In wireless communication systems, the asynchronization of the oscillators in the transmitter and the receiver along with the Doppler shift due to relative movement may lead to the presence of carrier frequency offset (CFO) in the received signals. Estimation of CFO is crucial for subsequent processing such as coherent demodulation. In this brief, we demonstrate the utilization of deep learning for CFO estimation by employing a residual network (ResNet) to learn and extract signal features from the raw in-phase (I) and quadrature (Q) components of the signals. We use multiple modulation schemes in the training set to make the trained model adaptable to multiple modulations or even new signals. In comparison to the commonly used traditional CFO estimation methods, our proposed IQ-ResNet method exhibits superior performance across various scenarios including different oversampling ratios, various signal lengths, and different channels.

A document by Riccardo Latella . Click on the document to view its contents.

This study explores the potential of quantum machine learning and quantum computing for climate change detection, climate modeling, and climate digital twin. We additionally consider the time and energy consumption of quantum machines and classical computers. Moreover, we identified several use-case instances for climate change detection, climate modeling, and climate digital twin that are challenging for conventional computers but can be tackled efficiently with quantum machines or by integrating them with classical computers. We also evaluated the efficacy of quantum annealers, quantum simulators, and universal quantum computers, each designed to solve specific types and kinds of computational problems that are otherwise difficult.

Earth observation has a crucial role in understanding and monitoring our planet’s health and changes. Spaceborne Synthetic Aperture Radar (SAR) has become a valuable technology for Earth monitoring, leading to a massive expansion of satellite launches. However, within the limited radio frequency (RF) band, Radio Frequency Interference (RFI) poses a significant challenge for SAR technology. RFI can have a significant impact on the overall system performance and particularly on SAR image quality. To analyze and solve the interference problem, a simulator/emulator is required at the RF level to emulate and analyze the effects of different RFI sources on the final focused spaceborne SAR image. This paper presents an open-source RF-level SAR emulator for spaceborne applications called SEMUS. SEMUS is an integrated end-to-end framework for realistic spaceborne SAR scenarios that can generate raw RF data (Level-0) for an arbitrary scene and reconstruct the final SAR-focused image (Level-1). Moreover, the emulator is capable of injecting arbitrary RFI waveforms into the raw SAR data. The simulation results prove SEMUS’s ability to generate high-quality Level-0 SAR data above Melbourne, Australia. Affirming its capability, SEMUS is able to reconstruct Level-1 free of RFI or contaminated with interference.

Blood glucose is typically measured using invasive methods such as finger pricking, which although accurate are not suitable for frequent use as they cause extreme pain, and do not provide provisions for continuous glucose monitoring. Recent studies have proposed non-invasive glucometers that are based on scientific principles such as optical polarimetry, thermal emission, and electromagnetic approaches, but are expensive, highly sensitive to external noise and environmental variations, have low signal-to-noise ratio (SNR), and poor glucose selectivity. Although developments in Near-Infrared Spectroscopy (NIRS) have overcome these limitations to a certain extent, they do not produce reliable measurements due to large calibration errors that often result in incorrect glucose readings. In this paper, we propose a robust particle-swarm optimization-based artificial neural network for non-invasive continuous glucose monitoring using the principles of NIRS. We show that the PSO-ANN approach outperforms the traditional backpropagation algorithm used in ANN training and several other regression algorithms with the lowest error metrics: MAE- 1.01, MSE-2.16, RMSE-0.97, R-sqaured score -0.976 and modified R-squared score -0.973. The paper also provides insights into the circuit design, sensors used, hardware-software integration, and clinical validation alongside providing an overview of HbA1c computation.The accuracy and reliability of the proposed system are analyzed using the Clarke Error Grid (CEG) with 93.9% of the obtained readings falling within zone A and 100% of the readings falling in the clinically accepted range (zones A and B). The paper also explores potential enhancements such as miniaturization of the prototype device for wearable applications and wireless connectivity.Â

In this paper, the performance of a  two-hop cooperative network is analyzed assuming the use of a channel-aware dynamic time-switching (TS) energy harvesting (EH) protocol at the battery-assisted EH full-duplex relay node. In such networks, node-level energy considerations are important for efficient battery utilization but have not attracted research attention. Two battery management schemes are considered. In the first scheme, fixed  battery energy (FBE) is added to the harvested energy.  To further improve performance, a new dynamic battery energy (DBE) scheme is considered in which the amount of battery energy drawn is channel dependent. The performance of the network is analyzed with both FBE and DBE schemes for static and dynamic TS schemes assuming a practical nonlinear EH model. Expressions are also presented for the average battery energy consumption in each case. It is demonstrated that joint optimization of  the average battery energy consumption and the TS parameter is important. It is also demonstrated that the proposed dynamic TS protocol provides substantial gains in throughput and average energy savings compared to the static TS protocol with both FBE and DBE.  The derived insights can aid a system designer. Computer simulations validate the derived expressions.

