Mobile users can open program in new tab for better viewing.

Open program in new tab

Day 1 07/12/2020
Room #1

Welcome Message by Organizing Committees 09:00 - 09:05

Germany time zone - CET

Welcome Message by EAI Conference Manager 09:05 - 09:10

Welcome Message by EAI Community Manager 09:10 - 09:15

Keynote Presentation by Professor Lin Wang 09:15 - 10:10

Coffe Break 10:10 - 10:20

Session 1: Mobile Networks 10:20 - 12:10

10:20 - 12:10
RedMesh: A WiFi-Direct Network Formation Algorithm for Large-Scale Scenarios

Device-to-device (D2D) communication enables collaboration between mobile devices, even when no communication infrastructure is available. In this context, WiFi-Direct presents itself as a technology able to provide D2D communication at WiFi speed and coverage range. Although its specification only addresses communication inside small groups (typically 8 devices), some solutions for inter-group communication have been proposed, and, atop such solutions, automatic network formation algorithms are now appearing. These initial proposals are, however, neither efficient for large scale scenarios, due to the use of broadcasts, nor effective, as they offer limited connectivity. In this paper we propose RedMesh, the first algorithm that creates mesh networks of off-the-shelf WiFi-Direct enabled devices, establishing connections that exclusively use unicast communication. The algorithm proved to be very effective, achieving full connectivity in 97.28% of the 1 250 tested scenarios with up to 250 nodes, in a total of 187 500 nodes.
Authors: António Teófilo (ISEL - Instituto Politécnico de Lisboa), João Lourenço (NOVA LINCS - NOVA University Lisbon), Hervé Paulino (NOVA LINCS - NOVA University Lisbon),
Hide Authors & Abstract

Show Authors & Abstract
10:20 - 12:10
An Empirical Analysis of the Progress in Wireless Communication Generations

The controversy and argument on the usefulness of the physical layer (PHY) academic research for wireless communications are long-standing since the cellular communication paradigm gets to its maturity. In particular, researchers suspect that the performance improvement in cellular communications is primarily attributable to the increases in telecommunication infrastructure and radio spectrum instead of the PHY academic research, whereas concrete evidence is lacking. To respond to this controversy from an objective perspective, we employ econometric approaches to quantify the contributions of the PHY academic research and other performance determinants. Through empirical analysis and the quantitative evidence obtained, albeit preliminary, we shed light on the following issues: 1) what determines the cross-national differences in cellular network performance; 2) to what extent the PHY academic research and other factors affect cellular network performance; 3) what suggestions we can obtain from the data analysis for the stakeholders of the PHY research.
Authors: Kevin Luo (Kobe University), Shuping Dang (KAUST), Chuanting Zhang (King Abdullah University of Science and Technology), Basem Shihada (KAUST), Mohamed-Slim Alouini (KAUST),
Hide Authors & Abstract

Show Authors & Abstract
10:20 - 12:10
Kairos: a self-configuring approach for short and accurate event timeouts in IoT

The Internet of Things (IoT) consists of embedded sensors that transmit events over a wireless network. Complex event processing provides powerful abstractions to aggregate and analyze relationships among event streams in real time, which can improve IoT application development and management. A key challenge are stream imperfections, because of the non-deterministic nature of IoT: events can be delayed due to variance in delay, or may even be missing due to packet loss. Timeouts are used to handle stream imperfections, distinguishing between delayed and lost events. For some applications, missing delayed events is costly, as the quality of the result depends on the presence of all inputs, but reacting too late to event non-arrivals can also lead to incorrect results. State-of-the-art results in timeouts that are impractically large when little to no missed events are tolerated by the application. We propose Kairos, a novel, self-configuring technique for determining event arrival timeouts in IoT that are both small and ensure that little to no events are missed, while also eliminating the overhead and complexity of user configuration. We evaluate our approach against the state-of-the-art using two representative IoT networks: SmartMesh IP and LoRaWAN. The results show that our solution is capable of reducing timeouts by up to two orders of magnitude even when little to no missed events are tolerated, thus satisfying the aforementioned application requirements.
Authors: Stefanos Peros (KU Leuven), Emekcan Aras (KU Leuven), Wouter Joosen (DistriNet-IBBT - KULeuven), Danny Hughes (KU Leuven),
Hide Authors & Abstract

Show Authors & Abstract
10:20 - 12:10
Visually-defined Real-Time Orchestration of IoT Systems

In this work, we propose a method for extending Node-RED to allow the automatic decomposition and partitioning of the system towards higher decentralization. We provide a custom firmware for constrained devices to expose their resources, as well as new nodes and modifications in the Node-RED engine that allow automatic orchestration of tasks. The firmware is responsible for low-level management of health and capabilities, as well as executing MicroPython scripts on demand. Node-RED then takes advantage of this firmware by (1) providing a device registry allowing devices to announce themselves, (2) generating MicroPython code from dynamic analysis of flow and nodes, and (3) automatically (re-)assigning nodes to devices based on pre-specified properties and priorities. A mechanism to automatically detect abnormal run-time conditions and provide dynamic self-adaptation was also explored. Our solution was tested using synthetic home automation scenarios, where several experiments were conducted with both virtual and physical devices. We then exhaustively measured each scenario to allow further understanding of our proposal and how it impacts the system's resiliency, efficiency, and elasticity.
Authors: Margarida Silva (Faculty of Engineering, University of Porto), João Dias (Faculty of Engineering, University of Porto), André Restivo (Faculty of Engineering, University of Porto), Hugo Ferreira (Faculty of Engineering, University of Porto),
Hide Authors & Abstract

Show Authors & Abstract
10:20 - 12:10
Performance Analysis of D2D-enabled Non-orthogonal Multiple Access in Cooperative Relaying System

Recently, device-to-device (D2D) enabled non-orthogonal multiple access (NOMA) is emerged as a new paradigm to improve spectrum efficiency. However, existing researches are confined to two-hop relaying scenarios, which restricts the coverage of the wireless networks. In this paper, we propose a novel multi-hop relaying scheme where both D2D transmitter and receiver are used as relays. Further, the cell-edge user employs maximal ratio combing (MRC) to enhance the quality of the received signal, of which the analysis becomes more complicated. Exact new closed-form expressions are provided for the cumulative distribution functions (CDFs) of the received signal-to-interference-plus-noise-ratio (SINR) at all users. Analysis on outage probability and ergodic capacity is presented and validated by simulations. It is demonstrated that the proposed scheme significantly outperforms typical two-hop relaying schemes.
Authors: Qian Cheng (University of Chinese Academy Science), Shunliang Zhang (Institute of Information Engineering.CAS), Dan Wang (University of Chinese Academy Science), Xiaona Li (Institute of Information Engineering, Chinese Academy of Sciences), Xiaohui Zhang (Institute of Information Engineering, Chinese Academy of Sciences),
Hide Authors & Abstract

Show Authors & Abstract

Lunch Break 12:10 - 12:40

Session 2: Smart Urban Mobility 12:40 - 14:50

12:40 - 14:50
Generation of Realistic Activity Scenarios for SUMO

The SUMO traffic simulator is a mainstream tool that allows to model and analyse traffic and mobility scenarios. Fully realistic scenarios can be appealing for many use cases, but they require an initial large investment of resources for their creation. In fact, the usual workflow comprises the manual creation of a statistics file with detailed information about the city and its properties, up to describing how many people live and work on each road. This step is followed by the application of the tool ACTIVITYGEN to generate activity-based traffic. Current alternatives are based on simple randomly generated traffic, such as by means of the tool randomTrips. We present a compromise between the two approaches, consisting of mathematical techniques to generate schools, city gates, population density with residential and industrial areas, and a city centre. We also introduce an accompanying tool, randomActivityGen, which implements the approach to create ACTIVITYGEN statistics files automatically. Evaluation of generated scenarios shows that population and industry density, schools, and city-gates are placed realistically through testing on five representative Danish cities. The approach is also compared with the output of the tool randomTrips and the LuST scenario.
Authors: Falke Carlsen (Aalborg University), Jacob Rasmussen (Aalborg University), Mathias Sørensen (Aalborg University), Nicolaj Jensen (Aalborg University), Michele Albano (Aalborg University),
Hide Authors & Abstract

Show Authors & Abstract
12:40 - 14:50
dFDA-VeD: A Dynamic Future Demand Aware Vehicle Dispatching System

