Communities in DSpace

Select a community to browse its collections.

Now showing 1 - 4 of 4

Recent Submissions

No Thumbnail Available
Item
Antennas For Non-Terrestrial Networks
(Università della Calabria, 2024-05-13) De Marco, Raffaele; Fortino, Giancarlo; Boccia, Luigi
This doctoral thesis focuses on antennas for Non-Terrestrial Networks (NTN) and satellite communications, showcasing two innovative antenna designs. First, a novel design for 2-D electronically steerable parasitic array radiator (ESPAR) is presented. This design is based on a 3×3 microstrip patch antenna array and it is intended to serve as a subarray within clustered phased array architecture. Compared to the existing varactor-based 2-D ESPAR designs, the proposed solution allows a continuous beam steering along eight different azimuthal planes, thus, extending the beam scanning capabilities to a conical region. Encouraged by the promising results, the concept is further extended into the Ka-band through integration into a large array configuration. This extension underscores the adaptability and practical viability of the proposed antenna concepts, paving the way for advancements in NTN and satellite communication technologies. The second contribution involves the design of an innovative dual-band dual-polarized transmitarray antenna (TA) utilizing multilayer frequency selective surfaces (MFSS) operating at K/Ka band. The proposed design achieves an optimal compromise between aperture efficiency and thickness, compared to the existing dual-band dualpolarized transmitarrays. This design approach allows for a cost-effective and lowprofile implementation, utilizing a single multilayer PCB without any air gap or vertical transition. An equivalent circuit model for the unit-cell has been formulated and examined, offering insights into the optimal phase control methodology and the transmission mechanism. The proposed design incorporates DL and UL unit-cells that are interleaved and autonomously control the transmitted phase for each band and polarization. Moreover, the proposed unit-cell and transmitarray are versatile, capable of scaling with varying frequencies, making them suitable for implementing other dual-band transmitarrays operating in two distinct frequency bands characterized by a substantial difference in upper and lower frequencies.
No Thumbnail Available
Item
GRAPH MINING AND MULTIMODAL REPRESENTATION LEARNING FOR EMERGING DECENTRALIZED SOCIO-ECONOMIC DOMAINS
(Università della Calabria, 2024-02-23) La Cava, Lucio; Fortino, Giancarlo; Greco, Sergio; Tagarelli, Andrea
Thumbnail Image
Item
On Chip Monitoring for Efficient Thermal Management
(Università della Calabria, 2024-05-10) Zambrano Chavez, Jose Benjamin; Lanuzza, Marco; Fortino, Giancarlo
Temperature is a critical physical parameter frequently monitored in electronic systems due to its substantial impact on performance and power consumption. In recent years there has been great effort to enhance the technology for development of temperature sensors. Partic- ularly, CMOS-compatible realizations have emerged as a promising alternative, showcasing different physical mechanisms for temperature sensor implementation, including BJTs, Ther- mal diffusivity and MOSFETs. MOSFET-based temperature sensors offer several advantages, such as low-voltage operation, high energy efficiency, and a compact footprint. Although they may sacrifice resolution and accuracy, this trade-off aligns with the prevalent trend in advanced System-On-Chips designed for Internet-of-Things (IoT) nodes, where these parameters can be relaxed since the priority is energy efficiency while being fully-integrated by handling directly a digital readout, also known as smart temperature sensors. This thesis focus on designing and evaluating energy-efficient smart temperature sensors based on MOSFET devices. The research encompasses two main achievements. The first involves a smart temper- ature sensor focused on low-voltage and low-power operation, with a nominal 350 mV of supply voltage and a power consumption of just 14 nW at 25 ◦C with a silicon footprint of 0.049 mm2, a resolution of 0.27 ◦C and achieving a Resolution Figure-of-Merit (R-FoM) of 0.034 nJ·K2. This sensor is tailored to meet the stringent constraints of IoT applications.. The second achievement focuses on a compact sensor targeted for Dynamic Thermal Management. The circuit exhibits a wide supply voltage operating range from 0.6 V to 1.8 V with an energy per conversion of 1.06 nJ, noise-limited resolution of 0.24 ◦C, a silicon area of 0.021 mm2, and an a R-FoM of 0.061 nJ·K2. The latter characteristic is particularly notable given the area constraint and the sensor’s ability to operate across a broad range of supply voltages
No Thumbnail Available
Item
New developments in deep learning for natural language processing and explanation
(Università della Calabria, 2024-01) Simeri, Andrea; Tagarelli, Andrea; Fortino, Giancarlo
This PhD thesis introduces LamBERTa, a pioneering BERT-based language understanding framework tailored for law article retrieval as a prediction task. LamBERTa addresses the challenges posed by a multi-class classification scenario featuring hundreds or thousands of classes and minimal per-class training instances, generated in an unsupervised manner. The study explores domain-adaptation coupled with task-adaptation to train LamBERTa models for the Italian Civil Code (ICC) article prediction, incorporating a recently defined Italian legal BERT and investigating the impact of enhancing the tokenizer with terms from the target legal corpus. The work’s contributions extend to the exploration of LamBERTa models’ explainability, employing model-level and instance-level explainability techniques. While initially focused on the Italian Civil Code, the LamBERTa architecture proves adaptable to other law code corpora such as the General Data Protection Regulation (GDPR), further powered by selfsupervised task-adaptive pre-training stages, with or without data enrichment based on recitals, marking the first comprehensive study of domain-adaptive and task-adaptive legal pre-trained language models for GDPR article retrieval. Furthermore, the thesis conducts a network analysis and mining study, presenting the first examination of citation networks inferred from ICC articles. Insights into structural features reveal hidden patterns, facilitating new interpretations and studies of the ICC. This methodology is adaptable to