacy of mobile users. However, in this paper, we argue and prove that simply removing the IDs is not sufficient for preserving mobile users’ privacy. We develop a mechanism that is able to extract the mobility patterns of users from cellular records and associate each record with a user, which means the ID is recovered, according to their mobility characteristics. We are able to associate 70%∼ 80% records with a user for two data sets collected from both mobile application and mobile operator side at the scale of several thousands to tens of thousands users. We find that the number of users released, the temporal and spatial granularity, and the speed of user movement are key factors that impact the privacy of the ID-removed anonymized cellular data.

12:00-13:00

Lunch

Lunch served in the Main Building Unicafé restaurant

Group Discussion

13:00-13:25

City-Scale Localization with Telco Big Data


Weixion Rao, Tongji University
It is still challenging in telecommunication (telco) industry to accurately locate mobile devices (MDs) at city-scale using the measurement report (MR) data. The MR measure parameters of radio signal strengths when MDs connect with base stations (BSs) in telco networks for making/receiving calls or mobile broadband (MBB) services. In this paper, we find that the widely-used location based services (LBSs) have accumulated lots of over-the-top (OTT) global positioning system (GPS) data in telco networks, and can be automatically used as training labels for learning accurate MR-based positioning systems. Benefiting from these telco big data, we deploy a context-aware coarse-to-fine regression (CCR) model in Spark/Hadoop-based telco big data platform for city-scale localization of MDs. Our experiments on real Shanghai city data show that our model achieves a mean error of 110m and a median error of 80m, outperforming the state-of-art range-based and fingerprinting localization methods.
13:25-15:00

Open Discussion


Open discussion session to discuss research ideas and collaborations.

15:00-15:30

Coffee Break

Coffee, tea and snacks

Closing Session

15:30-15:45

Closing Session


Sasu Tarkoma & Aaron Yi Ding
Conclusion for the workshop