ABSTRACT. Microwave sensing has proven to be a powerful enabler for real-time, non-invasive monitoring in biomedical, environmental, and chemical domains. In this talk, an overview of the research developed at the University of Bologna, with a focus on energy-efficient and wearable microwave sensors for hydration and ethanol detection will be presented. Our work spans multiple application domains—from human skin hydration sensing using compact resonant structures and machine learning, to ethanol detection via battery-less, wirelessly powered “filtennas” integrated with microfluidic channels. These solutions offer low-cost, scalable approaches for real-world diagnostics in healthcare and agricoltural applications. A key innovation I will highlight is the use of self-oscillating antennas (SOAs) for plant hydration monitoring. In this design a tree trunk becomes part of the oscillator’s load, allowing hydration levels to be inferred from changes in the system steady-state oscillation regime. This approach eliminates the need for external measurement circuitry, creating a truly autonomous sensing node ideal for smart agriculture and forestry.

ABSTRACT. In the last decade, the widespread hydrogeological instability has caused anomalies in several aspects of the natural earth life. Among them, the increase in landslides is an important concern for both the civil and scientific communities. In fact, the frequency of landslide events is progressively increasing, and with it, the demand for monitoring systems. Monitoring systems, on the other hand, can be implemented in a plethora of methods. In this regard, Wireless Sensor Networks (WSNs) are among the most innovative solutions, as they enable wireless landslide monitoring through the measurement of crucial parameters and their status assessment through long-range wireless technologies, such as the LoRa. This paper focuses on the implementation of a multi-sensorial WSN in the inhabited Macchia da Sole (TE) mountain locality, to monitor the main factors behind the landslide hazard, including the measurement of building cracks and underground water level. The main contribution of this paper is the level of autonomy achieved, as expressed by the average power consumption of 1.4 mW per node. Furthermore, the system integrates more types of sensors, and it is used in a zone with poor signal coverage.

ABSTRACT. This paper presents results of applications of artificial neural networks (ANNs) for representing the temperature-dependent behavior of a surface acoustic wave (SAW) resonator within the range from 0˚C to 100˚C. The temperature dependence of the admittance parameters of a SAW resonator (TO-39 device) with the nominal resonant frequency of 423.2 MHz is modeled using ANNs. The model is developed by using the data obtained by measuring the device characteristics in the mentioned temperature range. It is shown that the trained ANNs can be used for a fast prediction of the resonant frequency in the mentioned temperature range and further analysis of device behavior (frequency- or temperature- dependent ones).

ABSTRACT. The paper demonstrates an idea and performances of a novel solution introduced to resolve technological issues that are particularly important at higher microwave and millimeter wave frequencies. The concept presented here combines two filter technologies in a single composite filter - these are dielectric resonator filter technology and asymmetrical waffle iron technology. Dielectric resonator filter technology is used to realize the main bandpass filter response at millimeter frequencies while asymmetrical waffle iron technology is used to confine electromagnetic waves within cavities but without any conductive (i.e. “low ohmic”) and mechanical contacts between functional filter parts which is commonly required in classical solutions.

ABSTRACT. Open Ring resonator as a Microwave Microstrip Pressure Sensors on Textiles was tested in the range around 2.5 GHz. Rubber foam was used to apply pressure to the entire resonator. The measurement was performed for three cases of rubber foam: Two thicknesses and metallization of the upper part. The metallization on the textile is made of copper tape with a non-conductive adhesive so that it can be removed and shortened without damaging the textile. Also, the rubber foam is glued to the oscillator and can be removed. With this, an attempt was made to meet the sustainable development goals. The sensor is still in the experimental phase.

ABSTRACT. The Internet of Things (IoT) offers transformative potential in education by bridging theoretical knowledge with real-world applications. At the University of Novi Sad, students engage with IoT through a specialized course that emphasizes device prototyping, 3D design and 3D printing, microcontroller programming, data analysis, and device connectivity. The curriculum blends technical disciplines with problem-solving and interdisciplinary teamwork. Challenges like diverse student backgrounds and technical constraints are mitigated through simulation tools and prepared learning materials and exercises. This approach has significantly boosted student engagement, practical competence, and readiness for future industry demands. We have developed an educational IoT device that is used to demonstrate the topics throughout the curriculum and provide a testbed for student projects, while being inexpensive to produce and maintain.