As wireless optical communication becomes a key technology in many applications such as internet-of-underwater-things (IoWT) and inter-satellite communication, investigation of different modulation schemes has become essential. In this paper, we address the issue of whether it is possible to wirelessly transmit a modulated light signal at a frequency beyond the signal’s band-limit. We show that by using a superoscillation function as the amplitude modulating signal, it is possible to wirelessly transmit the optical signal through various media such as free-space, underwater and even turbulent underwater, and still preserve the superoscillatory nature of the signal with the property that it contains a component that oscillates faster than the signalâ\euro™s highest Fourier component. We also present cases where the received optical signal is distorted and quantify distortionÂ

This manuscript was submitted to IEEE/ACM Transactions on Audio, Speech and Language Processing and is currently undergoing review. In this work, we propose a fully adaptive kernel function for interior sound field interpolation that considers both directed and residual sound fields and always satisfies the Helmholtz equation. The method accomplishes this by assigning each component a different kernel function. The directed field represents sound field components of intense directionality that are sparsely distributed, and is represented by a superposition of kernels with strong directionality. The residual field represents the lower amplitudes and has much less predictable behavior, and thus was assigned a neural network weighted kernel function. We compared the proposed kernel to competing kernel formulations in numerical simulations and in real data experiments.

In the face of escalating global population, diminishing arable land, and an increasing demand for organic food, innovative farming systems are imperative. Traditional farming, characterized by excessive chemical usage, poses health and environmental concerns. Addressing these challenges, this research introduces a sophisticated system that amalgamates image processing, deep learning, and precision nutrient dosing for efficient soil-less farming. Utilizing a dataset of 24 GB, encompassing 4187 images of 11 plant varieties across three growth stages, a deep learning model was developed for automatic plant species identification, health assessment, and growth stage detection. The system’s prowess is exemplified with Butterhead lettuce, where the model’s insights guide nutrient dosing in an aeroponic tower. Integrating Arduino Uno for data acquisition and actuator control, with Central Computing Unit (Orange Pi 5) for backend management and deep learning application, the system’s architecture is robust and comprehensive. An Android application complements the system, offering real-time sensor data visualization, aeroponic tower initialization, and plant location tracking. The system also provides location-based growth recommendations, greenhouse-specific advice, and companion planting suggestions. In a month, the system projected a yield of 40 plants across 11 varieties. Drawing insights from various research papers, this system epitomizes the fusion of technology and agriculture, offering a promising solution for controlled environment agriculture and precision farming. This research not only advances soil-less farming practices but also addresses the increasing demand for organic food, presenting an efficient solution for large-scale cultivation.

Bila, a constructed language, represents a significant breakthrough in human-machine interaction, particularly in the field of robots within the context of Industry 4.0. This article outlines the author’s ambitious exploration of the three comprehensive stages involved in creating and implementing the Bila language based on Parallel Vietnamese Script 4.0 (CVNSS4.0), as well as integrating it into Bila Bot, a multifunctional humanoid robot software. Inspired by the Latin alphabet and Vietnamese language - CVNSS4.0, which was passionately developed by the authors (Kieu Truong Lam & Tran Tu Binh), the name “Bila” (shortened from Binh â\euro“ Lam) is constructed on a structured foundation for efficient communication. This article emphasizes the importance of clear vocabulary and grammar in the initial stages like CVNSS4.0, followed by the integration of simulated Vietnamese sounds and the creation of Bila Bot. It explores potential applications and benefits of Bila in the field of robotics, addressing challenges and limitations in implementation. Ultimately, the Bila language signifies a promising future for human-robot interaction and inter-robot communication, enabling more visual and efficient communication.

We propose a novel technique for identifying evoked components by using prior information of the stimulus time course as a guiding factor, allowing for modeling of trial variability, in a regularized optimization framework.

