We show for the first time that when a linearly polarized laser light incidents either a grating or a double slit, the output light is non-polarized. Namely, both grating and double slit depolarize the linearly polarized laser light.

In this work, we have proposed bit-interleave SRAM-based IMC architecture for AI edge devices.

This manuscript is the preprint version of a conference paper that is accepted for presentation and publication, International Conference on Future Energy Solutions (FES2023) to be held in Vaasa, Finland, 12-14, June 2023.

As artificial intelligence (AI) systems become more prevalent in various domains, ensuring their trustworthiness and efficiency becomes increasingly important. In this article, we present an analysis of the tradeoffs between different dimensions of trustworthy AI, including transparency, robustness, fairness, accountability, and efficiency. We further discuss the challenges and opportunities that arise when striving for a balance between these dimensions. We also offer recommendations for achieving an optimal balance. By understanding and addressing these tradeoffs, we can foster AI systems that perform efficiently but also maintain user trust and societal values.

In this paper, we propose a true full-duplex cell for space-constrained high-speed interconnects. The proposed interconnect can replace conventional simplex interconnects due to its ability to double the data rate with ease in PCB floor planning. The proposed ratioed driver-based self-interference (SI) cancellation technique uses an automatically adjustable on-chip background calibration which minimizes SI, to retrieve the far-end signal faithfully. The techniques used in the design relax linearity requirements and reduce power consumption for the transceiver. The SI cancellation is independent of the modulation format of the incoming signal from the far-end transmitter. Using the proposed technique reliable communication is possible between near-end and far-end transceivers without any synchronization. The proof-of-concept circuit has been designed and verified in a low-leakage 65 nm CMOS technology. The full-duplex nature of the transceiver can significantly reduce cross-talk in high-density interconnects compared to its simplex counterparts for the same throughput due to less stringent spacing requirements.

A new PRACH detection algorithm is proposed to process the Power Delay Profile of PRACH signal. The proposed algorithm not only signifies the peak magnitude (therefore boosts the detection probability in the low-SNR scenario) but also reduce the false peaks.

Dynamic Spectrum Sharing (DSS) is an enabler for a seamless transition from 4G Long Term Evolution (LTE) to 5G New Radio (NR) by utilizing existing LTE bands without static spectrum re-farming. In this paper, we propose a cross-band DSS scheme that utilizes the Multimedia Broadcast Multicast Service over a Single Frequency Network (MBSFN) feature of an LTE network and the Multicast Broadcast Service (MBS) feature of an NR network. The proposed DSS scheme utilizes LTE and NR resource controllers to assign muted MBSFN subframes on the LTE band and muted MBS subframes on the NR band based on traffic needs. In contrast to the state-of-the-art, the proposed DSS scheme does not require a coordination signaling channel between the LTE and NR networks. Instead, a machine learning-based Technology Recognition and Traffic Characterization (TRTC) system is used to identify and characterize traffic patterns. The LTE and NR resource controllers use the TRTC to sense the muted subframes and offload traffic accordingly.

The heat pump system is complicated to model dynamically because of the phase changing in both evaporator and condenser. The  moving boundary model tackles this problem by dividing the evaporator and condenser into variable-length zones, and their lengths  vary with the thermal state of the refrigerant. This document explains the full derivation process of the moving boundary model of an air source heat pump system.

Identifying and preserving community structures in a streaming graph is a very challenging task. However, many applications require the identification of these communities in very limited space and time. In this paper, we design Community Sketch, a small space data  structure that efficiently preserves communities. On query, it provides communities in constant time. With the use of community sketch data structure, a linear streaming community detection algorithm is proposed. Experimental results on the large real-world networks show that our algorithm outperforms other state-of-the-art algorithms in terms of quality metrics (NMI, F1-score, and WCC). Further, we propose an algorithm to produce benchmark network, namely, Temporal Community Benchmark Dataset (TCBD) which contains both true community labels and temporal information of edges. These synthetic networks are used to validate the proposed algorithm

In this study, we explore the potential of using a straightforward neural network inspired by the retina model to efficiently restore low-light images. The retina model imitates the neurophysiological principles and dynamics of various optical neurons. Our proposed neural network model reduces the computational overhead compared to traditional signal-processing models while achieving results similar to complex deep learning models from a subjective perceptual perspective. By directly simulating retinal neuron functionalities with neural networks, we not only avoid manual parameter optimization but also lay the groundwork for constructing artificial versions of specific neurobiological organizations.