ABSTRACT. An analysis of error probability of decoding of (U|U+V) construction is described in this paper. The derivation of the probability of decoding error is shown for (U|U+V) sequential decoder. The observed theoretical probability of error can be extended to (U|U+V) constructions with varying numbers of levels. The method is derived for an additive white gaussian noise channel (AWGN). The presented expression depends on channel parameters, code length and the number of errors that components can correct. Moreover, the task of maximizing the code rate of the (U|U+V) code with component codes having the Gilbert-Varshamov minimum distance is depicted. Simulation results demonstrate the coincidence of theoretical results with simulated probability of decoding failure. The Gilbert-Varshamov rates profile on signal-to-noise (SNR) ratio for different probability of decoding error is also shown.

ABSTRACT. The 16-bit floating-point format is a popular alternative to the widely used 32-bit floating-point format (FP32). This study analyzes the performance of two commonly employed 16-bit versions, dlfloat16 and bfloat16, for data with Laplacian and Gaussian distribution. The applied performance metric is the signal-to-quantization noise ratio (SQNR), allowed by using the connection between the floating-point format and piecewise uniform quantization. It is shown over a wide range of input data variances that SQNR results for Laplacian and Gaussian data differ slightly, indicating that both observed formats are robust to changes in the data distribution. SQNR analysis also suggested that dlfolat16 is a better option when increased calculation precision is required, while bfloat16 is better from a dynamic range perspective.

ABSTRACT. We analyze the performance degradation due to fractional Doppler shifts in multiuser uplink OTFS, when the OTFS channel estimation is performed with OMP algorithm. For the actual channel we use input output expressions in Doppler-Delay (DD) domain, where fractional Doppler shifts and delays are introduced. For estimation we use channel model in DD domain with delays and Doppler shifts on the grid to generate the OMP sensing matrix. The channel estimation does not account for the fractional Doppler shifts and delays, i.e. uses integer Doppler shifts and delays. To evaluate the algorithm performance for different fractional Doppler shifts we use the estimation Mean Square Error and the detection symbol error rate. The purpose is to ascertain the appropriate circumstances for OMP's application without significant performance loss. To improve channel estimation performance, we also use refined DD domain grid for OMP estimation i.e. larger dictionary of the sensing matrix.

ABSTRACT. In embedded systems, digital receivers play a vital role in interfacing with analog front-ends and acquiring high-speed data for further processing. They are particularly essential in applications involving real-time signal acquisition, demodulation, and analysis. This paper presents the design and implementation of a high-performance digital receiver, realized using the TI AFE7900EVM and AMD KCU116 development platforms. The receiver operates across a frequency range of several hundreds of MHz to 7.4 GHz, supporting a wide variety of modulation schemes and bandwidths up to 200MHz

ABSTRACT. This paper considers the main methods of user authentication based on brain EEG signals. It describes optimal strategies for preprocessing an EEG signal and proposes directions for further development of methods for user authentication and verification based on the user's response to stimulation using classification methods

ABSTRACT. This paper presents an experimental characterization of RADFET sensors. Two irradiation campaigns were carried out at TENMAK (Turkey) and the Vinča Institute (Serbia), employing Co-60 gamma sources with dose rates of 4.813 Gy/h and 17.04 Gy/h, respectively. RADFETs with different gate oxide thicknesses, including dual-gate high-k dielectric structures, were tested by monitoring threshold voltage shifts in response to accumulated radiation dose. The results indicate that devices with thicker oxide layers demonstrate greater sensitivity due to enhanced charge trapping. Moreover, lower dose rates were found to increase sensor sensitivity, particularly in devices with thinner oxides. These findings contribute to the optimization of RADFET design for reliable and accurate total ionizing dose measurements in space environments.

ABSTRACT. This paper presents an analysis the posture of the human body in a sitting position while working at the computer. Poses of the human body were captured using 9 inertial sensors placed on the human body, with an emphasis on upper body acquisition. Fourteen postures of improper holding of the human body while sitting were observed. The difference between incorrect and correct body posture is quantified through the angles of rotation of the sensor in 3 axes. The results show that the subject was in an incorrect body position more than 80% of the time, and that the greatest deviation in body position was observed in the area of the lower back.

