Program in Detail

The program is still subject to change.

Invited Speakers

Have a look at the details on our invited speakers.

Session 1: Information Security

Monday, September 11, 2023, 10:30 a.m.

Michael Hofmeier and Wolfgang HommelEnabling the JSON Web Signature Format to Support Complex and Identity-oriented Non-web Processes
Abstract: This paper examines what rules or extensions have to be applied to the JSON Web Signature format so that it can be used universally in identity-driven non-web processes where identities exchange data, documents or attestations in a decentralized manner but do not know each other. For this purpose, the format and the related process must fulfill certain requirements such as identifiability and support for multiple signatures at possibly different points in time. The German T-prescription was selected as the application scenario, since it involves special requirements for signatures, data protection and data transfer. For this scenario, all the necessary applications and libraries are implemented and the process is run through and analyzed.
Günter Fahrnberger

Bloom Filter-Based Realtime Risk Monitoring of SSH Brute Force Attacks

Abstract: Publicly shared hosts on the Internet appeal to well-behaving and mischievous clients in equal measure. Miscreants rapidly enumerate and attempt to capitalize on the hosts’ open ports. Specially CommandLine Interfaces (CLIs), such as Secure Shell (SSH), with odds of conquering unlimited permissions on such hosts allure culprits into conducting brute force attacks. Responsible personnel should not unclose SSH ports to the Internet unless inevitable. If opened, installable precautions, like anti-hammering, Intrusion Detection Systems (IDSs), or Intrusion Prevention Systems (IPSs), simply proffer protection with a rash of descriptive attack statistics. Beyond that, pertinent research assists with qualitative pattern-based realtime risk monitoring of SSH brute force attacks. This disquisition appraises such offenses’ danger more accurately than preceding methods with the support of a modified Bloom filter and attests the attained superiority over them.

Luis Almeida, Brayan Fernández, Daliana Zambrano, Anthony Almachi, Hilton Pillajo and Sang Guun Yoo

A Complete One-Time Passwords (OTP) Solution Using Microservices: A Theoretical and Practical Approach
Abstract: The objective of this paper is to share the knowledge required for developing a One-Time Password (OTP) system and the practical experience of developing a real one using microservices and different programming languages, remembering that the OTP is the most popular mechanism to carry out a two-factor authentication process. To achieve this purpose, an incremental iterative methodology was used that allowed the prototype to be implemented in different parts. The developed prototype was designed to work in mobile applications with iOS and Android operating systems. In this work, different types of OTP algorithms were used, such as HMAC-based One-Time Password, Time-based One-Time Password and True Random Number Generator. Additionally, for the development of the prototype, various tools and frameworks such as Flask, React Native and Flutter were used, which allowed the development of application components in an agile and efficient manner. The combination of these programming languages and tools resulted in a more efficient and effective implementation of the OTP generator prototype. Once the proposed system was developed, different types of tests were carried out to verify its optimal and efficient functionality.

Back to Program Overview, Top

Session 2: Environmental Protection

Monday, September 11, 2023, 14:30

Mohamed-Lamine Benzagouta, Hasnaa Aniss, Hacène Fouchal and Nour-Eddin El-Faouzi

Impact of Speed Limitation of Urban Road Traffic

Abstract: As vehicular communications continue to develop and evolve, data from probe connected vehicles can prove to be useful so to observe traffic dynamics more precisely. Cooperative Awareness Messages (CAM) give us detailed information about the location, velocity and heading of probe vehicles, whereas data from the Green Light Optimal Speed Advisory (GLOSA) give us information about the traffic lights at intersections such as the current phase, time to next phase and the speed advice that allows a driver to avoid a stop at a red light. In this work, we use data that is extracted from a smartphone application that simulates vehicular communications. We match the GLOSA records with the CAMs in order to observe the behavior of vehicles in the segments prior to an intersection. We characterize four different patterns of speed profiles for vehicles in segments with a traffic light intersection.
Jonas-Dario Troles, Richard Nieding, Sonia Valeska Simons and Ute SchmidTask Planning Support for Arborists and Foresters: Comparing Deep Learning Approaches for Tree Inventory and Tree Vitality Assessment Based on UAV-Data
Abstract: Climate crisis and correlating prolonged, more intense periods of drought threaten tree health in cities and forests. In consequence, arborists and foresters suer from increasing workloads and, in the best case, a consistent but often declining workforce. To optimise workflows and increase productivity, we propose a novel open-source end-to-end approach that generates helpful information and improves task planning of those who care for trees in and around cities. Our approach is based on RGB and multispectral UAV data, which is used to create tree inventories of city parks and forests and to deduce tree vitality assessments through statistical indices and Deep Learning. Due to EU restrictions regarding flying drones in urban areas, we will also use multispectral satellite data and fifteen soil moisture sensors to extend our tree vitality-related basis of data. Furthermore, Bamberg already has a georeferenced tree cadastre of around 15,000 solitary trees in the city area, which is also used to generate helpful information. All mentioned data is then joined and visualised in an interactive web application allowing arborists and foresters to generate individual and flexible evaluations, thereby improving daily task planning.
Karl Seidenfad, Maximilian Greiner, Jan Biermann and Ulrike LechnerCarbonEdge: Collaborative Blockchain-based Monitoring, Reporting and Verification of Greenhouse Gas Emissions on the Edge
Abstract: Decarbonization calls for efficient and trustworthy monitoring, reporting, and verification (MRV) of greenhouse gas (GHG) emissions. This article proposes a collaborative approach and exemplifies a concrete scenario: the biogas plant scenario. We propose a consortium blockchain as platform and aim to reduce costs and efforts by collaboration and automation in up-scaling GHG management. We present a demonstrator which employs an industry-grade radio frequency identification (RFID) security approach to manage certificate material, a low-code interface for interaction, and the operation of Hyperledger Fabric on industrial edge devices.