This paper presents the MagNet-AI platform as an online platform to demonstrate the “Neural Network as Datasheet” concept for B–H loop modeling and material recommendation of power magnetics across wide operation range. Instead of directly presenting the measured characteristics of magnetic core materials as time sequences, we employ a neural network to capture the B–H loop mapping relationships of magnetic materials under different excitation waveforms at different temperatures and dc-bias. Both LSTM and Transformer based neural network models are developed, verified, and compared. The training and inference process of the neural network are fully automated to minimize the impact of human error. The neural network can be used to rapidly predict B–H loops under different operating conditions, compare materials, and recommend materials for design. Neural networks are also effective in compressing the information contained in the raw database, and enable rapid material evaluation and comparison.

This paper has the sole purpose of emphasizing the fact that the movements of the human heart hold an important position in the field of heart ailment diagnosis. Our study brings out a new approach to the analysis of heart movements using machine learning. We have used the deep learning algorithms ResNet and BiLSTM for the classification and segmentation of the images from the video inputs and Explainable AI techniques LIME and SHAP have been applied in order to increase the interpretability and predictability of the model. During the training and testing phases, the downstream values of the tasks had been modified into the pre-trained set which was projected to self-supervised learning through the integration of SimCLR layers. The values thus obtained after the analysis of the pre-trained set helped us to evaluate the working of the model while predicting the next outcomes from a sequence of inputs. After the successful integration of these results with the current sequence of inputs, we have observed that the results generated by the model showed an increased learning rate of le-4 and the predictive outcomes becoming better with time. The highest accuracy value has been observed to be 92.5%. The values of the precision, recall, and f1 score also showed increasing trends with the passage of epochs and the final result suggested that the model is sufficient to provide proper medical assistance in terms of automation of the heart movement detection and it would be helpful for the early diagnosis of heart ailments.

Most existing methods for analyzing reconfigurable intelligent surface (RIS) channels are limited to free-space links and do not account for the mutual coupling of unit cells, or they fail to accurately model several types of realistic scenarios including near-field operation and interaction of RISs with rich multipath environments. Alternatively, some methods that can handle these scenarios are computationally intensive, such as pure full-wave analysis. We present an equivalence principle-based hybrid ray-tracing/full-wave method to model wave propagation in wireless communication channels enabled by RISs. This method uses ray-tracing to determine the incident waves on an RIS, then applies full-wave analysis to model the response of the RIS to these waves. Next, based on the equivalence principle, we introduce equivalent surface electric and magnetic current densities that generate the scattered fields produced by RIS. These equivalent sources are integrated with ray-tracing to derive the site-specific propagation model of the RIS channel. The proposed method readily accounts for multiple incident waves on an RIS, enabling the accurate analysis of propagation channels with RISs, with receivers in both the near and the far region of an RIS. We show that the accuracy of our method is comparable to that of full-wave analysis, through simple examples that are manageable by the finite-element method. Also, we experimentally validate the proposed technique by comparing simulation and measured data for an actual indoor radio environment with an anomalous reflection metasurface.

The residential services in the context of smart cities accu- mulate massive real-time inquiry data in natural language to describe the services in need. Such inquiry requests have diverse topics in the content and considerable variate length. Besides, the responsible departments that may handle the inquiry involve a large number of organizations, from metropolitan administration to local communities. Hence accumu- lated request data is primary and central to the service of accurately dispatching requests to responsible departments. The challenge is de- vising a data centric approach to fit the data with SOTA models and improve the request classification accuracy. In this paper, we analyze the factors of embedding tokens, data segmentation, model structures, and classification methods. We devise a unified modelling process with mul- tiple dataflows that combine these factors to observe their interactions. The experiment results demonstrate the compound effects and provide insights into how SOTA models respond differently to variations in these factors. The observations allow us to fine-tune the learning task at each stage and achieve a maximum 82.4% F1-Score.