With the rising demand of smart mobility, ride-hailing service is getting popular in the urban regions. These services maintain a system for serving the incoming trip requests by dispatching available vehicles to the pickup points. As the process should be socially and economically profitable, the task of vehicle dispatching is highly challenging, specially due to the time-varying travel demands and traffic conditions. Due to the uneven distribution of travel demands, many idle vehicles could be generated during the operation i different subareas. Most of the existing works on vehicle dispatching system, designed static relocation centers to relocate idle vehicles. However, as traffic conditions and demand distribution dynamically change over time, the static solution can not fit the evolving situations. In this paper, we propose a dynamic future demand aware vehicle dispatching system. It can dynamically search the relocation centers considering both travel demand and traffic conditions. We evaluate the system on real-world dataset, and compare with the existing state-of-the-art methods in our experiments in terms of several standard evaluation metrics and operation time. Through our experiments, we demonstrate that the proposed system significantly improves the serving ratio and with a very small increase in operation cost.
Authors: yang guo (Macquarie University), Tarique Anwar (Macquarie University), jian yang (macquarie University), Jia Wu (Macquarie University, Australia),
Hide Authors & Abstract

Show Authors & Abstract
12:40 - 14:50
Towards Bike Type and E-Scooter Classification With Smartphone Sensors

In this paper, we present a novel approach for identifying various bike types and e-scooters using sensor readings from the cyclist's smartphone. Bike type identification is necessary to provide context-aware navigation services that consider e-scooter- and bike-specific road conditions in route planning and improve safety and comfort by suggesting roads suitable for the cyclist's bike type. In addition, the idea of bike type identification is useful for advertising purposes or for improving VPA (Virtual Personal Assistants) capabilities with non-intrusive, bike or e-scooter specific suggestions. We employ a CNN (Convolutional Neural Network) deep learning approach to differentiate between various bike-types and e-scooters. The evaluation includes various roads, cyclists, bike types and smartphones. The results show that bike types are identified with average F1-scores, Accuracy and AUC of up to 0.92, 0.90 and 0.98 respectively.
Authors: Viktor Matkovic (University of Duisburg-Essen), Marian Waltereit (University of Duisburg-Essen), Peter Zdankin (University of Duisburg-Essen), Torben Weis (University of Duisburg-Essen),
Hide Authors & Abstract

Show Authors & Abstract
12:40 - 14:50
Learned Taxi Fare for real-life trip trajectories via Temporal ResNet Exploration

Accurate taxi fare forecasting in complex and crowded scenarios is an important building block to enabling intelligent transportation systems in a smart city.Given the observation,increasing popularity of taxi services such as Uber and Didi Chuxing in China, unable to collect large-scale taxi fare data continuously. Traditional taxi fare prediction methods mostly rely on time series forecasting techniques, which fail to model the complex non-linear spatial and temporal relations. To address those issues, we propose a Deep Multi-View Network called Temporal ResNet (TRES-Net) framework. Specifically, our proposed model consists of three views: (i) temporal view: modeling correlations between future taxi fare values with near time points, (ii) spatial view: to model deep spatial correlations, we further introduce a spatial similarity matrix that can learn from spatially similar taxi trips and capture the multi-modality of the motion patterns, and (iii) semantic view: to extract more taxi fare patterns, we integrate more factors such as trip distance, travel time, passenger count, tolls amount, tip amount, etc.. Extensive experiments on more than 700 millions NYC trips over several fare prediction benchmarks demonstrate that our method is able to predict taxi fare in complex scenarios and achieves state-of-the-art performance. Our large scale evaluation demonstrates that our system is (a) accurate—with the mean fare error under 1 US dollar and (b) capable of real-time performance
Authors: Sayda Elmi (National University of Singapore), Tan Kian-Lee (National University of Singapore),
Hide Authors & Abstract

Show Authors & Abstract
12:40 - 14:50
Speed Prediction on Real-life Traffic Data: Deep StackedResidual Neural Network and Bidirectional LSTM

Forecasting accurate traffic speed is of great importance to advanced traffic management systems but challenging problem as it is affected by many complex factors, such as events, inter-road traffic, traffic lights, period and weather conditions. Traffic speed prediction based on deep learning techniques has received much attention in recent years. However, the power of deep learning methods has not yet fully been exploited in traffic prediction in terms of the depth of the model architecture. This paper designs a deep-learning-based structure, called BiRNet, to forecast accurate traffic speed in each and every region in a city. More specifically, we employ the residual neural network and the bi-directional recurrent neural network to model the spatial and temporal closeness, respectively. A look-up layer is introduced to model the spatial scale of the prediction area. BiRNet learns to dynamically aggregate the output of these neural networks which is further combined with external factor learning to improve the spatially correlated time series data forecasting. Our extensive experiments on real-world trip data-sets generated in Singapore and NYC's road network, show that our proposed deep learning algorithm significantly outperforms the state-of-the-art learning algorithms.
Authors: Sayda Elmi (National University of Singapore), Kian-Lee Tan (National University of Singapore),
Hide Authors & Abstract

Show Authors & Abstract
12:40 - 14:50
Efficient discovery of emerging patterns in heterogeneous spatiotemporal data from mobile sensors

Heterogeneous sensor networks, including traffic monitoring systems and telemetry systems, produce massive spatiotemporal data. Geolocated time series data and timestamped trajectory data are generally produced from fixed and mobile sensors in these systems, offering the possibility to detect events of interest. Events of interest generally comprise emerging changes, including patterns of congestion in road, utility and communication networks. However, the discovery of these actionable events is challenged by the: i) spatiotemporal and heterogeneous nature of data produced by different sensors; ii) difficulty of detecting emerging patterns hardly noticeable at early stages; and iii) massive data size. This work proposes E2PAT, a scalable method to comprehensively detect emerging patterns from heterogoeneous sources of spatiotemporal data generated by large sensor networks. We combine differencing operations and spatial constraints to identify emerging patterns distributed along geographies of interest. We show that, by relying on these principles, E2PAT has linear-time efficiency on the input data size. In addition, we propose an integrative score to measure the relevance of emerging patterns and show its role to support pattern retrieval, promote usability, and guarantee the actionability of the found patterns. These contributions are experimentally assessed in the context of the Lisbon's road traffic monitoring system, a large-scale network of mobile and fixed sensors.
Authors: Francisco Neves (INESC-ID and Instituto Superior Técnico, Universidade de Lisboa), Anna Finamore (INESC-ID and Instituto Superior Técnico, Universidade de Lisboa), Rui Henriques (INESC-ID and Instituto Superior Técnico, Universidade de Lisboa),
Hide Authors & Abstract

Show Authors & Abstract

Coffe Break 14:50 - 15:00

Session 3: Internet of Things (IoT) and Cloud Networks 15:00 - 16:10

15:00 - 16:10
State Deployment in Fog Computing

The geographical distance between mobile devices and application servers (typically hosted in cloud datacenters) penalizes mobile distributed applications with an unavoidable latency and jitter that impacts the performance of distributed applications. Fog Computing architectures mitigate this impact by deploying fragments of state and computing power on surrogate servers at the network edge. However, the performance of Fog Computing depends of a middleware service able to monitor the application and deploy each fragment at the most convenient surrogate. This paper investigates and compares different algorithms to manage state deployment in run-time. Evaluation using a realistic dataset shows that different strategies contribute to increase the performance for distributed applications with distinct data access patterns.
Authors: Diogo Lima (University of Lisbon), Hugo Miranda (Universidade de Lisboa),
Hide Authors & Abstract

Show Authors & Abstract
15:00 - 16:10
Multi-Agent DRL Based Computation Offloading and Resource Allocation in UAV enabled IoT edge Network

The combination of the Internet of Things (IoT) edge network and unmanned aerial vehicle (UAV) we call it the Internet of UAVs (IoUAVS) is a promising feature of the next-generation network. Ultra-dense IoT devices are offloading a massive amount of data for computation on edge servers. Flying UAV is currently reducing the deployment costs of traditional base stations and supporting the scalability of the network. Therefore, in this scenario, resource allocation and IoT devices association with UAVs is a challenging part of ensuring the quality of service (QoS) of different IoT devices. To tackling large space problems, the existing methods are traditional single-agent learning mechanisms. Most of these techniques are not scalable to large networks in a multi-agent scenario because of computational complexity and massive network traffic. We proposed a model-free MADRL based computation offloading and resource allocation in IoUAVs network environment. Each IoT devices are an agent that collects information from the nearest agent and take actions to make decisions independently. The objective is to minimize the long-term reward function to reduce the computational costs while guaranteeing the QoS requirements of IoT devices. We formulate the Markov game theory problem and solve using a multi-agent deep deterministic policy gradient (MADDPG) method. The simulation results are shown to obtain the optimal offloading policy and resource allocation in a real-time environment.
Authors: Mohammed Abegaz (University of Electronic Science and Technology of China), Guolin Sun (University of Electronic Science and Technology of China), Gordon Owusu Boateng (University of Electronic Science and Technology of China), Bruce Mareri (University of Electronic Science and Technology of China), Wei Jiang (Institute for Wireless Communication and Navigation University of Kaiserslautern),
Hide Authors & Abstract