This study proposed a method for interpreting immunohistochemistry (IHC) images based on a graph convolutional network (GCN). Self-supervised transfer learning was employed to obtain cell nucleus segmentation masks, providing effective strong cues for a cell nucleus graph (CN-G). This study applys a GCN to end-to-end diagnostic classification tasks for IHC images, fully considering global distribution features and local details in images. We believe that our study makes a significant contribution to the literature because the proposed approach ensures high accuracy in the relevant tasks while addressing the challenges of the lack of labeled datasets and high number of sample pixels.

A document by Joonas Lahtinen . Click on the document to view its contents.

Diabetes is a metabolic disorder often diagnosed late and needs continuous blood glucose monitoring. We introduce GlucoBreath, a user-centric, cost-effective, and portable pre-diagnostic solution to address this global challenge. GlucoBreath addresses the urgent need for an accessible and non-intrusive diabetes detection device, offering affordability, mobility, and comfortable non-invasive diabetes testing, especially among economically weaker sections of society. GlucoBreath comprises (i) a non-intrusive multi-sensor Internet of Things device comprising multiple sensors detecting volatile organic compounds in breath, (ii) BreathProfiles dataset encompasses information from 492 patients, which includes demographic details, physiological measurements, and sensor readings derived by analyzing breath samples with our device, (iii) an innovative Machine Learning-based diabetes prediction system trained on the BreathProfiles dataset, and (iv) a user-friendly web interface for seamless device interaction and viewing diabetes reports. Given a person’s breath sample, demographics, and body vitals data as input, GlucoBreath predicts (a) if the person has diabetes. (b) If the person has diabetes, then the blood glucose level (BGL) of the person is moderate or high. GlucoBreath’s groundbreaking approach supersedes current methods, achieving an impressive mean accuracy of 98.4% using a Logistic Regression-AdaBoost stack-metamodel, marking a substantial 43.3% improvement over an existing method. Due to its portability, non-intrusiveness, and rapid response, GlucoBreath is a valuable pre-diagnostic tool that can facilitate the early detection of diabetes in many individuals. Further, the BGL prediction by GlucoBreath can help alert individuals to control their sugar consumption in case of a moderate BGL or visit a physician in case of a high BGL. Â

Disease forecasting is a longstanding problem for the research community, which aims at informing and improving decisions with the best available evidence. Specifically, the interest in respiratory disease forecasting has dramatically increased since the beginning of the coronavirus pandemic, rendering the accurate prediction of influenza-like-illness (ILI) a critical task. Although  methods for short-term ILI forecasting and nowcasting have achieved good accuracy, their performance worsens at long-term ILI forecasts. Machine learning models have outperformed conventional forecasting approaches enabling to utilize diverse exogenous data sources, such as social media, internet users’ search query logs, and climate data. However, the most recent deep learning ILI forecasting models use only historical occurrence data achieving state-of-the-art results. Inspired by recent deep neural network architectures in time series forecasting that benefit from self-attention, this work proposes the Regional Influenza-Like-Illness Forecasting (ReILIF) method for regional long-term ILI prediction. The proposed architecture takes advantage of diverse exogenous data, that are, meteorological and population data, introducing an efficient intermediate fusion mechanism to combine the different types of information with the aim to capture the variations of ILI from various views. The efficacy of the proposed approach compared to state-of-the-art ILI forecasting methods is confirmed by an extensive experimental study following standard evaluation measures.

Long-term time-series forecasting (LTSF) has received an increasing attention for its significant challenges and real-world applications. However, the previous studies under-explore the hierarchical timestamp information in LTSF. This information is crucial, especially for LTSF as failing to incorporate it may result in missing the global perspective of time series and important long-term trending effects, such as weekly and seasonal patterns. Therefore, we propose an interpretable hierarchical model called VH-NBEATS, which advances the N-BEATS model by addressing the aforementioned problem. VH-NBEATS comprises two essential blocks: the hierarchical timestamp block and the harmonic seasonal block to capture multi-diluted and trending effects. To address the high variability of time series, VH-NBEATS involves a stochastic autoencoder which significantly improves the standard deterministic approach. The experimental results are evaluated on five real-world datasets, showing state-of-the-art results for LTSF. We also prove that the VH-NBEATS framework can be easily incorporated into other ones, such as PathTST, leading to enhanced performance.

This paper introduces Continual Learning for Multilingual ASR (CL-MASR), a benchmark for continual learning applied to multilingual ASR. CL-MASR offers a curated selection of medium/low-resource languages, a modular and flexible platform for executing and evaluating various CL methods on top of existing large-scale pretrained multilingual ASR models such as Whisper and AWavLM, and a standardized set of evaluation metrics.

