About

Sandeep Gupta is a Professor of Electrical and Computer Engineering at the University of Southern California's Viterbi School of Engineering, within the Ming Hsieh Department of Electrical and Computer Engineering. He holds a doctoral degree in Electrical Engineering from the University of Massachusetts Amherst, earned in 1991, along with a master's degree in Computer Engineering from the same institution and a bachelor's degree in Electrical Engineering from the Indian Institute of Technology. His research focuses on design for testability, test, and validation of high-speed systems. Throughout his career, he has received numerous awards including the 2009 USC Distinguished Service Award of the Academic Senate, the 2006 IEEE Distinguished Lecturer honor, a 2000 Best Paper Award, and the NSF CAREER Award in 1998, among others. He has been recognized for his contributions to the field through various fellowships and honors, and he is actively involved in the academic community at USC, contributing to research, teaching, and departmental leadership.

Research topics

• Computer science
• Algorithm
• Electronic engineering
• Embedded system
• Reliability engineering

Selected publications

• Ice Thickness Estimation Beneath East Antarctica Using Teleseismic <i>P-wave</i> Coda Autocorrelation

  Journal of the Geological Society of India · 2026-03-01

  article

  ABSTRACT Understanding the structure and dynamics of Antarctica’s ice sheet is critical for assessing global climate change and sea-level rise. We estimated the ice thickness beneath three permanent seismic stations, BELA, SNAA, and MAIT, situated along the East Antarctic region, using the teleseismic P-wave coda autocorrelation technique. We processed verticalcomponent waveforms from teleseismic earthquakes (epicentre distance 30-90°) with Mw ≥ 5.5, recorded at these stations. Our results reveal significant spatial variability in ice thickness, ranging from approximately 0.92 ± 0.07 km at the MAIT station to 3.4 ± 0.13 km at the SNAA station, with BELA exhibiting an intermediate value of 2.82 ± 0.12 km. These variations reflect the influence of local subglacial topography and geological heterogeneities. These high-resolution estimates offer a valuable means to contribute existing ice thickness datasets and provide critical input for improved models of Antarctic ice dynamics.

  PublisherDOI

• Testing and Optimizing Web Applications with Continuous Integration/Continuous Deployment in Cloud Environments

  2025-02-21

  articleSenior author

  CI/CD pipeline solutions are receiving widespread adoption in order to speed up the delivery of web applications while at the same time being able to scale, secure, and optimize them. This study delves into the optimization of CI/CD pipelines in cloud systems, with a particular emphasis on AWS and Azure provided by Microsoft. The study, titled Testing and Optimizing Web Applications with Continuous Integration/Continuous Deployment in Cloud Environments, evaluates the performance of CI/CD pipelines using AWS Code Pipeline and Azure Pipelines, along with Terraform and Kubernetes for infrastructure management. By conducting a series of tests across three scenarios, including CI-only and CI+CD pipelines, the research compares deployment speeds, resource utilisation, and scalability under varying conditions. Selenium for browser testing, Jenkins for automation, and GitLab CI/CD for pipeline management are used for evaluating effectiveness, and effectiveness of the pipelines. The study can prove useful in establishing the features that differentiate between AWS and Azure specifically regarding the speeds of deploying their resources and scalability.

  PublisherDOI

• Asymmetric Predictive Testing for Aging in SRAMs

  2025-06-22

  articleSenior author

  To avoid corruption of user data, predictive testing methods have been proposed to identify SRAMs likely to fail in the near future due to aging. These methods use aggressive operating conditions, e.g., adjustments to wordline voltages or the power supply voltage, that are calibrated to provide high coverage of SRAMs likely to fail in the near future, but end up with some over-testing, i.e., spuriously identifying some chips as likely to fail. We first present our study which discovered that a large fraction of over-tested chips fail due to read faults triggered during read-1 operations. Our analysis identifies asymmetric aging in SRAM cells, which are more likely to store zeros, as the root cause for this. We build on this discovery to propose an asymmetric predictive testing method which performs writes using normal voltages, read-0 at aggressive voltages, and read-1 at less aggressive voltages. We demonstrate that this method significantly reduces over-testing by over $3 \times$ to $5 \times$, for low limits on under-testing. We also propose and use a new statistical sampling and simulation method to enable fast convergence and accurate evaluation of asymmetric predictive testing.