Show Authors & Abstract
15:00 - 16:10
Resource-aware log monitoring data transmission for Smart and IoT devices

Along with an increase in the number of IoT devices and computations progressively shifting to the network edge and mobile systems, the complexity of computer systems is on the rise. In order to keep them in operation their monitoring becomes important. However, classic monitoring systems do not cope well with lightweight solutions due to the large volume of monitoring data sent to the cloud. In this paper we present mechanisms which can be applied to monitoring systems to make them suitable for IoT and mobile devices. We discuss and evaluate the benefits and drawbacks of both mechanisms as well as the results they produce. Finally, we propose the concept of function-driven log processor that can be used to control the proposed mechanisms.
Authors: Tomasz Szydlo (AGH-UST), Krzysztof Zielinski (AGH-UST), Marcin Jarzab (Samsung Research Poland),
Hide Authors & Abstract

Show Authors & Abstract

Coffe Break 16:10 - 16:20

Workshop session 16:20 - 18:50

16:20 - 18:50
Machine Learning Approach to Manage Adaptive Push Notifications for Improving User Experience

In this modern connected world mobile phone users receive a lot of notifications. Many of the notifications are useful but several cause unwanted distractions and stress. Managing notifications is a challenging task with the large influx of notifications users receive on a daily basis. This paper proposes a machine learning approach for notification management based upon the context of the user and his/her interactions with the mobile device. Since the proposed idea is to generate personalised notifications there is no ground truth data hence performance metrics such as accuracy cannot be used. The proposed solution measures the diversity score, the click through rate score and the enticement score.
Authors: Adithya Madhusoodanan (National Institute of Technology Karnataka), Anand M (National Institute of Technology Karnataka), Kieran Fraser (ADAPT Centre, Trinity College Dublin, Ireland), Bilal Yousuf (ADAPT Centre, Trinity College Dublin, Ireland),
Hide Authors & Abstract

Show Authors & Abstract
16:20 - 18:50
New Ways to Estimate Blood Pressure, Heartrate Variability and SpO2 Via Smartphone Camera – Proof of Concept

Blood pressure (BP) and blood glucose levels are probably the two most commonly monitored cardiovascular parameters. We can use a smartphone camera to capture a fingertip video. The brightness variation represents a photoplethysmogram (PPG), which can be analyzed via Fast Fourier Transform (FFT), FFT power feature extraction, and MultiLayer Perceptron (MLP) training. It yields acceptable values for diastolic BP (DBP), BP range (RBP) and systolic BP (SBP). Heart Rate Variability (HRV) is estimated from the FFT power spectrum by comparing the main Heart Rate (HR) frequency power with that of neighboring frequencies. SpO2 is estimated via linear regression on hue variation in the video, which is related to O2 variation in the fingertip capillaries. The use of time- and resource-consuming MLP restricts the usability of the smartphone app. The code and documentation are publicly available. As this is only a proof of concept, suggestions for further improvement are made.
Authors: H.Lee Seldon (Glen Waverley, Victoria, Australia), Bee Theng Lau (Swinburne University of Technology), Yi Ling Ong (Swinburne University of Technology),
Hide Authors & Abstract

Show Authors & Abstract
16:20 - 18:50
Detection of Vertically-Oriented Texts in Images Containing Natural Scenes

Scene texts in natural images are an expressive means of communication. For example, texts available in an image provides important information, particularly texts on product packages and road signs. Therefore, scene text detection has been an important research area in the field of computer vision. However, scene text detection in natural images has been a challenging topic due to complicated scene backgrounds such as different font sizes and low image resolutions. Different methods on detecting texts in natural scene images have been proposed, especially in detecting horizontal texts, arbitrarily-oriented texts and curved texts, but not vertical texts. Hence, a research on vertical text detection was conducted. A framework of vertical text detector which mainly detects top-to-bottom vertical texts, bottom-to-top vertical texts and horizontally-stacked vertical texts was proposed. In this proposed framework, Multi-directional Text Detector (MTD) was implemented to detect the location of vertical texts in scene images. In this paper, a new VSTD-700 dataset with vertical text instances was introduced. Preliminary results on the performance of MTD showed precision of 87% on ICDAR 2013, 73% on ICDAR 2015, 71% on MSRA-TD500 and 87% on VSTD-700 datasets. Hence, the results showed that MTD is indeed able to detect arbitrarily-oriented scene texts and vertical scene texts simultaneously.
Authors: Yi Ling Ong (Swinburne University of Technology), Bee Theng Lau (Swinburne University of Technology), Almon Chai (Swinburne University of Technology), Chris McCarthy (Swinburne University of Technology),
Hide Authors & Abstract

Show Authors & Abstract
16:20 - 18:50
How far can Wearable Augmented Reality Influence Customer Shopping Behavior

We investigate if providing shoppers with Augmented reality (AR) as a shopping tool can lead to an increase in purchase rate compared with conventional shopping applications. In a "simulated shopping" study with two groups with a total number of 20 participants, a test group used a wearable AR device (HoloLens) as a primary shopping method while a control group used a tablet as a conventional 2D shopping device. According to the surveys collected from participants, 19 participants (95%) commented that the AR method was more joyful than conventional 2D method. Furthermore, all participants said that their experience with AR technology was more realistic. 16 participants (80%) believed that the AR method was more influential than the conventional method, but the other four participants (20%) said that neither of the methods had any impact on their buying intentions. Although the mean number of products added to the basket by each person in HoloLens experience and tablet experience was not significantly different (p=0.675), the mean number of bought items was significantly higher in HoloLens experience compared to the tablet experience (p=0.004). In conclusion, AR increased customer’s interest in shopping. While it had no significant impact on adding products to the shopping cart, it affected the rate of purchase.
Authors: Hamraz Javaheri (German Research Center for Artificial Intelligence (DFKI)), Maryam Mirzaei (SRH Hochschule Heidelberg), Paul Lukowicz (DFKI & TU Kaiserslautern),
Hide Authors & Abstract

Show Authors & Abstract
16:20 - 18:50
Towards a System for Aged Care Centres based on Multiuser–Multidevice Interactions in IoT Collectives

This paper explores a possible use-case of creating an integrated multiuser-multidevice interaction (2MUDI) model in IoT collectives, in particular, in an aged care centre environment. We have designed and developed a prototype, which we named KATE. The system comprises Internet-connected robot(s), multiple mobile devices and multiple users. Family members of the seniors admitted to aged care centres can monitor the seniors via the robot. Staff members, including doctors and nurses, who look after these seniors can also interact with and use the robot(s). This data can also be accessible by family members via an application on their mobile devices. We model the complex interactions and consider the implementation challenges, societal implications in a 2MUDI system and demonstrate its applicability with the KATE system.
Authors: Amna Batool (Deakin University), Seng W Loke (Deakin University), Niroshinie Fernando (Deakin University), Jonathan Kua (Deakin University),
Hide Authors & Abstract

Show Authors & Abstract
16:20 - 18:50
Proactive Connection Migration in QUIC

QUIC provides a secure, reliable and low-latency communication foundation for HTTP. QUIC uses the connection ID to uniquely determine a connection from client to server. After the user switches the network, the server recognizes the user request according to the connection ID and continues to provide services through the connection migration technology. This paper proposes a Proactive Connection Migration (PCM) mechanism for QUIC. PCM gives QUIC the ability to select the optimal network in a heterogeneous network environment. Firstly, PCM actively perceives the different networks available to users. Then, PCM integrates the network quality exploration of different paths into the user’s multiple request actions. Finally, PCM takes response delay and jitter into account, and uses online learning to fnd the optimal network for current Internet service. Experimental results show that, compared with original QUIC, the average response delay of QUIC with PCM is reduced by 59.43% at most.
Authors: Lizhuang Tan (Beijing Jiaotong University), Wei Su (Beijing Jiaotong University), Yanwen Liu (Beijing Jiaotong University), Xiaochuan Gao (China Unicom), Na Li (Shandong Branch of National Computer network Emergency Response technical Team/Coordination Center (CNCERT/SD)), Wei Zhang (Shandong Computer Science Center (National Supercomputer Center in Ji’nan) Qilu University of Technology (Shandong Academy of Sciences)),
Hide Authors & Abstract

