Generative Artificial Intelligence (AI) represents a significant advancement in AI, enabling the creation of synthetic data that closely mimics real data. This article provides a comprehensive overview of generative AI, including its definition and core concepts such as Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and autoregressive models. It explores applications across industries, including content creation, data augmentation, healthcare, gaming, fashion, finance, retail, cybersecurity, transportation, energy, education, entertainment, law, agriculture, real estate, and manufacturing. The article discusses future prospects of generative AI, highlighting enhanced creativity, improved human-AI collaboration, and personalized experiences. It also addresses dependencies and challenges, including data quality, computational resources, and ethical considerations. The benefits for businesses adopting generative AI are examined, emphasizing competitive advantage, enhanced customer experience, accelerated innovation, and improved decision-making. The article concludes by outlining the potential return on investment through increased efficiency, revenue growth, market expansion, and risk mitigation.

The steadily increasing number of wireless Internet of Things devices imposes new challenges on spectrum monitoring and spectrum management tasks. In this paper, we design a multifunctional cooperative sensor network to analyze electromagnetic (EM) situation in the monitored area. Specifically, we introduce a framework based on a long short-term memory (LSTM) deep neural network (DNN) to jointly localize a single transmitter (TX) and classify the waveform of the TX signal. The performance capability of separating fifth-generation New Radio (5G NR) communications standard signals from fourthgeneration Long Term Evolution (4G LTE) communications standard signals and Wi-Fi signals is investigated as a classification problem. Furthermore, the associated TX localization accuracy is measured regarding mean squared error (MSE). We show that in a basic scenario, it is possible to yield a reasonable TX localization performance without any prior knowledge of the signals' structure with just four sensors, in parallel getting reliable wireless waveform classification. Finally, we suggest future research directions for joint signal classification and TX localization in cooperative sensor networks.

The aim of this study, therefore, is to reinvent the future of computing systems in terms of performance, efficiency, and adaptability by identifying the "frontier" of AI-driven innovations in CPU design. * This document surveys front-end optimizations driven by AI, new architectural ideas, and upcoming paradigms that might revolutionize CPU technology through a thorough literature review as well as empirical evaluations. * With the help of advanced artificial intelligence methods like as machine learning, deep learning, and reinforcement learning, the exploration of new CPU architectures and optimization techniques permits the unleashing of computational powers beyond anything previously imagined. * AI methods on CPU architecture and address significant concerns such as scalability, energy efficiency, security, and reliability have been systematically explored. Furthermore, the transformative ability of AI-driven CPUs is also illustrated through real-world-case applications in several fields, e.g. autonomous systems, healthcare informatics, edge computing. This paper helps clear the roadmap for the next system of creative revolution in AI-imbued computing, defining potential directions for additional research and cooperation.

In conventional remote sensing change detection (RS CD) procedures, extensive manual labeling for bi-temporal images is first required to maintain the performance of subsequent fully supervised training. However, pixel-level labeling for CD tasks is very complex and time-consuming. In this paper, we explore a novel self-supervised contrastive framework applicable to the RS CD task, which promotes the model to accurately capture spatial, structural, and semantic information through domain adapter and hierarchical contrastive head. The proposed SSLChange framework accomplishes self-learning only by taking a single-temporal sample and can be flexibly transferred to mainstream CD baselines. With self-supervised contrastive learning, feature representation pre-training can be performed directly based on the original data even without labeling. After a certain amount of labels are subsequently obtained, the pre-trained features will be aligned with the labels for fully supervised finetuning. Without introducing any additional data or labels, the performance of downstream baselines will experience a significant enhancement. Experimental results on 2 entire datasets and 6 diluted datasets show that our proposed SSLChange improves the performance and stability of CD baseline in data-limited situations. The code of SSLChange will be released at https: //github.com/MarsZhaoYT/SSLChange

Malaria is a serious and potentially deadly vector-borne disease that affects millions of people around the world. According to the latest report by the World Health Organization, there are an estimated 247 million cases of malaria worldwide. Malaria is caused by Plasmodium parasites and is transmitted by the bite of Anopheles mosquitoes. Early detection of Malaria is crucial to detect to treat the same early and prevent any potential spread that may cause serious fatalities. In the past, there are many approaches reported that uses machine learning approaches for early detection of Malaria from images. But still there are avenues where the detection accuracies are less and may need more advanced models for easy and accurate detection. This work proposes an approach for automated classification of Malaria from cell images using Convolutional Neural Networks (CNN). When compared with some state-of-the-art approaches, our proposed approach shows better classification accuracy and outperformed them.

