Vitor Camargo Nardelli

Resource-constrained Internet of Things (IoT) devices motivate the investigation into cryptographic designs with measurable implementation costs. This work presents CHIMA (Cryptography cHaotic lIghtweight systeM and Authentication), an experimental system combining a key generator based on chaotic states, a Feistel cipher using the AES substitution box and a keyed bit permutation, and a Lesamnta-LW-based ciphertext check. Logistic, H’enon, and Lorenz systems are examined for key generation, and cipher variants with 64-bit and 128-bit blocks are evaluated. The study reports functional tests, avalanche measurements, NIST statistical tests, and implementation measurements on Arm Cortex-M0+, Cortex-M33, and Cortex-M7 microcontrollers. Controlled WiFi/MQTT and LoRa experiments assess communication interoperability. The results show that the chaotic key generator achieved high entropy and passed all NIST statistical tests, indicating its suitability for generating pseudorandom sequences. Based on the combined encryption and decryption times, CHIMA-64 requires 57.95–64.16% less time than PRESENT, while CHIMA-128 requires 23.60–59.36% less time than AES-128 and 13.32–47, as well as 10% less time than CLEFIA across the three platforms.

Astronomical observations often suffer from environmental variations that cause blurring in telescopes, compromising the quality and sharpness of the captured images. To mitigate these effects, this work proposes a methodology for detecting stars in blurred images by developing a mathematical model based on probabilistic approaches, exploring the relationship between the light dispersion of stars and a Gaussian distribution. The developed model incorporates two approaches to define the similarity structure: correlation and orthogonality similarity. The results obtained did not show false positives when using orthogonality similarity, achieving a precision of 100%, while the correlation method reached a precision of 95%, highlighting the potential of orthogonality similarity as an effective alternative to identify stellar patterns under blurred conditions. For a dataset with homogeneous star luminosity, an accuracy of 100% was achieved.

This work explores steganography as a tool for embedding hidden information, such as watermarks, into images. Encoding and decoding methods based on the Discrete Wavelet Transform (DWT) and Singular Value Decomposition (SVD) were proposed. The study provides a comprehensive comparison of the steganographic behavior of models using low-frequency, high-frequency, and multi-level sub-bands, evaluating the impact of different wavelet families, scaling factors, and decomposition levels. Metrics such as PSNR and MSE were used to assess the results, which were compared with the existing literature. The experiments demonstrated that high-frequency techniques offer the best balance between imperceptibility and extraction quality, achieving PSNR values of up to 51.19 dB when using the sym16 wavelet, while the multilevel model presented higher security at the cost of lower decoding quality. These findings help address the unexplored gaps in the literature, providing a detailed analysis of the possibilities and limitations of these techniques.

Fire detection plays a crucial role in safeguarding agricultural lands, pastures, and forested areas from catastrophic damage. The presence of diverse textures, colors, and shapes in these environments makes fire detection challenging, with many existing methods prone to high false-positive rates. In this work, we propose a two-stage fire detection method utilizing k-Nearest Neighbors (kNN). First, kNN is used to classify images as containing fire or not, achieving an accuracy of 81.57%. For images classified as containing fire, they are further divided into sub-images, and additional features are extracted. kNN is then applied again to classify these sub-images to localize the fire regions, achieving an accuracy of 84.81%. This approach is particularly effective in agribusiness contexts, where the shades of green dominate the landscape and the presence of orange and red is strongly correlated with fire. The proposed method offers a robust solution for the early detection and location of fires on agricultural land, supporting the protection of crops, pastures, and forest borders while enabling a rapid response to minimize economic and environmental losses.

The dimensional accuracy of a laser scanner has been extensively evaluated using various measurement methods and diverse reference standards. This study specifically focuses on two key considerations. Firstly, it assesses the dimensional accuracy of the laser scanner by employing another laser scanner, a handheld scanner, as the reference measurement method. Secondly, the study involves the use of three spheres fixed on each wall in both coplanar and non-coplanar positions within a laboratory room at SENAI ISI-SIM. The primary objective is to determine the dimensional accuracy between the centers of the coplanar and non-coplanar spheres up to 10 m. The comparison includes measurement uncertainties, as per ISO GUM standards, obtained using the laser scanner in a laboratory setting with controlled temperature and humidity. Analyzing non-coplanar dimensional accuracy enhances our understanding of the metrological performance of the laser scanner, particularly when assessing the dimensions of objects positioned randomly within a scanning scene.

The need to control the real-time location of assets is increasingly relevant worldwide. The Ultra-wideband (UWB) technology is an IoT solution for real-time locating systems (RTLS). The location of the asset is obtained by the signal exchange between a wireless tag (asset) and fixed anchors. The tag interacts with the fixed anchors, defining its position through the distances obtained by trilateration. This data is sent to the server through the gateway. It is well-known that this process has several sources of errors. However, the measurement uncertainty assessment of UWB technology is an important topic regarding its scope of use. This paper presents a task-specific measurement uncertainty evaluation for the UWB positioning system, according to the ISO GUM. It aims to propose a method to support decision-making regarding the possible uses of UWB technology. The position provided by the UWB is compared with reference points using Cartesian coordinates that are measured with a total station, providing metrological reliability. Using the information from the estimated uncertainty, one can define the minimum tolerance interval associated with UWB technology for a given use. A case study demonstrates the method.