arXiv subjects

Ankan Ghosh Dastider

Publications and source records attributed to Ankan Ghosh Dastider.

4 recordsLinked to original sources

arXiv2026

Hot LO Phonon-Induced RF Nonlinearity in GaN High-Electron-Mobility Transistors

Hot longitudinal optical (LO) phonons in GaN have recently been identified as a major factor degrading the DC performance of GaN high-electron-mobility transistors (HEMTs) by 30-60%, despite their ultrafast decay. However, their impact on large-signal RF performance, particularly RF linearity, remains poorly understood. Using full-band transport simulations of a fabricated GaN HEMT, we show that even ultrafast LO phonons with a lifetime of 30 fs degrade the output 1-dB compression point and the third-order output intercept power by ~3 dB compared to the case without LO phonon heating. Furthermore, our analysis reveals that improvements in transconductance ($g_\textrm{m}$) flatness do not necessarily translate into improved RF linearity because multiple nonlinear mechanisms contribute to the transistor response, and their combined effect cannot be captured by $g_\mathrm{m}$ flatness alone. This work clarifies a persistent ambiguity in the literature regarding using $g_\mathrm{m}$ flatness as a proxy for RF linearity and establishes intrinsic phonon-induced limits on the RF performance of GaN HEMTs.

physics.app-ph↗

arXiv2026

Physics-based Full-band GaN High-Electron-Mobility Transistor Simulation Suggests Upper Bound of LO Phonon Lifetime

Intrinsic limits to device performance arise from fundamental material properties that define the best achievable operation, independent of engineering constraints. In GaN high-electron-mobility transistors (HEMTs), hot longitudinal optical (LO) phonons can act as an intrinsic performance bottleneck by reducing electron saturation velocity, output current, and transconductance, which are key device metrics. While bulk GaN studies report LO phonon lifetimes of approximately 1 ps, leading to strong nonequilibrium phonon populations, ungated heterostructures show much shorter lifetimes of only tens of femtoseconds. Because direct measurement inside a HEMT channel is challenging, the true impact of hot phonons remains uncertain. Using full-band transport simulations of a fabricated GaN HEMT, we show that LO phonon lifetimes must be less than about 40 fs to reproduce measured I-V characteristics, consistent with ultrafast decay observed in GaN heterostructures. We further demonstrate that even these ultrafast lifetimes are not sufficient to eliminate hot phonon effects: the residual nonequilibrium LO population continues to limit the current density at high bias. Moreover, when the LO phonon lifetime exceeds a few tens of femtoseconds, a pronounced hot phonon bottleneck emerges, leading to substantial current-density suppression that is inconsistent with experiment.

physics.app-ph↗

arXiv2023

Topological, or Non-topological? A Deep Learning Based Prediction

Prediction and discovery of new materials with desired properties are at the forefront of quantum science and technology research. A major bottleneck in this field is the computational resources and time complexity related to finding new materials from ab initio calculations. In this work, an effective and robust deep learning-based model is proposed by incorporating persistent homology and graph neural network which offers an accuracy of 91.4% and an F1 score of 88.5% in classifying topological vs. non-topological materials, outperforming the other state-of-the-art classifier models. The incorporation of the graph neural network encodes the underlying relation between the atoms into the model based on their own crystalline structures and thus proved to be an effective method to represent and process non-euclidean data like molecules with a relatively shallow network. The persistent homology pipeline in the suggested neural network is capable of integrating the atom-specific topological information into the deep learning model, increasing robustness, and gain in performance. It is believed that the presented work will be an efficacious tool for predicting the topological class and therefore enable the high-throughput search for novel materials in this field.

cond-mat.mtrl-sci↗

arXiv2020

Privacy-Aware Activity Classification from First Person Office Videos

In the advent of wearable body-cameras, human activity classification from First-Person Videos (FPV) has become a topic of increasing importance for various applications, including in life-logging, law-enforcement, sports, workplace, and healthcare. One of the challenging aspects of FPV is its exposure to potentially sensitive objects within the user's field of view. In this work, we developed a privacy-aware activity classification system focusing on office videos. We utilized a Mask-RCNN with an Inception-ResNet hybrid as a feature extractor for detecting, and then blurring out sensitive objects (e.g., digital screens, human face, paper) from the videos. For activity classification, we incorporate an ensemble of Recurrent Neural Networks (RNNs) with ResNet, ResNext, and DenseNet based feature extractors. The proposed system was trained and evaluated on the FPV office video dataset that includes 18-classes made available through the IEEE Video and Image Processing (VIP) Cup 2019 competition. On the original unprotected FPVs, the proposed activity classifier ensemble reached an accuracy of 85.078% with precision, recall, and F1 scores of 0.88, 0.85 & 0.86, respectively. On privacy protected videos, the performances were slightly degraded, with accuracy, precision, recall, and F1 scores at 73.68%, 0.79, 0.75, and 0.74, respectively. The presented system won the 3rd prize in the IEEE VIP Cup 2019 competition.

cs.CV↗