Yamini Shankar

LiteTrack: Power-Efficient Wi-Fi-Based Indoor Localization for Smartphones

IEEE ANTS 2025 (Accepted for presentation)

Work carried out during internship at SRIB (Samsung R&D Institute India – Bangalore).

LiteTrack investigates how to make on-device, Wi-Fi-based indoor localization more power-efficient on smartphones. Through extensive experiments, we benchmark the minimum number of Wi-Fi scans required to achieve reliable localization while reducing battery consumption. We quantify the accuracy–energy trade-off across scan rates and propose a practical scanning policy that retains robust localization performance with significantly fewer scans—thereby extending battery life without sacrificing user experience.

Practical Defense Against Adversarial Wi-Fi Sensing

IEEE ANTS 2024

We present a black-box perturbation defense for CSI-based sensing that significantly reduces adversarial classifier accuracy while maintaining communication quality (≈1 dB median SNR delta). The approach targets realistic ISAC threats with minimal overhead and strong practicality.

Improving Network Resource Utilization for Distributed Wireless Sensing Applications

AIoT Workshop @ MobiHoc 2024

We compress and selectively subsample CSI streams before edge inference to improve end-to-end efficiency. Results across two testbeds show better latency, energy use, and throughput with negligible impact on sensing accuracy.