Logo des Repositoriums

Auflistung nach Schlagwort "Graph Theory"

1 - 4 von 4
  • Konferenzbeitrag
    Emission-Reducing Vehicle Routing in Food Logistics
    (INFORMATIK 2023 - Designing Futures: Zukünfte gestalten, 2023) Wolf, Armin; Cuno, Silke
    A web service for emission-reducing vehicle routing in food logistics is presented. This service computes emission-reduced round trips for the transportation of food products from (and to) a depot to (and from) regional locations like food stores or food hubs. Further, this service considers the travelling times between locations and the produced greenhouse gas emissions of the according drives as well as the time windows needed for pickup and delivery. The loading capacities of the vehicles and an efficient “last-in-first-out” packaging order of the vehicles are respected, with the consequence that goods picked-up last should be delivered and unloaded first. This is to avoid unnecessary work.
  • Textdokument
    Machine learning for optimizing disposition and planning of vehicles with near real-time IoT events at scale
    (INFORMATIK 2020, 2021) Daemi-Ahwazi, Anusch; Rost, Daniel
    Cargo vehicles today are equipped with power saving IoT devices measuring various aspects of the vehicle and cargo itself. The real-time stream of IoT events from the vehicles are sending large amounts of data each day, which needs to be correlated with each other and existing data sources to generate business value. The algorithmic challenges for discussion are the handling of noisy data and fast correlation of the sensor data as well as software engineering challenges to ensure the system(s) are highly performant and maintainable over the next decades.
  • Konferenzbeitrag
    MANTRA: A Graph-based Unified Information Aggregation Foundation for Enhancing Cybersecurity Management in Critical Infrastructures
    (Open Identity Summit 2023, 2023) Fuxen, Philipp; Hackenberg, Rudolf; Heinl, Michael P.; Ross, Mirko; Roßnagel, Heiko; Schunck, Christian H.; Yahalom, Raphael
    The digitization of almost all sectors of life and the quickly growing complexity of interrelationships between actors in this digital world leads to a dramatically increasing attack surface regarding both direct and also indirect attacks over the supply chain. These supply chain attacks can have different characters, e.g., vulnerabilities and backdoors in hardware and software, illegitimate access by compromised service providers, or trust relationships to suppliers and customers exploited in the course of business email compromise. To address this challenge and create visibility along these supply chains, threat-related data needs to be rapidly exchanged and correlated over organizational borders. The publicly funded project MANTRA is meant to create a secure and resilient framework for real-time exchange of cyberattack patterns and automated, contextualized risk management. The novel graph-based approach provides benefits for automation regarding cybersecurity management, especially when it comes to prioriization of measures for risk reduction and during active defense against cyberattacks. In this paper, we outline MANTRA’s scope, objectives, envisioned scientific approach, and challenges.
  • Textdokument
    The problem of packing modification-disjoint P3 – an overview and an improved heuristic approach
    (SKILL 2022, 2022) Dirks, Jona; Gerhard, Enna
    The problem of packing modification-disjoint P₃ – an overview and an improved heuristic approach We consider the problem of packing modification-disjoint induced P₃. This has not been fully researched so far. Induced P₃ are especially relevant to solve the cluster editing problem. We provide an overview and new insights for locating modification-disjoint P₃ packing within the complexity hierarchy. Accordingly, we further look into conflict graphs. In response to our theoretical results, we create a significantly improved heuristic based on the approach of Spinner (2019). We then analyze its efficiency empirically on a selection of generated and public datasets. Our results show that it is either better than existing heuristics when comparing solution size and running time.