Steven M. Hernandez

Portfolio for: Research

During my Ph.D in the Computer Science department at Virginia Commonwealth University so far, my research has primarily been focused on areas such as

Performing Wi-Fi Sensing with Off-the-shelf Smartphones 2020

PerCom 2020 Best Demo Award winner

WiFi sensing has recently attracted a lot of attention in providing a method for device-free sensing with standard WiFi devices through the use of Channel State Information (CSI). However, access to CSI from user-level applications is not provided by most WiFi devices, specifically ubiquitous smartphone devices. In this demonstration, we present our custom developed application for collecting, labeling and processing CSI on-device, in real-time, for standard off-the-shelf Android smartphones. We additionally demonstrate the use of our application on select device-free sensing tasks.

Paper: http://www.people.vcu.edu/~ebulut/percom20-demo.pdf

ESP32 Codebase: https://stevenmhernandez.github.io/ESP32-CSI-Tool/

WiFi Sensing with Channel State Information 2020

Developed ESP32 CSI Toolkit and Deep Learning System

Toolkit: https://stevenmhernandez.github.io/ESP32-CSI-Tool/
Conference: WoWMoM 2020
IEEE PerCom Demos 2020

Using TrinaryMC using our novel Perceived Direction Information (PDI), we can reduce the search area compared to other Monte Carlo Localization Techniques.

TrinaryMC 2020

Relative Localization for Mobile Devices

A device such as a smartphone, running our TrinaryMC algorithm uses knowledge of neighboring devices to predict their relative location within an area. Each device sends out beacon packets which allows neighboring devices to collect RSSI, the received signal strength indicator. RSSI is noisy and generally unpredictable, so instead of using the RSSI data raw, we instead create an adjacency matrix containing a novel PDI value (Perceived Direction Information) for each pair of neighboring devices.

Journal: Elsevier Journal of Network and Computer Applications
Conference: Proceedings of IEEE Consumer Communications & Networking Conference (CCNC)

RSSI-based Device Localization with Apache Spark Streaming 2018

Indoor Localization

This work looks at developing a streaming machine learning system developed in Apache Spark Streaming for RSSI-based indoor localization.

Apache Spark Streaming Codebase: https://github.com/StevenMHernandez/Spark-Streaming-OS-ELM