ABSTRACT. Bad sitting posture can cause the early development of musculoskeletal disorders. Consequently, a lot of research is done on sitting posture monitoring systems (SPMSs). We provide an overview of SPMS research, focusing on what is missing in the prior work. We synthesize a coherent 2D taxonomy of sensors used in SPMSs, review recent SPMS papers and radar based SPMS papers, and provide a clear overview of sensor type combinations used in multi-sensor SPMSs. Motion and pressure sensors are prevalent in the recent SPMSs, while radar was rarely used. There are many sensor combinations which have not been used in multi-sensor SPMSs. Based on our findings, we point out the possible paths of future research.

ABSTRACT. Communication through LEO satellites creates an opportunity to cover remote areas and provide a reliable connection in various scenarios. Optimizing this communication through channel forecasting creates multiple benefits by increasing communication reliability and communication speed. This can be done by using artificial neural networks and training them to forecast the future channel SNR values, based on previously observed ones. In this paper we analyze the influence of input sequence length on the spectral efficiency improvement in different scenarios in order to balance the tradeoff between computational resources and performance. The results show input sequence length has a moderate impact when performance is evaluated on individual channel conditions. In contrast, when evaluation is done on a group of channels with various conditions, the impact of input buffer length is significant. The results show promise in terms of practical computational requirements and open new directions for future work intended for real-world deployments.

ABSTRACT. In modern 4G and 5G networks, especially in dense urban areas, ensuring reliable coverage at the cell edge remains a major challenge, particularly for indoor users. To limit interference in OFDM systems, operators often apply overly conservative strategies, such as excessive downtilting and avoiding low-band deployment. This study analyzes an urban cluster in a Central American network where such practices led to poor service quality and low edge throughput. Using spectral efficiency and user experience metrics, underperforming sectors were identified and root causes classified, with emphasis on coverage gaps. Corrective actions, mainly low-band activation and tilt adjustment yielded significant SINR and user experience gains. The results highlight the need for balanced network design that jointly addresses coverage and interference, with spectral efficiency analysis as a key diagnostic tool.

ABSTRACT. IoT (Internet of Things) as a technology with its associated advantages and limitations has become indispensable part of day-to-day life. The last two decades have experienced a steady rise in the production and deployment of IoT devices reinforcing the importance of security and reliability aspect. Moreover, this technology has become an integral component of contemporary military systems. Thus, security in IoT has emerged as a critical concern, attracting considerable attention from researchers in recent years. This paper proposes a comprehensive survey of prevalent attacks on IoT devices, along with taxonomy structured around various relevant criteria. The aim of the survey is to highlight the security challenges in civilian and military IoT, considering the diversity of vulnerabilities and demands.

ABSTRACT. In this paper an automated egg counting system for conveyor belts using YOLOv11 detection and ByteTrack tracking is proposed. Nano and small YOLOv11 models are compared, achieving approximately 90% accuracy with low inference time. Results demonstrate that nano models provide comparable performance to larger variants while maintaining computational efficiency for agricultural automation.

ABSTRACT. Vibration‑based damage detection depends on precise low‑frequency estimation. Devices such as Arduino and Raspberry Pi are popular among non‑CS engineers for prototyping IoT systems, but their limited memory and processing power challenge conventional DSP methods. This work examines how these constraints shape implementation choices and whether tailored alternatives may offer a more suitable option than standard libraries.

ABSTRACT. This paper presents an ensemble learning-based framework for the recognition of stress states using IoT-based wearable photoplethysmography (PPG) device. Pulse rate variability (PRV) features were extracted from recorded PPG signals of 10 healthy subjects during the controlled experiment that included cognitive and rest tasks. Stacking and Voting ensemble classification approaches were applied for cognitive stress state recognition. Stacking classifier outperformed the Voting classifier, achieving 72% accuracy. This work confirms the feasibility of using IoT-based wearable PPG devices for automated stress detection.

ABSTRACT. This paper describes and discusses the concept of home-compatible lab exercises that are designed so that students of electrical engineering can do them on their own, at the leisure of their homes. The requirements for these labs are different than for the classical sets of lab exercises that are conducted in the laboratory environment. On the other hand, the labs have advantages over the remotely run lab exercises, and are more useful for the students, in comparison. A set of lab exercises has been developed within the framework of Erasmus+ Conn’Cor project, and evaluated by each of the partner institutions. The feedback received shows that the concept is a promising one, being embraced in special circumstances, but also as the general concept for specific courses.