other civil law code systems organized similarly to the ICC, supporting tasks like statute law retrieval, entailment, and question answering. The study also introduces LawNet-Viz, a web-based tool for modeling, analyzing, and visualizing law reference networks, aiming to assist legal experts and citizens in legal information processing and understanding. Finally, the thesis outlines the TrustSearch project’s ambitious goal to develop an AI tool that enhances the search experience promoting critical thinking. Two stance detection approaches, Definition-based and User-driven, are presented, showcasing the effectiveness of AI in transforming the search and data discovery experience. In particular, Chapter 1 introduces the motivations behind this research work, then it overviews recent works that address legal classification and retrieval problems based on deep learning methods and network analysis techniques as well as entailment resolution using Transformers and LLMs. Chapter 2 provides the reader with essential information about the research topic and it serves to establish the context around The Italian Civil Code (ICC) and The General Data Protection Regulation (GDPR). Chapter 3 presents our proposed framework for the civil-law article retrieval problem named as LamBERTa with the application on the ICC corpus. Within it, we investigate the performance of the models through an extensive quantitative and qualitative set of experiments. Furthermore, we study the effects of domain adaptive pre-training strategies and task adaptive training strategies and we scrutinize the concept of Explainability leveraging on the use of different techniques. Chapter 4 presents the LamBERTa models applied on the General Data Protection Regulation (GDPR) corpus, also powering them through self-supervised task-adaptive pre-training schemes. Chapter 5 explores the Italian Civil Code (ICC) from an unprecedented perspective based on network analysis. Furthermore, we present LawNet-Viz, a web-based tool for the modeling, analysis and visualization of law reference networks extracted from a statute law corpus. Chapter 6 is based on the research I carried out during the PhD visiting period abroad at (UPV) University Polytechnic of Valencia, Spain, where I actively participated in the TrustSearch project, a research project funded by the European Union, developing an AI tool that offers a new search experience to promote critical thinking. In this regard, we developed two different approaches, the first one which is based on the state of art LLMs prompt engineer, the second one consists is an application of an encoder-decoder Transformer-based model for the stance detection leveraging on the entailment task. Chapter 7 summarizes findings and discusses future studies which could extend the research.
Thumbnail Image
Item
A Logic-based Framework for Characterizing Nexus of Similarity within Knowledge Bases
(Università della Calabria, 2024-04-04) Ricioppo, Aldo; Terracina, Giorgio; Manna, Marco
Complex challenges frequently necessitate the establishment of significant connections among diverse entities. For instance, selecting the optimal CV, finding a suitable apartment or deciphering the causal factors behind a specific disorder are just a few examples, and each of them requires deep understanding of what characterizes a set of entities. In addition to the aforementioned examples, having to recognize similarities between entities is a recurring phenomenon in a multitude of real-world scenarios. Researchers from different fields have presented various methodologies over the last century to evaluate these similarities, commonly called similarity. In recent times, momentum has increased with the advent of “Google Sets”, leading to the fervent development of strategies to amplify a given set of entities while preserving their original shared interconnected properties. As a consequence, current methodologies can encompass, in a way or another, relevant interconnected properties shared by entities, a concept we term the nexus of similarity. Machines have demonstrated considerable prowess in handling similarity evaluations, often returning numerical scores as a result, and set expansions, thus giving the end user the opportunity to observe entities similar to those he was looking for. However, there is a notable gap in formally characterizing the nexus of similarity in a way that is intelligible by machines and interpretable by human intellect, especially considering that the attempts that have gained the most traction thus far are often bound to cases very specific, such as those made with respect to RDF graphs. To address these gaps, our endeavor contributes significantly to the existing literature. We aim to construct a novel framework grounded in logical constructs, designed to systematically and autonomously delineate the nexus of similarity. Our framework extends not only to pairs of entities but also to sets of tuples of entities, which we term anonymous relations, within a knowledge base. Furthermore, our analysis encompasses an in-depth examination of the computational complexity inherent in the proposed framework. Such an investigation affords a thorough insight into its feasibility and a subsequent evaluation of scalability. Both are critical components for the framework’s practical application. In summary, our study pioneers a novel, knowledge-driven approach capable of characterizing nexus of similarity and that can be used as a means to perform entity set expansion in a manner clear and intelligible to humans. Two of the principal components integral to our framework will be the semantic resources, specifically selective knowledge bases, that in their essence are knowledge bases equipped with a particular supplementary algorithm, and the explanation languages, of which we will take into consideration one in particular, which in our opinion has all the ideal characteristics to be considered a language suitable to reveal the nexus of similarity as best as possible. During the work, we will also justify our design choices. With the help of these means we aim to fill the perceptible gap in the characterization of nexus of similarity. At the same time, we will show how some of the current approaches to entity set expansion do not notice that by their very nature this kind of expansions should take the form of a taxonomy rather than a chain. To resolve this other gap, we will introduce the concept of expansion graph.