Back to Program Overview, Top

Session 3: Text Analysis

Thuesday, September 12, 2023, 9:00 a.m.

Jörg RothSupport for Fictional Story Writing and Copy Editing
Abstract: Fictional writing is the art of creating stories that are not based on real events or people. In recent years, book projects for smaller communities have become popular, not least due to the success of e-book platforms. In the context of story wringing, copy editing is an important step in the writing process as it ensures the overall quality concerning grammar, spelling and style. Currently, this step usually is executed by costly specialists. This is a problem for writers who are starting their careers. This paper presents a tool platform that simplifies the process of copy editing. Authors can execute first check steps themselves. The amount of further iterations with a commercial lector is reduced or may even be omitted. Our tool is able to apply even complex style rules that are based on deeper grammar analyses of texts.
Florian Würmseer, Stefan Wallentowitz and Markus FriedrichAutoNLP: A System for Automated Market Research Using Natural Language Processing and Flow-based Programming
Abstract: This paper introduces a novel architecture for an automated market research system that utilizes the Flow-based Programming (FBP) paradigm. The described system allows users to select topics of interest, and then automatically collects website content related to these topics. The system also offers different search modes for searching the collected text corpus, including stem-based and semantic search. These search modes utilize Natural Language Processing (NLP) techniques. The design of the system was developed to meet specific requirements and underwent a thorough assessment. We analyzed the system’s runtime efficiency with and without the use of flow-based logic and found scalability issues in the Python library used. Finally, we conducted a user study that evaluated the system’s usability, showing that the system can be used without requiring extensive training.

Hanan Batarfi, Olaa Alsaedi, Arwa Wali and Amani Jamal

Impact of Data Augmentation on Hate Speech Detection
Abstract: With the increase of social media platforms such as Facebook, Twitter, and YouTube, individuals from diverse cultures and societal backgrounds can communicate and express their viewpoints on several aspects of daily life. However, due to the differences in these cultures, along with the freedom of expression, hateful and offensive speech has increased and spread on these platforms. The detection of hate speech has significantly increased the interest of researchers in natural language processing (NLP). The OSACT5 shared task provides a new dataset that aims to detect the offensive language in addition to identifying the type of hate speech on Arabic social media. However, the available dataset is unbalanced, which leads to low performance, especially in the F1 score. Therefore, in this paper, we focused on overcoming such a problem by augmenting the text data. We ne-tuned and evaluated various pre-trained deep learning models in addition to augmenting the data to achieve the best performance. We observed that data augmentation increases the F1 score. After fine-tuning the QARiB model and augmenting the data we achieved the best F1 score of 0.49.

Back to Program Overview, Top

Session 4: Quantum Computing

Thuesday, September 12, 2023, 11:00 a.m.

