CoSDEO 2020

This Year’s Theme: Privacy and security in digital assistants

The 7th workshop on Context Systems Design, Evaluation and Optimization (CoSDEO2020) will be held in Austin, Texas, USA in conjunction with the 18th IEEE PerCom which is a premiere conference in Pervasive Computing and Communications (PerCom2020).

Program (2020-03-27) Now Online

09:15-09:30 Introduction
09:30-10:30 Keynote: Audio Security and Privacy (Andreas Nautsch, Eurecom)
Smart speakers, virtual assistants and interactive voice response systems successfully integrated as a part of our daily life. Intuitively, we use speech as a medium to seamlessly impart and exchange information for a plethora of services. Speech applications are found in personal use, health care and forensic sciences. Repurposing speech data is a potential threat to privacy infringement for speech being a very rich medium. Voice data, however, cannot only be exploited to characterize individuals but also to subvert system security: countermeasures of voice biometric systems are in high demand for detecting spoofing attacks on voice biometrics. An overview of the audio security and privacy at EURECOM is given on the ASVspoof challenge series; collaborations with legal and cybersecurity experts, and co-lead activities of the newly formed ISCA special interest group on security and privacy in speech communication.
10:30-12:00 Session 1 (Security)
Geunhyoung Kim; Im Young Jung: Encouraging Data Sharing for Safe Autonomous Driving
Autonomous or self-driving vehicles decide and enact various actions using state information without human interruption. More information can be gathered when autonomous vehicles can connect to and communicate with other vehicles and their environments, improving safe self-driving. However, information sharing requires data integrity, shared data verification, and incentive schemes to encourage information sharing. Therefore, this paper proposes an incentive scheme that also verifies data integrity to boost information sharing and hence improve autonomous driving safety.
Julian Graf, Katrin Neubauer, Sebastian Fischer, Rudolf Hackenberg: Architecture of an intelligent Intrusion Detection System for Smart Home
Increasing cyber-attacks on Internet of Things (IoT) environments are a growing problem of digitized households worldwide. The purpose of this study is to investigate how an intelligent Intrusion Detection System (iIDS) can provide more security in IoT networks through a novel architecture combining multiple classical and machine learning approaches. By combining classical security analysis methods and modern concepts of artificial intelligence, we increase the quality of attack detection and can therefore conduct dedicated attack suppression. The architectural image of the iIDS consists of different layers, which in parts achieve self-sufficient results. The results of the different modules are calculated by means of statement variables and evaluation techniques adapted for the individual module elements and subsequently combined by limit value considerations. The architecture image combines approaches for the analysis and processing of IoT network traffic and evaluates them to an aggregate score. From this result it can be determined whether the analyzed data indicates device misuse or attempted break-ins into the network. This study answers the questions whether a connection between classical and modern concepts for monitoring and analyzing IoT network traffic can be implemented meaningfully with a reliable architecture of an iIDS.
Invited talk (Lik-Hang Lee, Emerging role of Digital Assistant in Augmented Reality Driven Human-City Interaction)
This presentation will discuss the unique research issues and challenges in the intersection between augmented reality and smart cities. Designing interaction approaches for Mobile Augmented Reality (MAR) in city-wide urban scenarios are receiving rising attention. In our envisioned framework of Human-City Interaction, we foresee that MAR with complementary digital assistant will become critical enablers for immersive interaction in our smart cities. The proposed framework is on the basis of numerous works on MAR interaction design. We will discuss several illustrative examples to elaborate on the framework and accordingly highlight the importance of the digital assistant. The presentation will be concluded with future directions for Human-City Interaction.
12:00-13:30 Lunch Break
13:30-14:00 Position Talk (Stephan Sigg, Aalto University)
Position Talk: Ethically aligned voice user interfaces: Risks, Challenges and Opportunities. We witness the spread of voice user interfaces (VUIs)%through smartphones and specialized hardware, such as Siri, Alexa, or Cortana. As acoustical front-end for voice-based services, they take part in conversational social activities via a cornucopia of novel services and revolutionize customer interaction, human-device communication, as well as traditional human-to-human telecommunication applications~\cite{Porcheron2018}. Furthermore, additional services exploit emotional state, health status, gender, age or identity. For all of these, recent advances in speech communication, such as synchronized recordings from distributed microphones, accurate speaker localization and estimation of speech direction are bound to further raise the usefulness of these services. The integration of artificial intelligence (AI) during the processing is able to establish scene understanding capabilities and evolves VUIs into conversational bots, able to participate actively in human interaction.
14:00-14:30 Invited Talk (Xi Zheng, Software Engineering, AI and autonomous vehicles: Security assurance)
In this talk, I will first walk through some real industrial requirements and research challenges in autonomous vehicles. I will then talk about research works which can po- tentially solve these issues, mainly covering training, testing and anomaly detection for autonomous systems and driver behaviour detection. The talk will be broad but covering some state of the art interesting research questions and directions in autonomous vehicles safety and security assurance, and human vehicle inter- action which shall be suitable both for researchers and industry practitioners for in-depth enquiry and collaboration. Index Terms-Privacy and trust in voice user interfaces, Usable Security, Legislation, speech processing, Fuzzy cryptography
14:00-14:30 Breakout and collaboration

Call for Papers

The 7th CoSDEO-workshop aims to bring together researchers and practitioners working on the design, implementation, and evaluation of systems, algorithms or models for smart digital assistants.

