Theodore Kotsilieris, Stelios Kalogeropoulos, George Karetsos, Vassilis Loumos, Eleftherios Kayafas
PERFORMANCE EVALUATION OF MOBILE AGENT TECHNOLOGY FOR THE CONFIGURATION AND AUDITING OF SLSS IN DIFFSERV ENABLED IP NETWORKS

Nowadays we witness a growing interest for the efficient management of the QoS delivered by the service providers to the Internet users and the usage of SLAs. In such an environment the need to maximise customer satisfaction and reliability and consequently, the deriving obligation to quantify and measure service functions becomes apparent. As traditional centralized NMS manager-agent approach has stretched to its limits, we propose a way of bringing together the static nature of SNMP protocol with the Mobile Agent Technology (MAT) in IP network. In order to address lack of quality of service inherent in current Internet technology, we have chosen differentiated services architecture by implementing the corresponding DiffServ MIB proposed by IETF. The major objective of this paper is to compare the required time to configure and monitor DiffServ domain routers by using MAs and the client-server model. The results drawn from our research lend support to the claim that the performance enhancement by adaptating MAT to rigid network management area depends mainly on the traffic load on the link connecting the managed network with the manager.

Heikki Jokinen, Risto Ritala
VALUE ASSESSMENT OF MEASUREMENTS IN LARGE MEASUREMENT INFORMATION SYSTEMS

Modern industrial production processes are rich sources of measurement data, acquired in many forms and time periods. The integration of different forms of data and information has become more and more important in process industry. Decisions needed in managing processes in an optimal way force us to utilize the information in a way that maximizes its value. This paper discusses the value assessment of measurements and measurement information in large industrial systems. In particular, probabilistic description of information and the value of uncertainty and information in measurements concerning the performance of operational decision making are discussed.

Călin Ciufudean, Constantin Filote, Adrian Graur, George Mahalu
SCHEDULING SEQUENTIAL PROCESSES IN TREE NETWORKS

In this paper we focus on the problem of scheduling time-critical data flow over a measuring tree network. We assume that data are operated in discrete time and their arrival and deadline processes are arbitrary. Our goal is to determine a policy which transmits data with minimum extension time at every node (every link) in a tree network. The models existing in the literature do not consider simultaneous performance of various hardware components of a complex system. When a hardware component of the system fails, the system reconfiguration is often less than perfect. On this assumption we introduce an algorithm to model the availability of measuring systems with colored Petri nets (CPN). Regarding the fact that the availability of a measuring cell i (or branch in the network) is calculated with Markov chains, we model our system with stochastic CPN. The utility of our approach in alleviating the computational burden of measuring systems availability is illustrated via a Markov chain structure.

Leopoldo Angrisani, Mauro D’Arco, Rosario Schiano Lo Moriello, Michele Vadursi
OPTIMAL SAMPLING STRATEGIES FOR BAND-PASS MEASUREMENT SIGNALS

Data Acquisition Systems (DAS) play a fundamental role in a lot of modern measurement solutions. One of the parameters characterizing a DAS is its maximum sampling frequency. As well known from Nyquist theorem, for a correct digitization of analog signals, sampling frequency must be greater than a limit value, related to signal bandwidth. With regard to band-pass signals, which are very common in telecommunications, it is possible to single out some separated ranges of licit sampling frequencies that are significantly lower than such threshold. But, how to choose the sampling frequency conveniently? It seems that there is no instrument currently on the market that selects it properly, given the signal characteristics. The paper proposes a criterion of choice aimed at optimising the DAS resources and two algorithms that perform an automatic selection of the optimal sampling frequency.

Pier Andrea Traverso, Domenico Mirri, Gaetano Pasini, Gaetano Iuculano
CASCADED RECTANGULAR WINDOWS FOR LEAKAGE REDUCTION IN SPECTRAL ANALYSIS

A novel family of windows are proposed for digital spectrum analysis and the main figures of merit compared to those associated with both well-known and state-of-the-art windows available in the literature. The elements of the family, which are obtained by means of successive linear convolutions between elementary rectangular windows, present a very good side-lobe behaviour if a moderately high order of convolution is considered and allow for a strong reduction of the spectral leakage effect at the output of the analysis. The growth of the main-lobe within the spectrum of the proposed windows is acceptable for most of the applications, and the benefits deriving from the rapidly fading side-lobe signature can be appreciated even in presence of the need for a quite high frequency resolution, when the large storage capabilities and computational efficiency provided by modern sampling instrumentation are properly exploited.