Niels Neumann, Ward van der Schoot and Thom SijpesteijnQuantum Cloud Computing from a User Perspective
Abstract: Quantum computing is a rapidly progressing field: quantum computers are becoming more powerful and an increasing number of functionalities are offered via various quantum platforms and quantum software packages. Current quantum computers can be used almost exclusively via cloud services. It is expected that this will remain the case, at least in the near-term future. For successful adoption of quantum computing by the market, quantum cloud services should be user-centric. To that end, we explore quantum cloud computing from a user perspective. We describe a standardised overview of quantum cloud computing as a whole to create a common language to strengthen research and collaboration on quantum cloud computing. Specifically, we identify different types, information flows and relevant user functionalities of quantum cloud computing, based on their counterparts in classical cloud computing. Combined, this gives an overview of quantum cloud computing for the best user experience, paving the way towards user-centric quantum cloud computing.
Thom Sijpesteijn and Frank PhillipsonQuantum Approaches for Medoid Clustering
Abstract: The k-medoids problem is an important problem in data clustering, which aims to partition a set of data points into k clusters, where each cluster is represented by a medoid, i.e., a data point that is the most centrally located in the cluster. Quantum annealing might be helpful in finding the solution to this problem faster. In this paper we compare three approaches for using the quantum annealer and QUBO-formulations to solve the k-medoids problem. The first approach revolves around a QUBO that encodes the problem as a whole. This approach turns out not to scale well for bigger problem sizes. The QUBO in the second approach comes from the literature and solves only the problem of finding medoids: assigning the datapoints to clusters requires an additional step. The QUBO formulation in the third approach is the same as in the second, but with different penalty parameters. We show that the second and third approaches scale better in terms of complexity than the first approach. However, the original penalty parameters in approach 2 (i.e. those suggested in the literature) do not work well for bigger instances. Taking different parameters makes this approach much better in performance.
Stan G. van der Linde, Ward van der Schoot and Frank PhillipsonEfficient Quantum Solution for the Constrained Tactical Capacity Problem for Distributed Electricity Generation
Abstract: With the transition to sustainable energy sources and devices, the demand for and supply of energy increases ever more. With energy grids struggling to keep up with this increase, we need to ensure that supply and demand are correctly matched. The Tactical Capacity Problem tackles this issue by choosing the optimal location of sustainable power sources to minimise the total energy loss. We extend an existing quantum approach of solving this problem in two ways. Firstly, we extend the problem to include capacity constraints, resulting in the Constrained Tactical Capacity Problem. Secondly, we propose two ways of optimising the performance of the resulting model via variable reduction. These optimisations are supported by numerical results obtained on both classical and quantum solvers.

Back to Program Overview, Top

Session 5: Internet of Things

Wednesday, September 13, 2023, 9:00

Mohammad R.H. Mullick, Marcel Großmann and Udo R. KriegerA Feasibility Study of a Lightweight Fog Computing Architecture Integrating Blockchain Technology for Smart E-Health Applications
Abstract: Today, a variety of advanced Internet-of-Things applications like data collection apps for ambient assisted living services supporting elder patients or for the surveillance of smart homes are developed everywhere. Regarding such application scenarios and the requirement to securely disseminate the collected data, we present a fog computing architecture that integrates blockchain technology for privacy aware data sharing. The latter fog-blockchain prototype is based on a private, permissioned blockchain paradigm. We discuss a lightweight implementation of our proposed integrated fog node architecture using Hyperledger Fabric as blockchain with a peer-to-peer network among cheap single-board-computers as basic data collecting peers. Furthermore, we investigate the computing and network performance indices of our prototypical lightweight fog computing architecture.
Sabrina Hölzer, Lucie Schmidt, Wesley Preßler, Antonio Schulz and Christian ErfurthTowards Designing a User-Centered Local Community Platform to Foster Social Cohesion in a Multi-Generational Smart Community
Abstract: In recent years, the emergence of smart cities and other related initiatives has prompted a growing interest in the role of digitalization in the housing sector. Housing cooperatives are increasingly exploring new and innovative concepts of community living and social connectedness within neighborhoods. In the context of the research project "Multi-Generation Smart Community" (mGeSCo) we are investigating and testing digitization in various dimensions using a living lab approach in the "Smart Quarter" Jena-Lobeda, which is currently home to 240 residents. In cooperation with different stakeholders, network partners and residents, interdisciplinary solutions are being developed and explored in the dimensions of work, living, housing and caring. The neighborhood residents can obtain benefits from a diverse range of digital amenities. In order to enhance acceptance and effectiveness while improving the well-being of the community, both analog and digital methods of participation and communication are being integrated. For example, a community platform designed for a neighborhood can improve well-being by fostering community, social support, trust, engagement, and comfort with smart technology in homes. The paper aims to provide insights in the designing process, information on challenges and peculiarities of the neighborhood, as well as to form a preliminary approach for a user-centered design of a neighborhood platform. While the overall conceptualization will involve additional aspects (such as user experience, interface design, communication and content strategy, analytics, and security), the initial step will focus solely on user needs and the associated features and functions of the platform.
Marcel Großmann, Lukas Klinger, Vanessa Krolikowsky and Chandan SarkarEfficient Internet of Things Surveillance Systems in Edge Computing Environments Accessible via Web Based Video Transmissions over Low-Cost Hardware
Abstract: Video surveillance plays an important role in society, even though current systems are expensive, proprietary, and lack the portability to be useful in small-scale scenarios. In contrast, cheap Single Board Computers (SBCs) are easily deployed and integrated into existing Internet of Things networks, where sensor nodes can trigger various actors. Besides proprietary video transmission tools, the WebRTC framework enables access to Peer-to-Peer communications by all commonly available web browsers. By connecting these two paradigms, sensor nodes can detect predefined events and notify users with an announcement of an easily accessible video stream recorded by a cost eficient camera node. We are exploiting the video streaming capabilities of a popular SBC model, the Raspberry Pi, and evaluate the expected Quality of Experience in combination with the stream's resource utilization on the source node. The trial framework is made publicly available to conduct measurements on any upcoming hardware platform. Finally, we provide a prototype of a sensor-triggered video surveillance system based on container virtualization. Any user interacts with it by a state-of-the-art browser and every system administrator easily initiates the open-sourced system by running our micro-service stack.