Signal decomposition techniques aim to break down nonstationary signals into their oscillatory components, serving as a preliminary step in various practical signal processing applications. This has motivated researchers to explore different strategies, yielding several distinct approaches. A wellknown optimization-based method, the Variational Mode Decomposition (VMD), relies on the formulation of an optimization problem, utilizing constant bandwidth Wiener filters. However, this poses limitations in constant bandwidth and the need for constituent count. In this paper, a new method, namely Dynamic Bandwidth VMD (DB-VMD), is proposed to generalize VMD by addressing the Wiener filter limitations through enhancement of the optimization problem with an additional constraint. Experiments in synthetic signals highlight DB-VMDâ\euro™s noise robustness and adaptability in comparison to VMD, paving the way for many applications, especially when the analyzed signals are contaminated with noise.

Feature fusion is an effective solution for improving image retrieval performance. Although the more feature types, the better accuracy, complexity also increases. Applications in practice typically afford a limited number of feature types. Due to the strong complementarity, global and local features form an ideal combination for many fusion applications. However, the two kinds of features are intrinsically different in nature, thus cannot be fused in a straightforward way. In this work, we propose an integrated image retrieval and feature fusion framework for global and local features. It is based on inverted index fusion, a technique for efficient image retrieval. The core idea is to rank candidates by weighted voting during candidate selection, which is named pre-ranking. This procedure takes place before re-ranking, and is potentially superior to conventional late fusion. Extensive experiments on three public datasets show that the light-weight pre-ranking stage significantly contributes to accuracy, and brings substantial improvement when used together with re-ranking. Our method is robust and versatile, and can be applied to any scenario where inverted indexing is used. It is a promising technique for multimedia retrieval in the big data era.

Intelligent behavior is an emergent phenomenon observed in biological organisms across all scales. It describes the cooperative behavior of low complexity entities to accomplish complex tasks, which exceed their individual capabilities. This property is particularly important for the Internet of Bio-Nano Things (IoBNT), which consists of Bio-Nano Things (BNTs) used in the human body, where they face many restrictions, such as bio-compatibility and size constraints. In this paper, we present a novel BNT-architecture, called Molecular Nano Neural Networks (M3N), which allows the implementation of intelligence on the micro-/nano-scale. The proposed structure consists of compartments (low complexity entities) that are connected to each other to form a network. Based on reaction and diffusion of molecules in and between connected compartments, this network mimics an artificial neural network, which is an important step towards  artificial intelligence in the IoBNT. We provide design guidelines for the proposed M3N and successfully validate it by applying a regression and classification task.

In this letter, we consider a secondary network (SN) in which an ambient backscatter device (BD) utilizes the secondary transmitter (ST) signal to communicate its own information to the secondary destination (SD). Optimizing performance of such networks is complicated by signal reflections by the BD. It is shown in this work how the secondary transmit power and the reflection coefficient of the BD can both be jointly optimized using a simple restricted one dimensional search to satisfy the quality of service (QoS) constraints of SN as well as the primary network (PN) while maximizing performance of the backscatter link, which is termed the tertiary network (TN). Only statistical channel knowledge is used for this purpose. It is seen that careful optimization can improve spectral efficiency. Simulations validate the derived analytical expressions.

Unidentified continous functions are generalized classification for all signals. Author assumes that any unidentified continous function (any signal) as to be either digital (non-archimedian continuum) or analog (archimedian continuum). Author mainly founded a method to indicate a distinction for both continuum models. This method therefore can be used on to all unidentified continous functions in order to identify the continuum models of unidentified continuous functions.

BREN-Tech deals with the advancement of human brain and human body.It is a more effective and faster way where a chip like micro/macro machine is inserted into the body and many procees takes place in a short time.It involves minute comonents and can be connected to external systems as well.It can help towards the growth of human evolution which can be equal or more than the advancement of AI in the modern world

Voice likability is a critical factor in machine-human interaction. However, studies on speech likability typically does not apply the harmony theory in music, which suggests general rules for pleasant sounds. In this paper, I propose a new method that estimates the likability of vocal signals using the harmonic relation of pitch and the first formant (F1). I extract the pitch and F1 from the vowel signal and compute the average cent value between notes in the musical scale from each pitch and F1. A small cent value indicates a consonant relation between pitch and F1. I compared the calculated cent values with the MOS test results from ten speech samples. The results showed a clear correlation between the subjective MOS scores and the consonance of pitch and F1 in vowels.