The recent proliferation in smart digital assistants holds incredible potential for unprecedented services from (spontaneously networked) devices. Examples include directional audio reception, energy-gentle collaborative processing, unified user-interfaces spanning multiple collaborating devices, context-adaptive optimization of smart digital assistants behavior or device-capability-based intelligent task allocation. Most prominent examples for smart digital assistants are e.g. Alexa, Siri, Echo, Cortana, Google assistant, but also increasingly prominent chat-bots that are intergrated into pervasive applications. Such intelligent smart assistants exploit the recent avent of Artificial Intelligence for situation-dependent service composition conditioned on contextual stimuli from environmental perception. However, together with the technical challenges, privacy concerns, moralities and interface-design aspects of such intelligent machines need to be taken into account.

It is therefore the goal of this workshop to provide a publication and discussion platform for this growing community in which fundamental problems but also sophisticated approaches are presented and discussed.

Thus, we like to encourage practioners and scientists in all stages of their research, from first experiments to readily developed and evaluated systems, to submit their original work to allow a broad discussion with established field experts but also researchers relatively new to the field.

Even if work is in a very early stage, all submitted work must conform to typical scientific requirements; showing a good overview of the field and the specific area of the context of interest. If the paper is based on a technical system, the system should be described and discussed thoroughly.

Besides regular papers, we also encourage submission of visionary papers which must not describe completed research but contain ideas new to the field. These may be related to novel and convenient techniques for key generation and pairing, discussions on entropy and statistical properties of sensor-generated random sequences but also novel applications and designs of smart digital assistants. Topics of interest include but are not limited to:

  • Novel smart digital assistant systems
  • Context-based authentication and pairing
  • Attacks on smart digital assistants
  • Entropy and statistical analysis for keys generated in smart digital assistant systems
  • Novel sensor modalities for context-based authentication or security
  • Case studies in smart digital assistants
  • Distributed Audio-processing in smart digital assistants
  • Moralities and Ethics for smart digital assistants
  • Quantization approaches
  • Novel usable security/privacy mechanisms in the field of pervasive computing
  • Biometric and context-based authentication and pairing
  • Usability evaluations of new or existing security or privacy features
  • User studies in usable security and privacy
  • Lessons learned from the deployment and use of usable and pervasive security
  • User experience in smart digital assistants
  • Security schemes for sharply resource limited devices in Pervasive and IoT environments
  • Privacy and anonymization aspects
  • Novel interfaces fostering usable security and privacy
  • Entropy estimation and statistical analysis
  • Edge-support for usable security on resource limited pervasive devices
  • Lightweight security solutions
  • Detection and prevention of attacks
  • Authentication and access control
  • Trust in smart digital assistants
  • Orchestration and collaboration between devices

Important Dates

  • December 20, 2019: Paper and demo submission deadline
  • January 20, 2020: Notification of acceptance
  • January 31, 2020: Submission of camera-ready deadline

Submission Instructions

CoSDEO is single-blind. Papers may be no more than 6 pages in length (including references; Authors can purchase one additional page for the camera ready version). Papers in excess of page limits shall not be considered for review or publication. All papers must be typeset in double-column IEEE format using 10pt fonts on US letter paper, with all fonts embedded. The IEEE LaTeX and Microsoft Word templates, as well as related information, can be found at the IEEE Computer Society website. Workshop papers will be included and indexed in the IEEE digital libraries (Xplore).

Each accepted workshop paper requires a full PerCom registration (no registration is available for workshops only).

Submit Paper via EDAS


Workshop Organizers

Program Committee

  • Florian Alt, CODE Research Institute for Cyber Defense, Germany
  • Zinaida Benenson, FAU, Germany
  • Arne Bruesch, TU Braunschweig, Germany
  • Mads Graesboll Christensen, Aalborg, Denmark
  • Alexandra Craciun, XMOS, UK
  • Nick Evans, Eurecom, France
  • Sascha Fahl, Saarland University, Germany
  • Kassem Fawaz, University of Wisconsin, US
  • Thang Hoang, University of South Florida, US
  • Ansgar Kellner, TU Braunschweig, Germany
  • Le Ngu Nguyen, Aalto University, Finland
  • Markus Miettinen, TU Darmstadt, Germany
  • Andreas Nautsch, Eurecom, France
  • Petteri Nurmi, University of Helsinki, Finland
  • Dominik Schürmann, Confidential Technologies GmbH, Germany
  • Christian Stransky, Saarland University, Germany
  • Massimiliano Todisco, Eurecom, France
  • Isabel Transcoso, L2F, Portugal
  • Hien Truong, NEC Europe Ltd., UK
  • Dominik Wermke, Leibniz University Hannover, Germany
  • Yixin Zou, UMSI, US

Aalto University Fraunhofer AISEC PerCom 2020