RonPub -- Research Online Publishing
Not a member yet
199 research outputs found
Sort by
Towards Adaptive Actors for Scalable IoT Applications at the Edge
Full text linkTraditional device-cloud architectures are not scalable to the size of future IoT deployments. While edge and fog-computing principles seem like a tangible solution, they increase the programming effort of IoT systems, do not provide the same elasticity guarantees as the cloud and are of much greater hardware heterogeneity. Future IoT applications will be highly distributed and place their computational tasks on any combination of end-devices (sensor nodes, smartphones, drones), edge and cloud resources in order to achieve their application goals. These complex distributed systems require a programming model that allows developers to implement their applications in a simple way (i.e., focus on the application logic) and an execution framework that runs these applications resiliently with a high resource efficiency, while maximizing application utility. Towards such distributed execution runtime, we propose Nandu, an actor based system that adapts and migrates tasks dynamically using developer provided hints as seed information. Nandu allows developers to focus on sequential application logic and transforms their application into distributed, adaptive actors. The resulting actors support fine-grained entry points for the execution environment. These entry points allow local schedulers to adapt actors seamlessly to the current context, while optimizing the overall application utility according to developer provided requirements
Anonymous Shopping in the Internet by Separation of Data
Full text linkWhenever clients shop in the Internet, they provide identifying data of themselves to parties like the webshop, shipper and payment system. These identifying data merged with their shopping history might be misused for targeted advertisement up to possible manipulations of the clients. The data also contains credit card or bank account numbers, which may be used for unauthorized money transactions by the involved parties or by criminals hacking the parties' computing infrastructure. In order to minimize these risks, we propose an approach for anonymous shopping by separation of data. We argue for the feasibility of our approach by discussing important operations like simple reclamation cases and criminal investigations
Editorial of the Workshop on Very Large Internet of Things (VLIoT 2018)
Full text linkThe 2nd "Very Large Internet of Things" (VLIoT) workshop in conjunction with the 44th International Conference on Very Large Data Bases (VLDB) taking place in Rio de Janeiro, Brazil in 2018 is a forum for all researchers in the area of Internet of Things especially interested in related data management issues. This editorial of a special issue containing the workshop's papers provides an overview over the aims and scope of the workshop and the review procedure. Furthermore, we determine and shortly analyze a statistics of the topics addressed by the accepted papers
Identifying Malicious Nodes in Multihop IoT Networks using Dual Link Technologies and Unsupervised Learning
Full text linkPacket manipulation attack is one of the challenging threats in cyber-physical systems (CPSs) and Internet of Things (IoT), where information packets are corrupted during transmission by compromised devices. These attacks consume network resources, result in delays in decision making, and could potentially lead to triggering wrong actions that disrupt an overall system's operation. Such malicious attacks as well as unintentional faults are difficult to locate/identify in a large-scale mesh-like multihop network, which is the typical topology suggested by most IoT standards. In this paper, first, we propose a novel network architecture that utilizes powerful nodes that can support two distinct communication link technologies for identification of malicious networked devices (with typical singlelink technology). Such powerful nodes equipped with dual-link technologies can reveal hidden information within meshed connections that is hard to otherwise detect. By applying machine intelligence at the dual-link nodes, malicious networked devices in an IoT network can be accurately identified. Second, we propose two techniques based on unsupervised machine learning, namely hard detection and soft detection, that enable dual-link nodes to identify malicious networked devices. Our techniques exploit network diversity as well as the statistical information computed by dual-link nodes to identify the trustworthiness of resource-constrained devices. Simulation results show that the detection accuracy of our algorithms is superior to the conventional watchdog scheme, where nodes passively listen to neighboring transmissions to detect corrupted packets. The results also show that as the density of the dual-link nodes increases, the detection accuracy improves and the false alarm rate decreases
A Lightweight Network-Controlled Power Strip for Low-Cost Cluster Systems
Full text linkLow-cost clusters are not equipped with costly, sophisticated tools and cannot be controlled remotely. This work aims at addressing this issue and develops a lightweight network-controlled power strip, which enables administrators to monitor the cluster and perform operation via remote. The power strip is controlled via a web interface and a RESTful web service, which are implemented with the programming language Python and the web framework Flask. The solution is inexpensive and easy to implement and use. In this paper, we describe in detail the development and construction of the prototype of the solution and discuss its purchase cost and power consumption
Operation of Modular Smart Grid Applications Interacting through a Distributed Middleware
Full text linkIoT-functionality can broaden the scope of distribution system automation in terms of functionality and communication. However, it also poses risks regarding resource consumption and security. This article presents a field approved IoT-enabled smart grid middleware, which allows for flexible deployment and management of applications within smart grid operation. In the first part of the work, the resource consumption of the middleware is analyzed and current memory bottlenecks are identified. The bottlenecks can be resolved by introducing a new entity that allows to dynamically load multiple applications within one JVM. The performance was experimentally tested and the results suggest that its application can significantly reduce the applications' memory footprint on the physical device. The second part of the study identifies and discusses potential security threats, with a focus on attacks stemming from malicious software applications within the framework. In order to prevent such attacks a proxy based prevention mechanism is developed and demonstrated
Special Issue on High-Level Declarative Stream Processing
Full text linkStream processing as an information processing paradigm has been investigated by various research communities within computer science and appears in various applications: realtime analytics, online machine learning, continuous computation, ETL operations, and more. The special issue on "High-Level Declarative Stream Processing" investigates the declarative aspects of stream processing, a topic of undergoing intense study. It is published in the Open Journal of Web Technologies (OJWT) (www.ronpub.com/ojwt). This editorial provides an overview over the aims and the scope of the special issue and the accepted papers
Multi-Shot Stream Reasoning in Answer Set Programming: A Preliminary Report
Full text linkIn the past, we presented a first approach for stream reasoning using Answer Set Programming (ASP). At the time, we implemented an exhaustive wrapper for our underlying ASP system, clingo, to enable reasoning over continuous data streams. Nowadays, clingo natively supports multi-shot solving: a technique for processing continuously changing logic programs. In the context of stream reasoning, this allows us to directly implement seamless sliding-window-based reasoning over emerging data. In this paper, we hence present an exhaustive update to our stream reasoning approach that leverages multi-shot solving. We describe the implementation of the stream reasoner's architecture, and illustrate its workflow via job shop scheduling as a running example
Ontology-Based Data Access to Big Data
Full text linkRecent approaches to ontology-based data access (OBDA) have extended the focus from relational database systems to other types of backends such as cluster frameworks in order to cope with the four Vs associated with big data: volume, veracity, variety and velocity (stream processing). The abstraction that an ontology provides is a benefit from the enduser point of view, but it represents a challenge for developers because high-level queries must be transformed into queries executable on the backend level. In this paper, we discuss and evaluate an OBDA system that uses STARQL (Streaming and Temporal ontology Access with a Reasoning-based Query Language), as a high-level query language to access data stored in a SPARK cluster framework. The development of the STARQL-SPARK engine show that there is a need to provide a homogeneous interface to access both static and temporal as well as streaming data because cluster frameworks usually lack such an interface. The experimental evaluation shows that building a scalable OBDA system that runs with SPARK is more than plug-and-play as one needs to know quite well the data formats and the data organisation in the cluster framework
