Bernd Friedrichs

Bio: Bernd Friedrichs is an academic researcher. The author has contributed to research in topics: Adaptive filter & Message authentication code. The author has an hindex of 1, co-authored 4 publications receiving 2 citations.

Authentische und zuverlässige Mobilkommunikation für sicherheitsrelevante Anwendungen Teil II: Systemarchitektur und Einbettung in GSM

Abstract: A communication system for train control based on the GSM mobile radio standard was introduced in Part I. Cryptographical and error control mechanisms are required to ensure integrity and authentica-tion and have to be developed in accordance with the transmission standard, communication network and service definition. The message authentication code as the main security component for a single data block was analyzed and dimensioned by means of information theory in Part I. Methods for the security of the entire communication process including link management are discussed in Part II. Topics are the integration of the system components to achieve a generaI concept in the form of a layered architecture, an appropriate key management and cryptographical protocols as well as the adaptation to the GSM system

Analysis of Finite-Precision Adaptive Filters. Part I: Computation of the Residual Signal Variance

Abstract: Adaptive filters with variable coefficients have found many applications. The stochastic gradient least mean squares (LMS) algorithm is the most attractive adaptation scheme because of its computional simplicity. A fully digital implementation implies finite-precision adaptation algorithms with limited wordlength. For this condition exact theoretical investigations are currently available in literature only in part. This contribution presents new results for finite-precision adaptive filters. Part I investigates convergence in terms of the residual signal variance caused by coefficient dithering. In Part II [2], the probability density function of the residual signal is computed which allows the evaluation of error probabilities in digital receivers. Übersicht: Für adaptive Filter mit variablen Koeffizienten gibt es viele Anwendungen. Das stochastische Gradientenverfahren (LMS-Algorithmus) ist aufgrund der einfachen Rechenoperationen das attraktivste Adaptionsverfahren. Eine volldigitale Realisierung führt zu wertdiskreten Adaptionsalgorithmen mit beschränkten Wortlängen, deren exakte theoretische Untersuchung bisher nur in Ansätzen erfolgt ist. Es werden neue Ergebnisse zu wertdiskrctcn adaptiven Filtern präsentiert In Teil I wird das Konvergenzverhalten anhand der Varianz eines immer vorhandenen Restsignals untersucht. In Teil II [2] wird die Verteilungsdichtefunktion des Restsignals erfaßt, was die Berechnung von Fehlerwahrscheinlichkeiten in digitalen Empfängern ermöglicht. Für die Dokumentation: Adaptive Filter / Echokompensation / LMS-Algorithmus / Koeffizientenzittern / Quantisierung / Restfehler / Konvergenzgeschwindigkeit

Analysis of Finite-Precision Adaptive Filters Part II: Computation of the Residual Signal Distribution

Abstract: Adaptive filters are updated with the stochastic gradient least mean squares algorithm (LMS) in many applications. A generally remaining dithering of the coefficients causes a residual signal with a variance which was calculated in Part I [2]. If adaptive filters are used in digital receivers for data communication, additional information about the statistics of the residual signal would be useful for the calculation of the overall error probability. This contribution deals with the exact computation of the amplitude distribution of the residual signal, both for infiniteas finite-precision adaptive operation. As main result, the residual signal is established as Gaussian distributed even for adaptive filters with only one coefficient, if certain conditions are fulfilled. The widely used assumptions for the analysis of adaptive filtering are partly confirmed and otherwise partly refused. Übersicht: Fur adaptive Filter wird bei vielen Anwendungen das stochastische Gradientenverfahren (LMSAlgorithmus) zur Einstellung der Koeffizienten verwendet Ein grundsätzlich immer verbleibendes Zittern der Koeffizienten fuhrt zu einem Restsignal, dessen Varianz in Teil I [2] berechnet wurde. Beim Einsatz adaptiver Filter in digitalen Empfängern zur Datenübertragung sollten zur Berechnung der Fehlerwahrscheinlichkeit die statistischen Eigenschaften des Restsignals bekannt sein. Die Amplitudenverteilung des Restsignals wird hier sowohl für die weitkontinuierliche wie die wertdiskrete Adaption exakt berechnet Als Hauptergebnis erweist sich das Restsignal als normalverteilt schon für Filter mit nur einem Koeffizienten, sofern gewisse Voraussetzungen erfüllt sind. Die üblichen Annahmen für die Analyse adaptiver Filter werden teilweise bestätigt und teilweise widerlegt. Für die Dokumentation: Adaptive Filter / LMS-Algorithmus / Koeffizientenzittern / Quantisierung / Konvergenz von Verteilungen / Funktionalgleichungen / Differentialund Differenzengleichungen

Authentische und zuverlässige Mobilkommunikation für sicherheitsrelevante Anwendungen Teil I: Sicherheitsanforderungen und grundlegende Verfahren

Abstract: A communication system for train control based on the GSM mobile radio standard is introduced as an example of a safety-related application. However, data transmission is exposed to stochastic errors as well as intelligent cryptographical attacks. In order to prevent or detect such falsifications, methods of cryptography and error control coding are required, the message authentication code (MAC) beeing the most important component. Part I deals with the analysis of the MAC by means of information theory. Topics are its interpretation as an error detection code and its design and dimensioning in accordance with the constraints of security and reliability. Part II [1] presents the general concept for the safety-related application in the form of a layered architecture and the adaptation to the GSM system

Integrity check in a communication system

Abstract: A method of communication between a first node and a second node for a system where a plurality of different channels is provided between said first and second node. The method comprises the step of calculating an integrity output. The integrity output is calculated from a plurality of values, some of said values being the same for said different channels. At least one of said values is arranged to comprise information relating to the identity of said channel, each channel having a different identity. After the integrity output has been calculated, Information relating to the integrity output is transmitted from one of said nodes to the other.