Graduate Exam Abstract

Abstract: Accurate and robust indoor localization is essential for several applications ranging from navigation in large indoor spaces to location-based services on smartphones. However, achieving high localization accuracy presents several challenges, including device heterogeneity, external noise, privacy concerns, and the growing threat of adversarial attacks on machine learning (ML) models. These challenges are further complicated by the need for lightweight and efficient ML algorithms that can be deployed on resource-constrained devices like smartphones. This thesis addresses these challenges by proposing novel frameworks that enhance security, privacy, and noise-resilience while maintaining high accuracy. First, the SANGRIA framework leverages novel augmentation methods to mitigate the impact of noisy data and improve robustness. Its extension, FedHIL, adopts a federated learning approach to preserve user data privacy while enhancing model performance through a novel aggregation strategy. Additionally, the SENTINEL framework employs adversarial learning techniques to bolster resilience against malicious attacks and ensure secure localization. All proposed algorithms are lightweight, making them well-suited for real-time deployment on smartphones. The frameworks have been validated on real-world datasets collected from diverse indoor environments using publicly available smartphones. Experimental results confirm that these frameworks provide efficient solutions for modern indoor localization systems.

Adviser: Sudeep Pasricha
Co-Adviser: N/A
Non-ECE Member: Nikhil Krishnaswamy
Member 3: Anthony Maciejewski
Addional Members: Anura Jayasumana

Publications:
1. Saideep Tiku, Danish Gufran, and Sudeep Pasricha. "Multi-head attention neural network for smartphone invariant indoor localization." IEEE 12th international conference on indoor positioning and indoor navigation (IPIN), 2022.

2. Danish Gufran, Saideep Tiku, and Sudeep Pasricha. "VITAL: Vision transformer neural networks for accurate smartphone heterogeneity resilient indoor localization." 60th ACM/IEEE Design Automation Conference (DAC), 2023.

3. Danish Gufran, Saideep Tiku, and Sudeep Pasricha. "SANGRIA: Stacked autoencoder neural networks with gradient boosting for indoor localization." IEEE Embedded Systems Letters (ESL), 2023.

4. Danish Gufran, and Sudeep Pasricha. "FedHIL: Heterogeneity resilient federated learning for robust indoor localization with mobile devices." ACM Transactions on Embedded Computing Systems (TECS), 2023.

5. Saideep Tiku, Danish Gufran, and Sudeep Pasricha. "Smartphone invariant indoor localization using multi-head attention neural network." Machine Learning for Indoor Localization and Navigation, Springer International Publishing, 2023.

6. Danish Gufran, Saideep Tiku, and Sudeep Pasricha. "Heterogeneous device resilient indoor localization using vision transformer neural networks." Machine Learning for Indoor Localization and Navigation, Springer International Publishing, 2023.

7. Danish Gufran, Saideep Tiku, and Sudeep Pasricha. "STELLAR: Siamese multiheaded attention neural networks for overcoming temporal variations and device heterogeneity with indoor localization." IEEE Journal of Indoor and Seamless Positioning and Navigation (ISPIN-J), 2023.

8. Danish Gufran, and Sudeep Pasricha. "CALLOC: Curriculum adversarial learning for secure and robust indoor localization." ACM/IEEE Design, Automation & Test in Europe Conference & Exhibition (DATE), 2024.

9. Danish Gufran, Pooja Anandathirtha, and Sudeep Pasricha. "SENTINEL: Securing Indoor Localization against Adversarial Attacks with Capsule Neural Networks." IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 2024.

Program of Study:
ECE 581C
ECE 561
ECE 513
ECE 658
ECE 569
CS 542
CS 545
SYSE 541