Rapid progress in various fields of science and technology in recent years has caused various adverse and predictable negative effects on the environment. Today, one of the most important biological problems is environmental electromagnetic pollution. The intensity of its electromagnetic radiation has become so widespread that it is currently known as one of the hidden and silent forms of pollution. One means of communication that is an inseparable part of daily life is the mobile phone. Research and surveys show that mobile phone waves have a negative effect on the human body. Our goal in this article is to study the effects of mobile phone electromagnetic waves on the human eye. In order to check the effect of these waves, the specific absorption rate (SAR) criterion is used. Because the rate of specific absorption in the eye is determined according to the absorption of energy in the eye tissues and the conversion of this absorbed energy into thermal energy, they increase the temperature of the eye tissues. In other words, the temperature distribution in the eye directly correlates with the rate of specific absorption. The results have shown that the tissue of the eye is susceptible. When the eye is exposed to waves, the eye’s temperature increases, and the antenna angle relative to the eye is effective in the rate of specific absorption. According to studies, the conclusion of this research indicates that without observing safety points, including maintaining distance, they can cause dangerous bodily complications.

The work is devoted to the search for new vacuum-plasma coatings with high hardness to increase the durability of the compressor blades of the GTE of aircraft engines Ti-Al-N-based vacuum-plasma coatings obtained by Avinit technologies, which ensure the application of hard, high-quality coatings with dramatically reduced micro-arc damage, were selected as candidates. Avinit multilayer coatings have higher functional characteristics than TiN (microhardness, crack resistance, temperature resistance, erosion and corrosion resistance) and may be promising for applying erosion-resistant coatings for compressor blades. Avinit technologies are technologically closest to the vacuum-plasma technologies used in industrial production for applying TiN protective coatings. New multi-layered 2D nanocomposite wear-resistant ion-plasma hard coatings Avinit (TiN-AlN)n have been developed. The created software products made it possible to reach a qualitatively new level in terms of further modification and improvement of the designs of Avinit functional coatings, stability of technologies and improvement of their quality control when applying such coatings for use in the production of compressor blades of gas turbine engines of aircraft engines. Special attention is paid to methods of preliminary ion-plasma treatment of surfaces before coating. Metallographic studies of the chemical and phase composition and structure of Avinit (TiN-AlN)n coatings have been carried out. The thickness of the coatings is 7–9 μm, the microhardness is 34–35 GPa (compared to the serially used TiN coating: 27.4 GPa). The use of three-stage ion-plasma treatment in Avinit technologies using a double vacuum-arc discharge followed by the application of strengthening coatings in a single technological cycle eliminates the formation of cracks and ensures the production of tightly bonded, high-quality coatings of a given composition with the maximally reduced share of the droplet component. The developed coatings (TiN-AlN)n were applied to experimental batches of working compressor blades of GTE aircraft engines for bench tests.

In this paper, two open problems of the existing bounded UDE-based robust control method are investigated: the input constraint with multiple control inputs, and the performance degradation caused by the introduced additional time-varying control variable. A modified bounded UDE-based robust control method is developed for the MIMO system with the inputs constrained by a general positive defined matrix, which is a more general condition of the norm constraint for control inputs. A nonlinear designed is further proposed to modified the direct multiplication of the additional time-varying variable to mitigate the performance degradation of the existing bounded UDE-based robust control, and the guidance of the nonlinear design procedures is provided. The proposed design is applied to the vector control of a PMSM drive system, and the simulation studies are provided to validate the proposed design under voltage constraint of the vector control with both normal operation and voltage constraint triggered.

We  consider active hypothesis testing with multiple heterogeneous agents. Each agent has access to its own set of experiments, has different action costs and forms its own beliefs. Additionally,  each experiment carries a global cost, and the agents must try to keep the expected cumulative cost below a certain threshold. We study a centralized and a decentralized scenario. Under the centralized scenario, the agents  are instructed how to act by a central controller. Under the decentralized scenario they communicate and exchange information over a directed graph. For each scenario we propose  separate deep reinforcement learning algorithms based on proximal policy optimization. Solutions to the decentralized problem start from fully decentralised training, and progressively introduce two levels of centralisation during training.  We assess the proposed algorithms in an example of anomaly detection over sensor networks, considering three different decentralised communication settings. We infer that all algorithms achieve the required accuracy level considerably faster than a single deep reinforcement learning agent, while satisfying the expected cost constraints when required. Under the assumption that the communication  graph is fully connected, the decentralised agents perform just as well, and sometimes better than the centralised  controller. Out of all decentralised algorithms, the one that uses a global critic is by far the best performing and can compete with the central controller even when the communication graph is not complete.