  PublisherDOI

• Evaluation of quality of life, voiding, and sexual dysfunction following urogynecological fistula repair

  Asian Journal of Medical Sciences · 2025-11-30

  articleOpen accessSenior author

  Background: The urinary tract is at risk of injury during pelvic operations and complicated labor. Such injuries may lead to urogynecological fistula (UGF) whereby the urine bypasses the continence mechanisms and involuntarily leaks through genital tract. UGFs significantly impact quality of life, voiding function, and sexual health. This study evaluates outcomes following fistula repair surgery. Aims and Objectives: Assessing quality of life, voiding, and sexual dysfunction following urogenital fistula repair and Comparison of the parameters pre- and postoperative period. Secondary objective was identifying the prevalence of vesicovaginal fistula (VVF) in a tertiary care center. Materials and Methods: A cross-sectional study was conducted over 18 months at R. G. Kar Medical College and Hospital, including 50 patients with urogenital fistulas. Quality of life was measured using 36-item short-form health survey, voiding patterns using Incontinence Questionnaire-Urinary Incontinence Short Form, and sexual function using female sexual function index, assessed preoperatively and 6 months postoperatively. Results: In our study, the most commonly performed procedure was open transabdominal repair 23 (46.0%), followed by laparoscopic VVF repair 13 (26.0%) and ureteric reimplantation 11 (22.0%). Transvaginal approaches were less frequent, with only 2 (4.0%) undergoing transvaginal repair and 1 (2.0%) open transvaginal repair, indicating a preference for abdominal surgical techniques in managing these cases. The value of z is 5.1512 and P&lt;0.00001. The result is considered significant at P&lt;0.05. Conclusion: Urogynecological fistula repair greatly enhances quality of life by restoring urinary continence and improving social and emotional well-being. While voiding and sexual functions often improve, some patients may experience persistent dysfunctions. Ongoing follow-up and holistic care are essential to address these issues and ensure optimal long-term recovery and reintegration.

  PublisherOA PDFDOI

• 3D crustal structure of Kumaun-Garhwal (central) Himalaya from joint inversion of surface wave and body wave dataset.

  2025-03-14

  preprintOpen accessSenior author

  Understanding the crustal structure of the Himalayas and the geometry of the underthrusting Indian plate beneath the Himalayan arc provides crucial insights into regional tectonics and enhances earthquake hazard assessment in the region. This study focuses on the Kumaun-Garhwal Himalaya, a region that is proposed as a potential site for a future great earthquake. We obtained the 3D compressional wave (Vp), shear wave (Vs), and P-to-S wave velocity ratio (Vp/Vs) of the region. To achieve this, we employed joint inversion of body wave and surface wave datasets. This integrated approach overcomes the limitations of individual methods, providing a comprehensive view of the crustal structures. The analysis involved inverting the arrival times of 515 local earthquakes recorded at 41 broadband stations spanning the region and also analyzed continuous waveforms recorded by these stations between November 2006 and June 2008. The resulting crustal velocity structure and relocated earthquake hypocenters reveal a flat-ramp-flat geometry of the Main Himalayan Thrust (MHT). Furthermore, the findings offer critical insights into the crustal composition and its role in earthquake generation. These results enhance our understanding of the region's tectonic framework and contribute to better assessment and mitigation of seismic hazards in the Himalayan arc.Keywords: Seismic tomography; Continental tectonics; Main Himalayan Thrust; crustal imaging.

  PublisherDOI

• Exploring the Efficiency of Docker Containers in Cloud-Based Application Deployment Using AWS

  2025-02-21 · 1 citations

  articleSenior author

  The quick evolution of web apps has revolutionised people’s daily lives and careers. Hybrid clouds are becoming more common as a means of accessing resources on demand via cloud computing. Using Docker as its foundation, this article presents a framework that can automatically grow web applications in a hybrid cloud. Web applications are deployed on hybrid cloud hosts, and their resources may be adjusted by adding or removing docker containers, a technique used in this research. This study explores the efficiency of Docker containers in cloud-based application deployment on Amazon Web Services (AWS) by comparing two primary deployment models, Amazon EC2 and Amazon ECS. Performance testing was conducted using Apache JMeter, and development was supported by Visual Studio Code. Performance metrics, such as throughput, response time and error rate, were measured across varying instance types and configurations. The results also depicted the variation in performances; in the step load condition, EC2 has an ultra-low error percentage of $0.01 \%$ with a mean throughput rate of 35.89 operation/sec, and ECS has a higher throughput rate of 80.95 operation/sec but with a higher error percentage of 2.22 are observed. Even under peak load, their EC2’s rejection rate went to $2.93 \%$ and throughput reduced to 25.41, though ECS achieved a better high throughput of 34.45 per second operations at a slightly higher rejection rate of $8.86 \%$. These results indicate the ideal cost benefits of Docker container stand alone in achieving larger scale and better throughout relative to VM deployments that provide fewer errors and better reliability in large traffic loads. Thus, this work presents real-world experience useful for organisations that seek to enhance IaaS resource performance and elasticity in AWS environments.