Back to Program Overview, Top 

Session 6: Short Papers

Wednesdy, September 13, 2023, 11:30 a.m.

Chérifa Boucetta, Laurent Hussenet and Michel HerbinImproved Euclidean Distance in the K Nearest Neighbors Method
Abstract: The KNN algorithm is one of the most famous algorithms in data mining. It consists in calculating the distance between a query and all the data in the reference set. In this paper, we present an approach to standardize variables that avoids making assumptions about the presence of outliers or the number of classes. Our method involves computing the ranks of values within the dataset for each variable and using these ranks to standardize the variables. We then calculate a dissimilarity index between the standardized data, called the Rank-Based Dissimilarity Index (RBDI), which we use instead of Euclidean distance to find the K nearest neighbors. Finally, we combine the Euclidean distance and the RBDI index taking into account the advantage of both dissimilarity indices. In essence, the Euclidean distance considers the Euclidean geometry of the data space while RBDI is not constrained by distance or geometry in data space. We evaluate our approach using multidimensional open datasets.
Udit Nath, Bhaskar Jyoti Medhi, Anuradha Deori, Maharaj Brahma, Mwnthai Narzary and Pranav Kumar SinghAutoBookFinder: A Case Study of Automated Book Rack Identification in Library through UGV
Abstract: The misplacement of books in libraries often results in wasted time when searching for a specific book on the wrong shelf. To address this issue, we created a novel, low-cost prototype Unmanned Ground Vehicle (UGV) equipped with a camera and Wi-Fi modules. The UGV wirelessly sends data to a local server, which uses image processing to determine the location of the book. A web user interface was created to enable users to easily locate the exact shelf of the desired book. This solution has not been previously explored and was tested in our Institute’s library. The adoption of UGVs greatly improved the overall user experience, transforming the manual and time-consuming process of book-finding into an automated and time-saving operation.
Moritz WeberAIoT Chances in Resource-Constrained Environments via Manual Pruning
Abstract: The challenges of AIoT devices in a resource-constrained environment are extensive. Even in areas where new AI opportunities are only being identified, an AIoT implementation can be more effective than traditional AI approaches. This paper proposes relevant solutions to address resource constraints, utilizing the pruning method. Furthermore, it presents a theoretical and strategic procedure for manual pruning, outlining how to navigate emerging challenges and identify potential limitations. To illustrate the efficacy of such applications and how they facilitate new AIoT use cases, the energy sector serves as an example scenario. The paper signify a foundation in tailoring AI applications to IoT devices with greater precision, establishing a solid foundation for future adaptations.
Frank PhillipsonEnd-to-End Quality of Service Management in the Physical Internet: a Digital Internet Inspired Multi-Domain Approach
Abstract: For the layer `System Level Functionality' of the Phyisical Internet, it is needed to estimate end-to-end performance characteristics of transportations that visit multiple logistic domains. This paper proposes an approach based on a Digital Internet functionality: a combination of a Service Level Agreement registry and a Quality of Service processor. Existing SLA-calculus gives tools for the QoS-processor to combine the SLA-parameters for all possible end-to-end paths and gives the QoS-processor the possibility to propose the best path given the required performance. A realistic implementation is proposed using a multi objective/constraint approach and a related communication form between the domain owner and the QoS Processor.

Back to Program Overview, Top