Show Authors & Abstract
16:20 - 18:50
Consumer Wearables and Affective Computing for Wellbeing Support

Wearables equipped with pervasive sensors enable us to monitor physiological and behavioral signals in our everyday life. We propose the WellAff system able to recognize affective states for wellbeing support. It also includes health care scenarios, in particular patients with chronic kidney disease (CKD) suffering from bipolar disorders. For the need of a large-scale field study, we revised over 50 off-the-shelf devices in terms of usefulness for emotion, stress, meditation, sleep, and physical activity recognition and analysis. Their usability directly comes from the types of sensors they possess as well as the quality and availability of raw signals. We found there is no versatile device suitable for all purposes. Using Empatica E4 and Samsung Galaxy Watch we have recorded physiological signals from 11 participants over many weeks. The gathered data enabled us to train a classifier that accurately recognizes strong affective states
Authors: Stanislaw Saganowski (Wroclaw University of Science and Technology), Przemyslaw Kazienko (Wroclaw University of Science and Technology), Maciej Dzieżyc (Wroclaw University of Science and Technology), Patrycja Jakimów (Wroclaw University of Science and Technology), Joanna Komoszyńska (Wroclaw University of Science and Technology), Weronika Michalska (Wroclaw University of Science and Technology), Anna Dutkowiak (Wroclaw University of Science and Technology), Adam Polak (Wroclaw University of Science and Technology), Adam Dziadek (Capgemini), Michał Ujma (Capgemini),
Hide Authors & Abstract

Show Authors & Abstract
Day 2 08/12/2020
Room #1

Session 4 - Session 4: Human-Centered Computing 09:00 - 12:00

09:00 - 12:00
SpiderHand: Supporting Arachnophobia Therapy Using Motion-Controlled Robotic Manipulator

Although robots have become an everyday companion for medical personnel in various domains, few solutions have been crafted to facilitate therapy of psychological disorders. Treatment of phobias is one of the areas which could highly benefit from introducing device-mediated interactions. To address the challenges of arachnophobia therapy, we designed SpiderHand - robot-based system for quasi-direct interaction with unpleasant creatures. We first identified the challenges of cognitive-behavioral therapy through interviews with a professional therapist and elicited functional requirements through experimental inquiry. Then, we designed and implemented an operational research prototype of motion-controlled robotic manipulator including remote touch features. We contribute a complete implementation toolkit for constructing DIY robot manipulator and provide insights on designing interactive robots for psychological therapy.
Authors: Julia Dominiak (Institute of Applied Computer Science, Lodz University of Technology, Poland), Mikołaj Woźniak (Institute of Applied Computer Science, Lodz University of Technology, Poland), Andrzej Romanowski (Institute of Applied Computer Science, Lodz University of Technology, Poland), Zbigniew Chaniecki (Institute of Applied Computer Science, Lodz University of Technology, Poland), Krzysztof Grudzień (Institute of Applied Computer Science, Lodz University of Technology, Poland),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
Ubi-Interact

Ubi-Interact is a framework designed to facilitate interactive applications consisting of a variety of individual systems and devices that are distributed over a (local) network. It is designed to allow easy integration of devices and exchange of data based on extendable common data formats. It features edge computing capabilities, allowing to analyze and transform device data. The same computing modules can also be used to manage an arbitrary number of devices in an abstract fashion, allowing ad-hoc handling of (dis-)connecting devices and making devices with similar components/capabilities interchangeable while lowering the necessity to re-implement parts of the overall system behaviour. Specification and implementation of these applications should be modular, extendable and re-usable so users can share and improve on the work of others. Performance, re-usability of once established capabilities and easy integration of devices are the main points of interest for developing this framework.
Authors: Sandro Weber (Technical University Munich), Daniel Dyrda (Technical University Munich), Marian Ludwig (Technical University Munich), Gudrun Klinker (Technical University Munich),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
VR Book: A Tangible Interface for Smartphone-based Virtual Reality

In this work we present three prototypes of the tangible VR Book: a visual marker-based solution to tangible interaction for smartphone-based Virtual Reality (VR). Smartphone-based VR represents a low barrier to entry in VR experiences given that many people nowadays own a smartphone device and that VR headsets for these devices are affordable and quick to set up when compared to desktop-based VR. Tangible interaction in smartphone-based VR has not been much explored, in our opinion, despite the fact that it can result in easy to use and engaging experiences. In this work, we explore a marker-based solution to object tracking that allows tangibles to be created in an easy and cheap way, maintaining the overall system accessible. We describe a design space for visual marker-based tangible interaction and three prototypes of a tangible VR Book that explore different aspects of the design space. We also present user feedback on their expectations regarding the interaction with the VR Book.
Authors: Jorge Cardoso (University of Coimbra, CISUC, DEI), Jorge Ribeiro (University of Coimbra, DEI),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
Finger Air Writing - Movement Reconstruction with Low-cost IMU Sensor

In this paper, we present a novel and generic handwriting system to write on an imaginary canvas and reconstruct it to a 2-D projection plane in real-time. The presented solution is based on a finger-worn low-cost setup using an Inertial Measurement Unit (IMU) with 9 degrees of freedom (DoF) from the accelerometer, gyroscope, and magnetometer. It is based on a simple sensor fusion algorithm to construct a real-time interactive atmosphere between the user and remote 2D projection screen. Finger air-writing system has the full potential to be used in virtual and augmented reality applications. The placement of the sensor on the finger gives complete freedom and allows for natural hand gesture movements without holding anything. Furthermore, we also performed a human-based preliminary evaluation with overall recognition rate of above 90% to establish the utility of low-cost IMUs in the handwriting domain to attract the attention of the research community in this direction.
Authors: Junaid Younas (DFKI & TU Kaiserslautern), Hector Margarito (Technical University Kaiserslautern), Shizen Bian (German Research Centre for Artificial Intelligence), Paul Lukowicz (DFKI & TU Kaiserslautern),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
LAMAR: Lidar based Adaptive Multi-inhabitant Activity Recognition

Human Activity Recognition is a key building block of many emerging applications such as ambient-assisted living and disability care. While HAR has made significant progress with the advancement of deep neural networks and emerging RF sensors, employment of LiDAR (Light detection and Ranging) sensor has seen few lights due to its cost and computational complexity. Given the high promise of accurate LiDAR point-clouds, we develop LAMAR (LiDAR-based Adaptive Multi-inhabitant Activity Recognition), a low-resolution low-cost solid-state LiDAR-based multiple-inhabitant activity recognition system. LAMAR employs efficient signal processing methods and novel adaptive deep learning model towards developing a multi-inhabitant Adaptive HAR system. More specifically, we propose (i) a voxelized feature representation based real-time point-cloud fine-tuning method, (ii) clustering (DBSCAN and BIRCH) and Adaptive Order Hidden Markov Model based multiple person tracking technique and (iii) a novel adaptive deep learning-based domain adaptation technique to improve the accuracy of HAR in presence of data scarcity and diversity (device, location and population diversity). We evaluated our framework via (i) a real-time collected data from 6 participants and (ii) one publicly available LiDAR activity data (28 participants) which provided promising HAR performances in multiple inhabitants scenario (max. 94%) scenarios with a 63% improvement of multiple person tracking than state-of-art framework
Authors: Mohammad Arif Ul Alam (University of Massachusetts Lowell), Md Mahmudur Rahman (University of Massachusetts Lowell), Fernando Mazzoni (University of Massachusetts Lowell), Jared Widberg (University of Massachusetts Lowell),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
Let's Forget About Exact Signal Strength: Indoor Positioning based on Access Point Ranking and Recurrent Neural Networks