  PublisherDOI

• Persistent PSD trends: a tool for seismic landslide detection

  Acta Geophysica · 2025-10-16

  articleSenior author
  PublisherDOI

• Analysing the Impact of Cloud-Native Architectures on Web Application Performance Using AWS Lambda

  2025-05-16 · 1 citations

  articleSenior author

  Serverless computing, exemplified by AWS Lambda, offers dynamic, scalable infrastructure solution. This paper investigates performance of AWS Lambda, API Gateway, and DynamoDB in web applications, focusing on scalability, resource utilisation, and response time. Through an experimental framework, impact of varying workloads on cold start latency and system efficiency is assessed. Results show that as request volumes increase, cold start times decrease from 600 ms at 10 requests to 350 ms at 1000 requests, while response times improve from 500 ms to 250 ms. The serverless architecture demonstrates superior performance over traditional server-based models, with more efficient resource management and faster scaling under fluctuating traffic. The study highlights benefit of serverless computing for dynamic workloads and emphasises the need for further research on optimising AWS Lambda configurations for enhanced performance.

  PublisherDOI

• Assessing the Performance of Machine Learning Approaches for Cyber Attack Detection to Improve Cybersecurity

  2025-02-21 · 2 citations

  articleSenior author

  Cybercrime is a major problem all around the world since it causes countries and people to lose a lot of money every day. Recent developments have brought security models and prediction tools based on AI to address the difficulties of cyberattack detection and prevention. The purpose of this research is to assess how well a CNN model can identify cyberattacks on the CSE-CIC-IDS2018 dataset. The dataset underwent extensive preprocessing to address duplication, missing values, and feature normalisation using Min-Max scaling. An 80-20 train-test split was applied, and the CNN model was implemented alongside comparison models, including Deep Neural Networks (DNN), NB, and LSTM networks. Performance measures such as F1-score, recall, accuracy, and precision were used to assess the model’s efficiency. The CNN model had better results than its competitors in recognising and categorising cyberattacks, with a $95 \%$ F1 score, $99 \%$ recall, $92 \%$ precision, and 98.31% accuracy. Visualisation of the training and testing trends highlighted the model’s robustness with minimal overfitting, though some fluctuations in testing loss indicated areas for further optimisation. These results establish the CNN model as a reliable and robust approach for enhancing cybersecurity through efficient cyber-attack detection.

  PublisherDOI

• Enhancing Software Quality with Fault Detection and Prediction based on AI-Driven Model

  2025-01-24

  articleSenior author

  Early defect detection and prediction are essential in software engineering to reduce costs and improve quality. This study presents an AI-driven approach for fault detection and prediction employing ML techniques on the CM1 dataset from NASA’s software metrics program. The data preprocessing steps include normalisation, feature selection, and the use of the ADASYN algorithm to enhance minority class detection. Faults are predicted using ML classifiers: Extra Trees (ET), Support Vector Machine (SVM), KNN, RF, MLP, and NB. F1 score, recall, accuracy, and precision are a metrics utilised to assess a models. When it comes to defect prediction, the Extra Trees Classifier (ETC) model outperforms all of the others with an accuracy91.51%, precision91.76%, and recall99%. Although the Naive Bayes model shows lower accuracy (77.61%) and efficiency, it offers a simpler approach for fault detection. Future research in this area should be guided by these findings, which highlight the promise of AI-driven approaches to enhance frameworks for software defect prediction and quality assurance.

  PublisherDOI

Frequent coauthors

• Hugh A. Bruck

  University of Maryland, College Park

  75 shared

• Dana S. Nau

  University of Maryland, College Park

  44 shared

• Rishi K. Malhan

  University of Southern California

  40 shared

• Brual C. Shah

  University of Southern California

  39 shared

• Ariyan M. Kabir

  University of Southern California

  36 shared

• Prahar M. Bhatt

  University of Southern California

  33 shared

• Krishnanand N. Kaipa

  Norfolk State University

  32 shared

• Jason M. Gregory

  32 shared

Education

• Ph.D., Electrical Engineering

  University of Southern California

  2000

• M.S., Electrical Engineering

  University of Southern California

  1996

• B.S., Electrical Engineering

  University of Southern California

  1994

Awards & honors

• 2009 USC Distinguished Service Award of Academic Senate
• 2006 IEEE Distinguished Lecturer
• 2005 Other Awards
• 2000 Best Paper Award
• 2000 Nominated for Best Paper Award