Refine
Year of publication
Document Type
- Conference Proceeding (145) (remove)
Conference Type
- Konferenzartikel (143)
- Konferenz-Abstract (1)
- Sonstiges (1)
Is part of the Bibliography
- yes (145)
Keywords
- Eingebettetes System (5)
- Kommunikation (5)
- Messung (3)
- Security (3)
- Sensortechnik (3)
- cryptography (3)
- Applikation (2)
- Energieversorgung (2)
- Internet der Dinge (2)
- Internet of Things (2)
Institute
- ivESK - Institut für verlässliche Embedded Systems und Kommunikationselektronik (101)
- Fakultät Elektrotechnik und Informationstechnik (E+I) (bis 03/2019) (88)
- Fakultät Elektrotechnik, Medizintechnik und Informatik (EMI) (ab 04/2019) (54)
- Fakultät Maschinenbau und Verfahrenstechnik (M+V) (2)
- Fakultät Medien (M) (ab 22.04.2021) (1)
Open Access
- Closed Access (81)
- Closed (31)
- Open Access (28)
- Bronze (9)
- Diamond (4)
- Gold (1)
Training deep neural networks using backpropagation is very memory and computationally intensive. This makes it difficult to run on-device learning or fine-tune neural networks on tiny, embedded devices such as low-power micro-controller units (MCUs). Sparse backpropagation algorithms try to reduce the computational load of on-device learning by training only a subset of the weights and biases. Existing approaches use a static number of weights to train. A poor choice of this so-called backpropagation ratio limits either the computational gain or can lead to severe accuracy losses. In this paper we present TinyProp, the first sparse backpropagation method that dynamically adapts the back-propagation ratio during on-device training for each training step. TinyProp induces a small calculation overhead to sort the elements of the gradient, which does not significantly impact the computational gains. TinyProp works particularly well on fine-tuning trained networks on MCUs, which is a typical use case for embedded applications. For typical datasets from three datasets MNIST, DCASE2020 and CIFAR10, we are 5 times faster compared to non-sparse training with an accuracy loss of on average 1%. On average, TinyProp is 2.9 times faster than existing, static sparse backpropagation algorithms and the accuracy loss is reduced on average by 6 % compared to a typical static setting of the back-propagation ratio.
Security in IT systems, particularly in embedded devices like Cyber Physical Systems (CPSs), has become an important matter of concern as it is the prerequisite for ensuring privacy and safety. Among a multitude of existing security measures, the Transport Layer Security (TLS) protocol family offers mature and standardized means for establishing secure communication channels over insecure transport media. In the context of classical IT infrastructure, its security with regard to protocol and implementation attacks has been subject to extensive research. As TLS protocols find their way into embedded environments, we consider the security and robustness of implementations of these protocols specifically in the light of the peculiarities of embedded systems. We present an approach for systematically checking the security and robustness of such implementations using fuzzing techniques and differential testing. In spite of its origin in testing TLS implementations we expect our approach to likewise be applicable to implementations of other cryptographic protocols with moderate efforts.
Die immer weitreichenderen Anwendungen des Smart Metering und des Smart Grid stellen immer höhere Anforderungen an Kommunikationstechnologien, die die Zielkonflikte aus Echtzeitfähige, Stabilität, Kosten und Energieeffizienz möglichst anwendungsoptimiert und auf einem immer höheren Niveau lösen. Insbesondere im Bereich der so genannten Primärkommunikation zwischen einem Sensor- oder Aktorknoten und einem Datensammler mit Gatewayfunktionalität konnten in den vergangenen Jahren wesentliche Fortschritte erzielt werden. Zu nennen sind hierbei insbesondere die Aktivitäten der ZigBee Alliance rund um den offenen Spezifikationsprozess des ZigBee Smart Energy Profiles (SEP) und der OMS-Gruppe beim ZVEI, die auf dem Wireless M-Bus nach EN13757-4 aufbauen, der sich seinerseits lebhaft und zielgerichtet weiter entwickelt. Der Beitrag diskutiert die vorhandenen Einschränkungen und die verfügbaren Lösungsansätze. Er illustriert diese anhand einiger öffentlich geförderter Projekte, an denen das Team des Autors beteiligt ist.
This paper presents the elements and the results from the European research project inCASA (Integrated Network for Completely Assisted Senior Citizen’s Autonomy), which designed and implemented a seamless integration of heterogeneous systems and network protocols for regionally distributed telecare and telehealth applications. The integration includes a multitude of physical interface, the transcoding of data models using embedded middleware, and a backend system with open interfaces. The implementation was verified in field tests in five European countries.
Immer mehr Anwendungen der Heim- und der Gebäudeautomatisierung werden vernetzt, weil damit erweiterte Funktionen ermöglicht oder Kosten gespart werden können. Dabei führt eine Reihe von Aspekten zu einem erhöhten Risiko für diese vernetzten Systeme. Gegenwärtig arbeiten verschiedene Gruppen an Sicherheitslösungen für die vernetzte Heim- und Gebäudeautomatisierung. Der Beitrag gibt einen Überblick über diese Aktivitäten und zeigt die wesentlichen Entwicklungsrichtungen auf.
Efficient, secure and reliable communication is a major precondition for powerful applications in smart metering and smart grid. This especially holds true for the so called primary communication in the Local Metrological Network (LMN) between meter and data collector, as the LMN comes with the most stringent requirements with regard to cost, range, as well as bandwidth and energy efficiency. Until today, LMN field tests are operated all over the world. In these installations, however, energy autarkic systems play a marginal role. This contribution describes the results of the framework 7 (FP 7) WiMBex project (“Remote wireless water meter reading solution based on the EN 13757 standard, providing high autonomy, interoperability and range”). In this project an energy autarkic water meter was developed and tested, which follows the specification of the Wireless M-Bus protocol (EN 13757). The complete system development covers the PCB with the RF transceiver and the microcontroller, the energy converter and storage, and the software with the protocol. This contribution especially concentrates on the design, the development and the verification of the routing protocol. The routing protocol is based on the Q mode of EN13757-5 (Wireless M-Bus) and was extended by an additional energy state related parameter. This extension is orthogonal to the existing protocol and considers both the charge level and the charge characteristics (rate of occurrences, intensity). The software was implemented in NesC under the operating system TinyOS. The system was verified in an automated test bed and in field tests in UK and Ireland.