Positioning is a key task in many different contexts. In the last decades, it has considerably evolved, but, while there are a lot of systems that offer a quite good performance in outdoor scenarios, the indoor realm is still under exploration. Among existing technologies and techniques for indoor positioning, the most popular one makes use of WiFi fingerprints. Such an approach has many advantages; however, its adoption as a standard for everyday life is limited due to issues like the (time) costly radio map construction, and radio signal strength fluctuations in indoor environments. In this paper, we present a novel solution for indoor positioning based on deep learning, that ignores as much as possible signal strengths, in order to reduce the adverse effects associated with their usage. It exploits signal strength only to generate a ranking-based representation of the access points associated with a fingerprint. By developing and testing two recurrent neural network models, we show that the proposed approach is able to achieve a positioning performance, based on access point ranking, comparable to the one achieved by state-of-the-art algorithms on multiple publicly available indoor datasets. As additional benefits, compared to existing ones, the developed solution is considerably more robust to signal fluctuations and simpler in terms of the considered data.
Authors: Nicola Saccomanno (University of Udine), Andrea Brunello (University of Udine), Angelo Montanari (University of Udine),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
Feature Recommendation by Mining Updates and User Feedback from Competitor Apps

Competition in mobile applications is becoming more and more intense with the increase in popularity of smart phones and mobile devices. Previous research shows that app developers spent considerable amount of time in exploiting user feedback to improve their apps. However, relying on own user feedback is insufficient for app survival in such competitive environment. It is highly important for an app developer to learn from competitors in order to keep the rank higher or become topper in the store (e.g., Google Play). In this work, we present an approach to automatically classify and rank popular and un-popular features from the competitor apps. We follow 3 steps- (1) extract features from competitor app updates (i.e., whatsNew) and user feedback (i.e, reviews), (2) filter and group the review features that are relevant to whatsNew features, then, classify whatsNew features to binary classes, such as, popular and unpopular, (3) rank and prioritize those popular and unpopular features from competitors in order to recommend developers to adopt or avoid those features. The ranking of whatsNewfeatures are done based on whatsNew-to-review relevance, user sentiments, and popularity of those features. We conduct extensive experiments on 840 updates and 262000 reviews of 84 different competitor apps of 10 categories. We found encouraging results from the experiments and the empirical evaluation validates the efficacy of our approach,hence, potential usefulness to the developers.
Authors: MD KAFIL UDDIN (Swinburne University of Technology), He Qiang (Swinburne University of Technology), Han Jun (Swinburne University of Technology), Chua Caslon (Swinburne University of Technolgy),
Hide Authors & Abstract

Show Authors & Abstract
09:00 - 12:00
Watching the Watchers: Resource-Efficient Mobile Video Decoding through Context-Aware Resolution Adaptation

Mobile computing evolution is critically threatened by the limitations of the battery technology, which does not keep pace with the increase in energy requirements of the mobile applications. A novel approach for reducing the energy appetite of mobile apps comes from the Approximate Computing field, which proposes techniques that in a controlled manner sacrifice computation accuracy for higher energy savings. Building on this philosophy we propose a context-aware mobile video quality adaptation that reduces the energy needed for video playback, while ensuring that a user's quality expectations with respect to the mobile video are met. We confirm that the decoding resolution can play a significant role in reducing the overall power consumption of a mobile device and conduct a user study with 22 participants to investigate how the context in which a video is played modulates a user's quality expectations. We discover that a user's physical activity and the spatial/temporal properties of the video interact and jointly influence the minimal acceptable playback resolution, paving the way for context-adaptable approximate mobile computing.
Authors: Octavian Machidon (University of Ljubljana), Tine Fajfar (University of Ljubljana), Veljko Pejovic (University of Ljubljana),
Hide Authors & Abstract

Show Authors & Abstract

Lunch Break 12:00 - 12:30

Session 5: Machine Learning and Mobile Networks 12:30 - 14:40

12:30 - 14:40
Deep Reinforcement Learning for UAV-Assisted Emergency Response

In the aftermath of a disaster, the ability to reliably communicate and coordinate emergency response could make a meaningful difference in the number of lives saved or lost. However, post-disaster areas tend to have limited functioning communication network infrastructure while emergency response teams are carrying increasingly more devices, such as sensors and video transmitting equipment, which can be low-powered with limited transmission ranges. In such scenarios, unmanned aerial vehicles (UAVs) can be used as relays to connect these devices with each other. Since first responders are likely to be constantly mobile, the problem of where these UAVs are placed and how they move in response to the changing environment could have a large effect on the number of connections this UAV relay network is able to maintain. In this work, we propose DroneDR, a reinforcement learning framework for UAV positioning that uses information about connectivity requirements and user node positions to decide how to move each UAV in the network while maintaining connectivity between UAVs. The proposed approach is shown to outperform other greedy heuristics across a broad range of scenarios and demonstrates the potential in using reinforcement learning techniques to aid communication during disaster relief operations.
Authors: Isabella Lee (University of Illinois at Urbana-Champaign), Vignesh Babu (University of Illinois at Urbana-Champaign), Matthew Caesar (University of Illinois at Urbana-Champaign), David Nicol (University of Illinois at Urbana-Champaign),
Hide Authors & Abstract

Show Authors & Abstract
12:30 - 14:40
A Flock-of-Starling Optimization Algorithm with Reinforcement Learning Capability

Aiming at the issues of weak learning capability and low learning efficiency of classic Particle Swarm Optimization (PSO), we propose a flock-of-starling optimization algorithm (StarlingOpt) based on reinforcement learning theory to improve PSO. Agent, as the protagonist of the algorithm, is modeled to imitate the behavior of flock of starlings. Attention mechanism is applied for agent to pay attention to its own state as well as its k-neighbor’s state so that agent can focus on more attentions to those valuable state information. Meanwhile, attention alignment is proposed for agent to integrate multiple attentions for rapid learning and state update. The experimental results show that the proposed algorithm can effectively accelerate learning speed of agent in optimization process, which enable improve the capability and efficiency to find the optimal solution on optimization problem.
Authors: Rong Xie (School of Computer Science, Wuhan University, China),
Hide Authors & Abstract

Show Authors & Abstract
12:30 - 14:40
A Deep Reinforcement Learning Approach to Fair Distributed Dynamic Spectrum Access

This paper investigates the task how to achieve fairness in distributed dynamic spectrum access (DSA). Specifically, we consider a cognitive radio network scenario with multiple primary users (PUs) and secondary users (SUs). Each PU operates in a licensed channel. We assume that there is no coordination between PUs and SUs, and no coordination among SUs. The key challenges for SUs are to: (1) avoid collisions with PUs, (2) avoid collisions with other SUs, (3) fair access of spectrum resources in an uncoordinated system, (4) deal with different PU activity patterns, (5) deal with spectrum sensing errors. To address these challenges, we propose a deep reinforcement learning (DRL) approach and an associated reward function to achieve fair access to spectrum resources. Specifically, we use the method of Dueling Double Deep Q-Networks with Prioritised Experience Replay (D3QN-PER) as DRL algorithm for each SU. In our simulation experiments, we demonstrate that the proposed approach performs better than existing DRL methods.
Authors: Syed Qaisar Jalil (The university of Newcastle, Australia), Mubashir Husain Rehmani (Cork Institute of Technology, Ireland), Stephan Chalup (1 The University of Newcastle, Newcastle, Australia),
Hide Authors & Abstract

Show Authors & Abstract
12:30 - 14:40
Towards Enhancing Fault Tolerance in Neural Networks

Fault Tolerance in Neural Networks is critical for applications that require reliable computation for long duration. The inherent fault tolerance of Neural Networks can be improved with regularization, however, the current techniques exhibit a trade-off between generalization and classification accuracy. We model a Neural Network as two distinct functional components: a Feature Extractor with an unsupervised learning objective and a Fully Connected Classifier with a supervised learning objective. Traditional approaches to train the entire network using a single supervised learning objective are insufficient to achieve the objectives of the individual functional goals optimally. In this work, a novel two phase framework with multi-criteria objective function combining unsupervised training of the Feature Extractor followed by supervised training of the Classifier Network is proposed. In the Phase I, the unsupervised training of the Feature Extractor is modeled using two games solved simultaneously in the presence of Neural Networks with conflicting objectives. The goal is to generate robust features for the input and smoothen the feature space to match with a prior distribution. In Phase II, the Feature Extractor is combined with the Fully Connected Classifier for fine-tuning on the classification task. The proposed two phase training algorithm is evaluated on four architectures with varying model complexity on standard image classification datasets: FashionMNIST and CIFAR10.
Authors: Vasisht Duddu (University of Waterloo), D Vijay Rao (Institute for Systems Studies and Analyses), Valentina Balas (Aurel Vlaicu University of Arad),
Hide Authors & Abstract

Show Authors & Abstract
12:30 - 14:40
Movement and Orientation Visualization using Wearable Inertial Sensors

