Gpu-based many-core architecture emulation: A double level approach

The efficient emulation of a many-core architecture is a challenging task, each core could be emulated through a dedicated thread and such threads would be interleaved on an either single-core or a multi-core processor. The high number of context switches will results in an unacceptable performance. To support this kind of application, the GPU computational power is exploited in order to schedule the emulation threads on the GPU cores. This presents a non trivial divergence issue, since GPU computational power is offered through SIMD processing elements, that are forced to synchronously execute the same instruction on different memory portions. Thus, a new emulation technique is introduced in order to overcome this limitation: instead of providing a routine for each ISA opcode, the emulator mimics the behavior of the Micro Architecture level, here instructions are date that a unique routine takes as input. Our new technique has been implemented and compared with the classic emulation approach, in order to investigate the chance of a hybrid solution.

Evolution of the structural and fracture design of the ISS payloads due to the new launchers, specifically applied to Biolab IEC-2 projects

The relatively young discipline of astronautics represents one of the scientifically most fascinating and technologically advanced achievements of our time. The human exploration in space does not offer only extraordinary research possibilities but also demands high requirements from man and technology. The space environment provides a lot of attractive experimental tools towards the understanding of fundamental mechanism in natural sciences. It has been shown that especially reduced gravity and elevated radiation, two distinctive factors in space, influence the behavior of biological systems significantly. For this reason one of the key objectives on board of an earth orbiting laboratory is the research in the field of life sciences, covering the broad range from botany, human physiology and crew health up to biotechnology. The Columbus Module is the only European low gravity platform that allows researchers to perform ambitious experiments in a continuous time frame up to several months. Biolab is part of the initial outfitting of the Columbus Laboratory; it is a multi-user facility supporting research in the field of biology, e.g. effect of microgravity and space radiation on cell cultures, micro-organisms, small plants and small invertebrates. The Biolab IEC are projects designed to work in the automatic part of Biolab. In this moment in the TO-53 department of Airbus Defence & Space (formerly Astrium) there are two experiments that are in phase C/D of the development and they are the subject of this thesis: CELLRAD and CYTOSKELETON. They will be launched in soft configuration, that means packed inside a block of foam that has the task to reduce the launch loads on the payload. Until 10 years ago the payloads which were launched in soft configuration were supposed to be structural safe by themselves and a specific structural analysis could be waived on them; with the opening of the launchers market to private companies (that are not under the direct control of the international space agencies), the requirements on the verifications of payloads are changed and they have become much more conservative. In 2012 a new random environment has been introduced due to the new Space-X launch specification that results to be particularly challenging for the soft launched payloads. The last ESA specification requires to perform structural analysis on the payload for combined loads (random vibration, quasi-steady acceleration and pressure). The aim of this thesis is to create FEM models able to reproduce the launch configuration and to verify that all the margins of safety are positive and to show how they change because of the new Space-X random environment. In case the results are negative, improved design solution are implemented. Based on the FEM result a study of the joins has been carried out and, when needed, a crack growth analysis has been performed.

Deep learning models for the automatic analysis of volleyball plays

Collecting and analysing data is an important element in any field of human activity and research. Even in sports, collecting and analyzing statistical data is attracting a growing interest. Some exemplar use cases are: improvement of technical/tactical aspects for team coaches, definition of game strategies based on the opposite team play or evaluation of the performance of players. Other advantages are related to taking more precise and impartial judgment in referee decisions: a wrong decision can change the outcomes of important matches. Finally, it can be useful to provide better representations and graphic effects that make the game more engaging for the audience during the match. Nowadays it is possible to delegate this type of task to automatic software systems that can use cameras or even hardware sensors to collect images or data and process them. One of the most efficient methods to collect data is to process the video images of the sporting event through mixed techniques concerning machine learning applied to computer vision. As in other domains in which computer vision can be applied, the main tasks in sports are related to object detection, player tracking, and to the pose estimation of athletes. The goal of the present thesis is to apply different models of CNNs to analyze volleyball matches. Starting from video frames of a volleyball match, we reproduce a bird's eye view of the playing court where all the players are projected, reporting also for each player the type of action she/he is performing.

Production Lot Detection through Synthetic Generated Data

Correctness of information gathered in production environments is an essential part of quality assurance processes in many industries, this task is often performed by human resources who visually take annotations in various steps of the production flow. Depending on the performed task the correlation between where exactly the information is gathered and what it represents is more than often lost in the process. The lack of labeled data places a great boundary on the application of deep neural networks aimed at object detection tasks, moreover supervised training of deep models requires a great amount of data to be available. Reaching an adequate large collection of labeled images through classic techniques of data annotations is an exhausting and costly task to perform, not always suitable for every scenario. A possible solution is to generate synthetic data that replicates the real one and use it to fine-tune a deep neural network trained on one or more source domains to a different target domain. The purpose of this thesis is to show a real case scenario where the provided data were both in great scarcity and missing the required annotations. Sequentially a possible approach is presented where synthetic data has been generated to address those issues while standing as a training base of deep neural networks for object detection, capable of working on images taken in production-like environments. Lastly, it compares performance on different types of synthetic data and convolutional neural networks used as backbones for the model.

OpenMP task scheduling strategies to mitigate hardware variability in tightly-coupled shared memory clusters

In questa tesi sono stati apportati due importanti contributi nel campo degli acceleratori embedded many-core. Abbiamo implementato un runtime OpenMP ottimizzato per la gestione del tasking model per sistemi a processori strettamente accoppiati in cluster e poi interconnessi attraverso una network on chip. Ci siamo focalizzati sulla loro scalabilità e sul supporto di task di granularità fine, come è tipico nelle applicazioni embedded. Il secondo contributo di questa tesi è stata proporre una estensione del runtime di OpenMP che cerca di prevedere la manifestazione di errori dati da fenomeni di variability tramite una schedulazione efficiente del carico di lavoro.

Noise mapping of highly trafficked roads of the province of Bologna according to D.Lgs. 194/2005

The European Community has stressed the importance of achieving a common understanding to deal with the environmental noise through community actions of the Member States. This implies the use of harmonized indicators and specific information regarding the values of indicators, the exceedance of limits and the number of people and dwellings exposed to noise. The D.Lgs. 149/2005 in compliance with the European Directive 2002/49/EC defines methodologies, noise indicators and types of outputs required. In this dissertation the work done for the noise mapping of highly trafficked roads of the Province of Bologna will be reported. The study accounts for the environmental noise generated by the road infrastructure outside the urban agglomeration of Bologna. Roads characterized by an annual traffic greater than three millions of vehicles will be considered. The process of data collection and validation will be reported, as long as the implementation of the calculation method in the software and the procedure to create and calibrate the calculation model. Results will be provided as required by the legislation, in forms of maps and tables. Moreover results regarding each road section accounted will be combined to gain a general understanding of the situation of the overall studied area. Although the understanding of the noise levels and the number of people exposed is paramount, it is not sufficient to develop strategies of noise abatement interventions. Thus a further step will be addressed: the creation of priority maps as the basis of action plans for organizing and prioritizing solutions for noise reduction and abatement. Noise reduction measures are reported in a qualitative way in the annex and constitute a preliminary research.