Reversely computed dynamic temporary weights introduce a novel and significant enhancement to the adaptability and accuracy of large language models. By dynamically recalculating the weights of key hidden layers during inference, our methodology significantly improves the model's performance across various natural language processing tasks. Experimental results demonstrated substantial increases in accuracy, response time, and computational efficiency when compared to the baseline performance. The integration of dynamic weights enabled the model to adjust its internal parameters in real-time, resulting in more precise and context-aware predictions. Statistical analysis confirmed the significance of these improvements, providing robust validation for the proposed enhancements. This research not only advances the state-of-the-art in language model optimization but also paves the way for more intelligent and adaptable AI systems. Future work will address computational overhead and explore broader applicability to other neural network architectures.

Integrated electrical and heating systems(IEHS) are drawing huge attention because of their cost-effective and flexible operational capabilities. Optimal dispatch of renewable units in an IEHS is challenging because of their intermittent and stochastic nature. Uncertainties in the renewable energy source (RES) forecast, electrical, and thermal loads can cause voltage and temperature violations compromising the operational security of the system. In this paper, a rule-based dynamic dispatch is proposed to handle the increasing levels of uncertain RES and ensure the system's security. A chance-constrained (CC) co-optimization algorithm considering the uncertain loads and RES is developed. In the proposed method, electric boiler power and curtailment of RES are dynamically dispatched to reduce operational costs and network losses. The developed method is tested on the IEEE 15 bus distribution system and UK Generic Distribution System (UKGDS) 95 bus test system models. The results show that the proposed method significantly reduces the voltage deviations and temperature violations compared to the deterministic case.

A general-purpose instrumentation amplifier must be dc-coupled and have a differential input to handle both differential and single-ended input signals. It also must exhibit low input noise in both voltage and current, to accommodate a wide range of signal source impedances. Additionally, having a differential output is desirable to allow direct connection to current high-resolution analog-to-digital converters (ADCs) which have differential inputs. There are commercially available devices with \(e_n\) voltage noise spectral densities as low as 1 nV √Hz ⁄ but present high \(e_n\) current noise spectral densities \(i_n\) of a few pA/√Hz. On the other hand, there are also devices with \(i_n\) as low as a few fA/√Hz but presenting  \(e_n\)   around 10 nV/√Hz. To obtain low values of both \(e_n\) and \(i_n\) , a fully differential circuit topology combining discrete Junction Field Transistors (JFET) and Operational Amplifiers (OA) is proposed. Design equations, stability analysis, and experimental results are presented. As an example, a fully differential instrumentation amplifier has been designed, built, and tested showing \(e_n\) <1 nV/√Hz @1 kHz and \(i_n\) <10 fA/√Hz @1 kHz. The proposed topology finds applications such as front ends for measuring and testing instruments, industrial instrumentation, and audio circuits.

The exponential increase in Internet of Things (IoT) devices coupled with 6G pushing towards higher data rates and connected devices has sparked a surge in data. Consequently, harnessing the full potential of data-driven machine learning has become one of the important thrusts. In addition to the

Researchers have sought solutions to interconnect different radio access technologies, enabling them to coexist in the current and future heterogeneous network environments to obtain additional benefits. This paper presents SVORA, an innovative approach that integrates the concept of Virtualised/Open (V/O)-Radio Access Network (RAN) with a Service-Based Architecture (SBA) in order to optimize the energy consumed by the network and minimize the communication delay. In the context of SVORA, the Delay-aware Energy Efficiencybased RAN selection algorithm (DEER) is proposed. DEER employs utility functions and considers Quality of Service (QoS) metrics such as throughput, resource block utilization, delay, and energy consumption measurements when finding the RAN to best support the requested service. DEER is highly flexible, being able to be tuned to support critical delay services, prioritizes energy saving, or observe a balanced approach. Testing shows that the SVORA-based DEER was able to increase the number of users and the aggregate data rate by about 10%, whereas the delay and energy consumption decreased by approximately 90% and 50%, respectively.