Activity recognition using wearable sensors has widespread and important applications in different domains including healthcare, safety and behavior monitoring. Almost all the solutions for activity recognition are data-driven, and so, understanding the data and their characteristics is fundamental toward developing effective solutions. Visualization is perhaps the most effective approach for getting insight into data as well as communicating the insights with others. A novel visualization method that would provide additional utility to the existing methods is of utmost desire. However, such novel methods for visualization are rarely invented, particularly in the area of wearable and mobile sensing. This paper presents novel methods for visualizing movement and orientation using inertial sensors. It demonstrates the use of our methods for visualizing several activities and gestures. We also developed an efficient method for smoking puff detection leveraging the visualization methods.
Authors: Md Abu Sayeed Mondol (University of Virginia), John Stankovic (University of Virginia),
Hide Authors & Abstract

Show Authors & Abstract
12:30 - 14:40
Positioning with Map Matching using Deep Neural Networks

Deep neural networks for positioning can improve accuracy by adapting to inhomogeneous environments. However, they are still susceptible to noisy data, often resulting in invalid positions. A related task, map matching, can be used for reducing geographical invalid positions by aligning observations to a model of the real world. In this paper, we propose an approach for positioning, enhanced with map matching, within a single deep neural network model. We introduce a novel way of reducing the number of invalid position estimates by adding map information to the input of the model and using a map-based loss function. Evaluating on real-world Received Signal Strength Indicator data from an asset tracking application, we show that our approach gives both increased position accuracy and a decrease of one order of magnitude in the number of invalid positions.
Authors: Hannes Bergkvist (Sony), Paul Davidsson (Malmö University), Peter Exner (Sony),
Hide Authors & Abstract

Show Authors & Abstract

Coffe Break 14:40 - 14:50

Session 6: Sensing and Energy Optimization 14:50 - 17:00

14:50 - 17:00
Optimization of rechargeable battery lifespan in wireless networking protocols

Energy optimization is one of the major issues in the telecommunication field and particularly in wireless networks. This optimization is an essential condition for the ubiquity of wireless and mobile networks. Recent studies show that the Information and Communications Technology sector (ICT) (production, distribution, and use) amounts to 1.4\% of overall global CO2e emissions. Almost all works on wireless telecommunication protocols focus on a short term vision of energy consumption consisting of the extension of non-rechargeable battery lifespan, the reduction of the consumed energy amount, or the energy use planning. We discuss how a long-term vision of energy consumption can be conducted by taking into account the device's state of health and particularly the rechargeable battery status. We discuss how the improvement of the devices lifespan impacts on the network lifespan and the energy consumption in production and distribution phases and hence significantly improves energy efficiency. We illustrate the impact of this new paradigm on the routing problem in wireless sensor networks. We compared the new routing approach with well-known methods in terms of nodes and system lifespan. The obtained results show that consideration of the state of health of batteries extends the system life significantly and improves the success rate of data transmission.
Authors: Hakim Mabed (UBFC / FEMTO-ST), Mohamed Batta (UBFC/FEMTO-ST), Zibouda Aliouat (LRSD Laboratory, Setif-1 University),
Hide Authors & Abstract

Show Authors & Abstract
14:50 - 17:00
Exploiting SWIPT for IoT NOMA-based Diamond Relay Networks

Combining non-orthogonal multiple access (NOMA) with cooperative relaying networks has been considered as a viable approach for increasing the spectral efficiency and capacity enhancement of the system. With the exponential growth of the Internet of Things (IoT), integrating simultaneous wireless information and power transfer (SWIPT) with NOMA and diamond relaying networks can increase capacity and energy efficiency at the same time. Moreover, a diamond network topology has been established as a standard cooperative relaying model in the 3rd Generation Partnership Project (3GPP). Therefore, in this paper, we investigate SWIPT for IoT NOMA-based diamond relay networks wherein a source node transmits two symbols to a destination node through two energy harvesting (EH) based decode-and-forward relay nodes using the principle of downlink NOMA. The two EH based relay nodes harvest the energy from the source's signal and then transmits the decoded symbol to the destination node using the uplink NOMA protocol. Analytical expressions for the achievable rate of the symbols and the considered system are derived and verified with the simulation results. Moreover, an asymptotic closed-form solution for the achievable rate is also provided for mathematical tractability and simplified analysis. Our results demonstrate the effect of energy harvesting parameters and the effectiveness of the considered system over the similar system model using conventional orthogonal multiple access.
Authors: Ashish Rauniyar (University of Oslo, Norway), Paal Engelstad (University of Oslo, Norway), Olav Østerbø (Telenor Research, Norway),
Hide Authors & Abstract

Show Authors & Abstract
14:50 - 17:00
Reliability Model for Incentive-Driven IoT Energy Services

We propose a novel reliability model for composing energy service requests. The proposed model is based on consumers' behavior and history of energy requests. The reliability model ensures the maximum incentives to providers. Incentives are used as a green solution to increase IoT users' participation in a crowdsourced energy sharing environment. Additionally, adaptive and priority scheduling compositions are proposed to compose the most reliable energy requests while maximizing providers' incentives. A set of experiments is conducted to evaluate the proposed approaches. Experimental results prove the efficiency of the proposed approaches.
Authors: Amani Abusafia (The University of Sydney), Athman Bouguettaya (The University of Sydney),
Hide Authors & Abstract

Show Authors & Abstract
14:50 - 17:00
Formation-based Selection of Drone Swarm Services

Swarm of drones are increasingly being asked to carry out missions that can't be completed by one drone. Particularly, in delivery, issues arise due to the swarm's limited flight endurance. Hence, we propose a novel formation-guided framework for selecting Swarm-based Drone-as-a-Service (SDaaS) for delivery. A detailed study is carried out to highlight the effect of swarm formations on energy consumption. Two SDaaS selection approaches, i.e. Fixed and Adaptive, are designed considering the different formation decisions a swarm can take. The proposed framework considers extrinsic constraints including wind speed and direction. We propose SDaaS selection algorithms for each approach. Experimental results prove the efficiency of the proposed algorithms.
Authors: Balsam Alkouz (The University of Sydney), Athman Bouguettaya (The University of Sydney),
Hide Authors & Abstract

Show Authors & Abstract
14:50 - 17:00
Elastic Composition of Crowdsourced IoT Energy Services

We propose a novel type of service composition, called elastic composition, which provides a reliable framework in a highly fluctuating IoT energy provisioning settings. We rely on crowdsourcing IoT energy (e.g., wearables) to provide wireless energy to nearby devices. We introduce the concepts of soft deadline and hard deadline as key criteria to cater for an elastic composition framework. We conduct a set of experiments on real-world datasets to assess the efficiency of the proposed approach.
Authors: Abdallah Lakhdari (The university of Sydney), Athman Bouguettaya (The University of Sydney), Sajib Mistry (Curtin university), Azadeh Ghari Neiat (Deakin University), Basem Suleiman (The university of Sydney),
Hide Authors & Abstract

Show Authors & Abstract
14:50 - 17:00
InsideOut: Model to Predict Outside CO Concentrations from Mobile CO Dosimeter Measurements Inside Vehicles

The current pollution measurement methodology is coarse-grained where the pollution measurements are spatiotemporally few and far in-between. Our vision is to provide broadly accessible, fine-grained pollution information to a variety of end-users, and in turn, allow them to make better informed decisions using a new, more accurate information stream. To this end, this study proposes a new neural network model to estimate Carbon Monoxide (CO) concentrations outside vehicle from crowd-sourced CO measurements inside vehicles measured using mobile devices (dosimeters). End-users can benefit from the fine-grained pollution information generated by this prediction model along with data from direct measurements. A neural network is used to model the dynamic relationship between the CO measurements inside and outside a moving vehicle. The resulting neural network model is then used to predict outside CO concentrations from CO measurements inside vehicles. Mobile dosimeters were used inside and outside vehicles to collect measurements used in training a neural network. For this regression task, a new neural network architecture was designed using Convolutional layers and Gated Recurrent Unit (GRU) layers. The results show that outside CO concentrations can be estimated from inside vehicle CO measurements with high accuracy. The proposed neural network model provides a promising new and novel source of fine-grained pollution information along with direct measurement streams
Authors: Srinivas Devarakonda (Rutgers University), Senthil Chittaranjan (other), Daehan Kwak (Kean University), Badri Nath (Rutgers University),
Hide Authors & Abstract

