Search

In this paper we describe the activities towards the design of a common framework for the Instrument Control and Data Processing Units for the three scientific payload instruments on board the joint ESA-JAXA SPICA mission, currently at the end of its phase A study. In this context, we started a program to assess modular architectures based on the use of a quad-core fault-tolerant LEON4 SPARC V8 processor on a SpaceWire network. We will describe the results of our initial tests using both Asymmetric Multi processing (AMP) and Symmetric Multi Processing (SMP) configurations. In addition, the possibility to adopt the RTEMS real time operating system, already space qualified on single core processors, will be evaluated both in terms of latency performances and of dynamical allocation of the resources. Finally, we will present the outline of the way forward for the next phases of the SPICA project.

Downlink synchronisation error estimators suited for OFDM-TDMA wireless LAN are presented. The focus is on the timing error estimator, which is of the Early-Late type and has a low complexity, making it attractive for high speed systems. It also allows a large sharing of the computational burden with the frequency error estimator. The estimator is presented and analysed. Simulations are performed to asses the performance of the synchronisation system over indoor channels.

A new duration intrinsic model for improved speech recognition by HMM techniques is presented. Assuming an exponentially decaying time dependency of the states loop probability, the duration density can be factorized and a path early pruning theorem demonstrated. As a consequence, computational complexity is greatly reduced with respect to explicit models, whereas recognition performances improve considerably.

An information based, two-stage DP procedure is proposed for phonemic HMM structure inference. It operates on the phonetic description of speech provided by means of an ergodic HMM. The resulting HMM, while comprising a quite large number of states, share the observation densities originated by the EHMM.

By means of digital processing of vocal signals we have automatically extracted some quantitative meaningful features which are related to vibratory pattern perturbations of the vocal folds, when affected by pathological processes. Here we give the methods and techniques of vocal signal analysis and the procedure for the evaluation, in a multidimensional space, of the parameters extracted, to estimate their capability of discrimination between the normal and pathologic classes.

We have determined the levels of DNA-polymerases-alpha and-beta in fibroblasts obtained from normal subjects and from patients with Xeroderma Pigmentosum (XP) belonging to three different complementation groups and to the variant form. The assays have been performed in crude extracts and after fractionation on sucrose gradients. The levels of alpha and beta-polymerases in the different cases of XP were found to lie within the same range as the control values, and no correlation was found with the severity of the symptoms. The sedimentation coefficients of the two polymerases from all the pathological lines were identical to those of the normal fibroblasts.