Wireless sensor network (WSN) has found widespread applications in many domains such as smart home and healthcare systems. Group communication for WSN offers enhanced energy efficiency and reduced network overhead. However, existing group communication protocols encounter difficulties in managing the rekeying process efficiently for a dynamic group or require computationally expensive public key primitives for shared secret handshaking to overcome this challenge. Moreover, the security of most of the existing protocols rely primarily on safe keeping some secrets on the group members' devices. To overcome these limitations, we propose a novel physical unclonable function (PUF)-based lightweight secure group communication protocol. PUF enables lightweight key-less authentication for resource constrained IoT devices. The proposed protocol utilizes the device's PUF and the one-time pad (OTP) to eliminate secure key storage at both group verifier and prover nodes. The proposed protocol also supports efficient group key renewal for the admission and withdrawal of group members. We achieve this by using a full binary tree as a secret vault for sharing and updating distributed secrets with Chinese Reminder Theorem. Meantime, this data structure also reduces the computation and communication complexity for key renewal to O(log 2 N) at the cluster head and O(1) at the sensor nodes. A comparative analysis shows that the proposed protocol surpasses related protocols in terms of security features and overheads in computation, communication as well as secret storage requirements.

This document presents a simple method that allows the use of time-domain simulation to extract an AC small signal model of the power stage of a power converter. Such extraction can be important for the simulation of feedback loop gain and the stability of complex power converter topologies, such as resonant converters or converters with complex switch control. With the presented approach, a single simulation run in the time domain is sufficient to extract a small signal model, resulting in a time-efficient extraction process. The method is based on the step response simulation, where the input control signal source (duty cycle, switching frequency, etc.) is applied as a step function. The output parameter of the converter (output voltage, current, etc.) is recorded during the time-domain simulation and then transformed into a SPICE netlist using Laplace transform statements. The Laplace transform in s-domain is supported by most SPICE simulators, such as LTspice, PSpice, etc. The presented algorithm and model are automatically generated using a developed open-source software tool. Thus, the free software LTspice, along with the presented open-source tool, can be used as a free alternative to commercial SPICE programs like PSpice, PADS, Simetrix, MATLAB, etc.

Massive MIMO (mMIMO) communication systems are recognized as key enablers of next-generation wireless networks. However, the orthogonal pilot assignments typical of multiple-input multiple-output (MIMO) systems are not wellsuited to emerging use cases characterized by short channel coherence intervals and high number of connected user equipments (UEs). In this work, we propose a novel approach for intracell pilot reuse that leverages the spatial features of correlated mMIMO channels to attain low pilot contamination while using a small number of pilot sequences. The first part of the proposed solution is a groundbreaking clustering algorithm termed Kbeams which splits the complex intra-cell pilot allocation into tractable problems without significant loss of optimality. Then, we introduce a heuristic approach called best-first pilot assignment (BFPA) designed to manage intra-cluster pilot assignments by minimizing interference among the most contaminating UEs. We evaluate the performance of our proposed solution (K-beams+BFPA) in terms of sum-normalized mean-squared error (NMSE) and sum-rate under various challenging network setups. Simulation results show that our approach is a robust alternative to more computationally demanding benchmarks.

Safe road crossing by autonomous wheelchairs can be affected by several environmental factors such as adverse weather conditions influencing the accuracy of artificial vision. Previous studies have addressed experimental evaluation of multi-sensor information fusion to support road-crossing decisions in autonomous wheelchairs. In this study, we focus on the fine-tuning of tracking performance and on its experimental evaluation against outdoor environmental factors such as fog, rain, darkness, etc. It is rather intuitive that those factors can negatively affect the tracking performance; therefore our aim is to provide an approach to quantify their effects in the reference scenario, in order to detect conditions of unacceptable accuracy. In those cases, warnings can be issued and system can be possibly reconfigured to reduce the reputation of less accurate sensors, and thus improve overall safety. Critical situations can be detected by the main sensors or by additional sensors, e.g., light sensors, rain sensors, etc. Results have been achieved by using an available laboratory dataset and by applying appropriate software filters; they show that the approach can be adopted to evaluate video tracking and event detection robustness against outdoor environmental factors in relevant operational scenarios.