Show Authors & Abstract
Day 3 09/12/2020
Room #1

Session 7: Socially-aware ML and IoT 09:30 - 11:00

09:30 - 11:00
A Bilateral Recommendation Strategy for Mobile Event-Based Social Networks

Mobile Event-Based Social Network (EBSN) platforms, such as Meetup and Plancast, have become increasingly popular for online organization of offline (in-person) events. The problem of the existing techniques in ESBN is that they do not reflect the bilateral (two-way) nature of efficient event planning: 1) events enroll more influential participants, and 2) participants are arranged to events they are most interested in. In this paper, we address this weakness by formally defining the bilateral recommendation problem and making two contributions to solving this problem: (a) by analyzing all types of the user’s behaviors during the selection session, we can accurately predict which event the users will eventually choose to participate in, and (b) by introducing the concepts of interpersonal similarity and interaction strength in EBSNs, we can calculate the interactive influence of users. We report the results of extensive experiments on real datasets that confirm the improved precision, effectiveness and scalability of our proposed bilateral recommendation strategy as compared to the state of the art.
Authors: Yu Zhang (University of Muenster, Germany), Sergei Gorlatch (University of Muenster, Germany),
Hide Authors & Abstract

Show Authors & Abstract
09:30 - 11:00
k-Anonymous Crowd Flow Analytics

Measuring pedestrian dynamics using the signals sent from smartphones has become popular. Notably, Wi-Fi-based systems are currently widely deployed. However, many such systems have also become subject to serious debate due to privacy infringement. For some time, secure hashing of a smartphone's unique MAC address was considered to be sufficient, yet this method has been overruled by Europe's General Data Protection Regulation which states that an individual should not be identifiable from any dataset without explicit prior consent. In this paper, we propose a novel anonymization technique that essentially anonymizes detected smartphones immediately at the sensor before any data on such a detection is stored for further analysis. Our solution borrows from the notion of k-anonymity, while avoiding its well-known drawbacks that lead to de-anonymization. Moreover, while ensuring what we coin detection k-anonymity, we also ensure high accuracy of counting measures when dealing with realistic pedestrian flows within crowds. We evaluate our solution both in a simulated environment and in a realistic environment reproducing real-life settings.
Authors: Valeriu - Daniel Stanciu (University of Twente), Maarten van Steen (University of Twente), Ciprian Dobre (University Politehnica of Bucharest), Andreas Peter (University of Twente),
Hide Authors & Abstract

Show Authors & Abstract
09:30 - 11:00
IAM - Interpolation and Aggregation on the Move: Collaborative Crowdsensing for Spatio-temporal Phenomena

Crowdsensing allows citizens to contribute to the monitoring of their living environment using the sensors embedded in their mobile devices, e.g., smartphones. However, crowdsensing at scale involves significant communication, computation, and financial costs due to the dependence on cloud infrastructures for the analysis of spatio-temporal data. This limits the adoption of crowdsensing by activists although sorely needed to inform our knowledge of the environment. As an alternative to the centralized analysis of crowdsensed observations, this paper introduces a fully distributed interpolation-mediated aggregation approach running on smartphones. To achieve so efficiently, we model the interpolation as a distributed tensor completion problem, and we introduce a lightweight aggregation strategy that anticipates the likelihood of future encounters according to the quality of the interpolation. Our approach thus shifts the centralized postprocessing of crowdsensed data to distributed pre-processing on the move, based on opportunistic encounters of crowdsensors through state-of-the-art D2D networking. The evaluation using a dataset of quantitative environmental measurements collected from 550 crowdsensors over 1 year shows that our solution significantly reduces –and may even eliminate– the dependence on the cloud infrastructure, while it incurs a limited resource cost on end devices. Meanwhile, the overall data accuracy remains comparable to that of the centralized approach.
Authors: Yifan Du (Inria Paris), Françoise Sailhan (CNAM Paris), Valérie Issarny (Inria Paris),
Hide Authors & Abstract

Show Authors & Abstract
09:30 - 11:00
Towards a Machine Learning-driven Trust Evaluation Model for Social Internet of Things: A Time-aware Approach

The emerging paradigm of the Social Internet of Things (SIoT) has transformed the traditional notion of Internet of Things (IoT) into a social network of billions of interconnected smart objects by integrating social networking facets into the same. In SIoT, objects can establish social relationships in an autonomous manner and interact with the other objects in the network based on their social behaviour. A fundamental problem that needs attention is establishing of these relationships in a reliable and trusted way, i.e., establishing trustworthy relationships and building trust amongst objects. In addition, it is also indispensable to ascertain and predict an object's behaviour in the SIoT network over a period of time. Accordingly, in this paper, we have proposed an efficient time-aware machine learning-driven trust evaluation model to address this particular issue. The envisaged model deliberates social relationships in terms of friendship and community-interest, and further takes into consideration the working relationships and cooperativeness (object-object interactions) as trust parameters to quantify the trustworthiness of an object. Subsequently, in contrast to the traditional weighted sum heuristics, a machine learning-driven aggregation scheme is delineated to synthesize these trust parameters to ascertain a single trust score. The experimental results demonstrate the significance of the proposed model.
Authors: Subhash Sagar (Macquarie University), Adnan Mahmood (Telecommunications Software & Systems Group, Waterford Institute of Technology, Republic of Ireland), Michael Sheng (Macquarie University), Munazza Zaib (Macquarie University, Sydney), Wei Zhang (The University of Adelaide, Adelaide),
Hide Authors & Abstract

Show Authors & Abstract

Coffe Break 11:00 - 11:10

Session 8: Security and Privacy 11:10 - 12:20

11:10 - 12:20
Quantifying Privacy Leakage in Graph Embedding

Graph embeddings have been proposed to map graph data to low dimensional space for downstream processing (e.g., node classification or link prediction). With the increasing collection of personal data, graph embeddings can be trained on private and sensitive data. For the first time, we quantify the privacy leakage in graph embeddings through three inference attacks targeting Graph Neural Networks. We propose a membership inference attack to infer whether a graph node corresponding to individual user's data was member of the model's training or not. We consider a blackbox setting where the adversary exploits the output prediction scores, and a whitebox setting where the adversary has also access to the released node embeddings. This attack provides an accuracy up to 28\% (blackbox) 36\% (whitebox) beyond random guess by exploiting the distinguishable footprint between train and test data records left by the graph embedding. We propose a Graph Reconstruction attack where the adversary aims to reconstruct the target graph given the corresponding graph embeddings. Here, the adversary can reconstruct the graph with more than 80\% of accuracy and link inference between two nodes around 30\% more confidence than a random guess. We then propose an attribute inference attack where the adversary aims to infer a sensitive attribute. We show that graph embeddings are strongly correlated to node attributes letting the adversary inferring sensitive information (e.g., gender or location).
Authors: Vasisht Duddu (INSA Lyon, CITI, Inria), Antoine Boutet (INSA Lyon, CITI, Inria), Virat Shejwalkar (University of Massachusetts Amherst),
Hide Authors & Abstract

Show Authors & Abstract
11:10 - 12:20
MIRES: Recovering Mobile Applications based on Backend-as-a-Service from Cyber Attacks

Many popular mobile applications rely on the Backend-as-a-Service (BaaS) cloud computing model to simplify the development and management of services like data storage, user authentication and notifications. However, vulnerabilities and other issues may lead to malicious operations on the mobile application client-side and malicious requests being sent to the backend, corrupting the state of the application in the cloud. To deal with these attacks after they happen and are successful, it is necessary to remove the immediate effects created by the malicious requests and subsequent effects derived from later requests. In this paper, we present MIRES, an intrusion recovery service for mobile applications based on BaaS. MIRES uses a two-phase recovery process that restores the integrity of the mobile application and minimizes its unavailability. We implemented MIRES in Android and with the Firebase platform and made experiments with 3 mobile applications that showed results of 1000 operations reverted in less than 1 minute and with the mobile application inaccessible only for less than 15 seconds.
Authors: Diogo Vaz (INESC-ID / IST), David Matos (INESC-ID / IST), Miguel Pardal (INESC-ID / IST), Miguel Correia (INESC-ID / IST),
Hide Authors & Abstract

Show Authors & Abstract
11:10 - 12:20
My House, My Rules: A Private-by-Design Smart Home Platform

