The evaluation of electrical energy flows is at the base of all operations connected to the commercial transactions in energy market. Typically, the measurement apparatuses are made and characterized to work in ideal sinusoidal conditions. But, with increasing diffusion of power electronic devices, the typical characteristics of power network are far from sinusoidal ones. With the aim to link the cost of energy to the level of power quality, it is necessary to determine the energy but also the indexes which characterize the power quality. In this paper the design and implementation of a multi processor measuring instrument based on multiple digital signal processors (DSPs) is discussed. This system can embody a number of special signal processor and analog I/O modules, specifically designed to obtain the flexibility and scalability to meet the needs of real-time measurement applications and low cost. After a prototype description, a part of analysis algorithm is reported. The paper is completed by preliminary results of metrological characterization.

The capacity of the communication network is a key metric for both network planning and management, and performance assessment of applications that take advantage of the network. Several tools for capacity measurement are present in literature, which provide satisfying results when used in wired networks. On the contrary, they provide inaccurate results on wireless networks. The paper presents a refined study of processing times at each protocol layer, which suggests that capacity estimates provided by the existing tools are affected by a systematic error. A suitable measurement methodology and a related testbed are proposed to experimentally validate the theoretical results. The experimental outcomes demonstrate the validity of the study, and permit to explain why existing tools are inaccurate on wireless links.

A heuristic procedure of classification, the quantum-inspired classifier (QIC), exploiting search space exploration and resource exploitation of quantum computing on a software basis is proposed. The application to the problem of speed sequence classification for vehicle emission factor determination based on drive styles is shown. Experimental results are discussed by showing the QIC capability of converging better and faster than classical evolutionary algorithms.

the project proposes to accomplish a new system architecture of real-time video inspection of textile materials, which should ensure a fault tracking-down, rate of over 99%, in the presence of disturbing elements from the industrial environment. The inspection procedure stipulates that the system should learn the correct model of the textile material, so that it is able to notice the deviations that might appear during its inspection. The model changes are analyzed with the help of many detection algorithms to separate real flaws from normal variations, accepted in the material. After the detection, follows the localization of the flaw, settling the coordinates, its size and recording it in the flaw library. This is a work-in-progress that has to be finished in august 2008.

We present a new method for the identification of bubble nucleation in superheated droplet detectors. This is accomplished through some elaborate signal processing techniques applied to the acoustical recording of the nucleation events. These include the application of wavelets, chirp-z transform and pulse shape identification procedures to locate temporally and validate the nucleation together with back-propagation algorithms for spatial localization. Experimental results are given for the spatial localization at different temperatures and pressures.