Antenna synthesis is becoming increasingly challenging given the tight requirements for C-SWAP (cost, size, weight and power) reduction while maintaining stringent electromagnetic performance specifications. While recent reports have investigated developments in AI (primarily deep learning) to address this challenge, such methods have limited capability for multi-objective constrained inverse design and the ability to handle large shape sets. We present a multi-objective antenna synthesis methodology employing deep neural network surrogate-driven evolutionary algorithm. Specifically, we propose a branched-output convolutional neural network architecture for creating the surrogate model and showing its ability to predict multiple electromagnetic parameters over an operating frequency range. The proposed methodology and its synthesis capabilities are showcased by considering the set of polygonal-patch antennas for operation in the 2 to 3 GHz spectral band. Specifically, we report the design, fabrication and experimental characterization of three polygon-shaped patch antennas, each fulfilling different objectives, one being a single band with resonance at 2.45 GHz, the second antenna is dual-band with resonance at 2.2 GHz and 2.9 GHz, and the third one being wide-band centered around 2.7 GHz. The reported methodology enables rapid synthesis (in seconds), produces verifiable sound designs and is promising for furthering data-driven design methodologies for electromagnetic wave device synthesis.

The energy associated to the elecromagnetic (EM) field due to thermal agitation in homogeneous ohmic media is evaluated. To this goal I propose an expansion of electromagnetic fields in terms of orthogonal modal functions that are eigen-vector of Maxwell's. These fields are recognized as the degrees of freedom of the field and the energy associated to each of them is represented using the Johnson thermal voltage sources associated to the dissipative part of the impedance of the modes. At low frequency the conduction currents are dominant and in phase with the agitation of the electrons: the thermal energy is thus efficiently transferred to the EM field. At higher frequencies the radiated displacement currents overshadow the conduction currents. However, since the displacement currents are out of phase with the agitation of the electrons, the thermal energy is not efficiently transferred to the EM field. The model, which is entirely analytical, first provides the thermally generated EM energy per unit of volume. Then, relying on a number of simplifications commonly adopted in radiometry, provides the energy radiated outside the warm bodies. This latter radiated energy is finally compared with recent measurements for moderate conductivity silicon presented in a companion paper [11]. The model's accuracy with respect to measurements is within 1 dB, while the application of Planck's law, even mediated by the emissivity, is clearly off target.

The thermal energy radiated by silicon wafers is characterized experimentally in the mm and sub-mm wave ranges. The samples are heated and the energy they radiate thermally is captured by horn antennas and then detected. The energy radiated is in the tens and hundreds of pico-Joule per second ranges. The Zero Bias Schottky Detectors adopted are characterized by Noise Equivalent Powers in the few pico-watts, rendering the measurements possible. For moderate conductivities the measurements show a descending pattern as a function of the frequency which is not in line with expectations from Planck's law. Should these experiments have been available 120 years ago, the community of the time would probably have been concerned about avoiding the THz catastrophe, rather than the Ultraviolet catastrophe! A companion paper provides a possible theoretical classical explanation of the results [1].

The implementation details of an electronic design automation tool using a novel reinforcement learning-assisted automated analog layout design flow (RALF) are presented. The tool generates the layout of an analog circuit with a SPICE netlist as the input, using the SkyWater Technologies SKY130 process design kit and the open-source VLSI layout tool Magic. Two different device placement strategies are implemented, one based on reinforcement learning, and the other based on simulated annealing. Automatic recognition and optimized treatment of structures like differential pairs, current mirror loads, and resistor strings are implemented using subgraph isomorphism. This tool also naturally supports hierarchical circuit topologies. After the placement, a two-stage routing algorithm connects the devices. The first stage is a wire planner, which plans the routing on a rough tile-based grid and uses a negotiation-based algorithm to eliminate estimated routing congestions. The second stage is a detailed router, which facilitates the previously planned guiding to lay out the resources in the routing space. A gridless approach, based on the expansion of obstacles in the routing area, enables this tool to route with variable wire widths as required by user-defined constraints.