Smart home technology has gained widespread adoption. However, several instances of massive corporate surveillance and episodes of sensor data breaches have raised many privacy concerns amongst potential consumers. This paper presents PatrIoT, a private-by-design IoT platform for smart home environments. PatrIoT revisits the typical architecture of existing IoT platforms, and provides an alternative design where the home owner retains full ownership and control of smart device generated data. It leverages Intel SGX to prevent unauthorized access to the data by untrusted IoT cloud providers, and offers home owners an intuitive security abstraction named flowwall which allows them to specify easy-to-use policies for controlling sensitive sensor data flows within their smart homes. We have built and evaluated a PatrIoT prototype. Most of the participants in a field study considered PatrIoT to be easy to use, and the supported policies to be useful in protecting their privacy.
Authors: Igor Zavalyshyn (UCLouvain), Nuno Santos (INESC-ID), Ramin Sadre (UCLouvain), Axel Legay (UCLouvain),
Hide Authors & Abstract

Show Authors & Abstract

Lunch Break 12:20 - 12:50

Session 9: M-Health 12:50 - 14:40

12:50 - 14:40
AI-Fairness Towards Activity Recognition of Older Adults

Though, many existing researches focus on different WSN integration, signal processing and intelligent detection to recognize different activities, none of the existing works address the Artificial Intelligence (AI) fairness in activity recognition of older adults domain. We argue that AI-fairness towards detecting activities of older adults is different than to fairness for other protected attributes such as age, gender or race due to existence of diversity of same activity (walking using crutches, walker). However, the diversity also exists among the same older adults based on time and space as well. The above constraints limit the AI capabilities and causes unfair detection of daily activities that has potential impacts on healthcare interventions for older adults. We investigate, first of its kind, AI-fairness of activity recognition for older adults using a single wearable WSN sensor in presence of diverse disabilities. In this regard, we (i) employ signal processing and Bi-directional LSTM model to recognize diverse multi-label activities of older adults using single WSN; (ii) identify the existence of biases in activity recognition considering the age and functional ability as protected attributes; (iii) mitigate the biases by applying different bias mitigation techniques in different stages of machine learning; and (iv) experimentally evaluate proposed frameworks using older adults data collected from a real-time smart home system deployed in an older home.
Authors: Mohammad Arif Ul Alam (University of Massachusetts Lowell),
Hide Authors & Abstract

Show Authors & Abstract
12:50 - 14:40
Security Awareness of End-Users of Mobile Health Applications: An Empirical Study

Mobile systems offer portable and interactive computing – empowering users – to exploit a multitude of context-sensitive services, including mobile healthcare. Mobile health applications (i.e., mHealth apps) are revolutionizing the healthcare sector by enabling stakeholders to produce and consume healthcare services. A widespread adoption of mHealth technologies and rapid increase in mHealth apps entail a critical challenge, i.e., lack of security awareness by end-users regarding health-critical data. This paper presents an empirical study aimed at exploring the security awareness of end-users of mHealth apps. We collaborated with two mHealth providers in Saudi Arabia to gather data from 101 end-users. The results reveal that despite having the required knowledge, end-users lack appropriate behaviour , i.e., reluctance or lack of understanding to adopt security practices – compromising health-critical data with social, legal, and financial consequences. The results emphasize that mHealth providers should ensure security training of end-users (e.g., threat analysis workshops), promote best practices to enforce security (e.g., multi-step authentication), and adopt suitable mHealth apps (e.g., trade-offs for security vs usability). The study provides empirical evidence and a set of guidelines about security awareness of mHealth apps.
Authors: Bakheet Aljedaani (The University of Adelaide), Aakash Ahmad (College of Computer Science and Engineering, University of Ha'il, Saudi Arabia), Mansooreh Zahedi (The University of Adelaide), M. Ali Babar (The University of Adelaide),
Hide Authors & Abstract

Show Authors & Abstract
12:50 - 14:40
AutoCogniSys: IoT Assisted Context-Aware Automatic Cognitive Health Assessment

Cognitive impairment has become epidemic in older adult population. The recent advent of tiny wearable and ambient devices, a.k.a Internet of Things (IoT) provides ample platforms for continuous functional and cognitive health assessment of older adults. In this paper, we design, implement and evaluate \emph{AutoCogniSys}, a context-aware automated cognitive health assessment system, combining the sensing powers of wearable physiological (Electrodermal Activity, Photoplethysmography) and physical (Accelerometer, Object) sensors in conjunction with ambient sensors. We design appropriate signal processing and machine learning techniques, and develop an automatic cognitive health assessment system in a natural older adults living environment. We validate our approaches using two datasets: (i) a naturalistic sensor data streams related to Activities of Daily Living and mental arousal of 22 older adults recruited in a retirement community center, individually living in their own apartments using a customized inexpensive IoT system (IRB #HP-00064387) and (ii) a publicly available dataset for emotion detection. The performance of AutoCogniSys attests max. 93% of accuracy in assessing cognitive health of older adults.
Authors: Mohammad Arif Ul Alam (University of Massachusetts Lowell), Nirmalya Roy (University of Maryland Baltimore County), Sarah Holmes (University of Maryland Baltimore), Aryya Gangopadhyay (University of Maryland Baltimore County), Elizabeth Galik (University of Maryland Baltimore),
Hide Authors & Abstract

Show Authors & Abstract
12:50 - 14:40
TinnituSense: a Mobile Electroencephalography (EEG) Smartphone App for Tinnitus Research

Tinnitus is a disorder or symptom that causes phantom noise sensation in the ears without presence of any external sound source. Tinnitus is understood as a problem caused by underlying damage in the inner-ear. However, recent studies have shown that tinnitus is also influenced by complexities in non-auditory brain areas. Among different brain-imaging techniques, mobile Electroencephalography (EEG) can be a viable solution in better understanding the influencing factors in the brain causing tinnitus, but real-time analysis of EEG in real-world environments is faced by unique challenges and limitations. We present the first pure smartphone-based solution to acquire and analyze EEG data in real time and in everyday settings, as well as in any other scenario which does not allow large setups. More specifically, we propose TinnituSense a smartphone app for EEG recordings and visualization, and evaluate this app to claim the feasibility of our approach. On one hand, the proposed approach will open the opportunities to perform brain-imaging in real-world environment. On the other hand, the developed app will allow tinnitus researchers to collect evidence for new facts regarding tinnitus with the help of ambulatory brain-imaging data.
Authors: Muntazir Mehdi (Institute of Distributed Systems, Ulm University), Florian Diemer (Institute of Distributed Systems, Ulm University), Lukas Hennig (Institute of Distributed Systems, Ulm University), Albi Dode (Institute of Databases and Information Systems, Ulm University), Ruediger Pryss (Institute of Clinical Epidemiology and Biometry, University of Wuerzburg), Winfried Schlee (Clinic and Policlinic for Psychiatry and Psychotherapy, Regensburg, Germany), Manfred Reichert (Institute of Databases and Information Systems, Ulm University), Franz Hauck (Institute of Distributed Systems, Ulm University),
Hide Authors & Abstract

Show Authors & Abstract
12:50 - 14:40
MobDL: A Framework for Profiling Deep Learning Models: A case study using Mobile Digital Health Applications

Smart mobile devices coupled with the Internet of Things (IoT) and Artificial Intelligence (AI) have emerged as a key enabler of modern digital health applications. While cloud computing is now a well established paradigm for analysing IoT captured data in mobile health applications, on-board analysis of data using AI approaches such as Deep Learning (DL) is gaining significant momentum. This is driven primarily by advances in on-board resources enabling modern mobile devices to execute complex DL models, while also offering improved response time and accuracy for rapid decision-making, and enhanced user privacy. While the number of mobile digital health applications that use IoT and DL is increasing, progress is currently impeded by a lack of framework for profiling and evaluating the performance of DL models on mobile devices. To this end, we propose MobDL, a framework for profiling and evaluating DL models running on smart mobile devices. We present an architecture of this framework and devise a novel evaluation methodology for conducting quantitative comparisons of various DL models running on mobile devices. Three diverse digital health applications that use heterogeneous data (e.g. image, time series) are introduced. We conduct extensive experimental evaluations using several DL models that have been developed using the data sets obtained for the three digital health applications to validate the efficacy of the proposed MobDL framework.
Authors: Abdur Rahim Mohammad Forkan (Swinburne University of Technology), Prem Prakash Jayaraman (Swinburne University of Technology), Rohit Kaul (Swinburne University of Technology), Yuxin Zhang (Monash University), Chris McCarthy (Swinburne University of Technology), Pari Delir Haghighi (Monash University), Rajiv Ranjan (Newcastle University, UK),
Hide Authors & Abstract

Show Authors & Abstract

Closing Message 14:40 - 14:50