Chiluka Rishith Reddy

Connecting Physical Signals to Intelligent Systems.

Electronics and Instrumentation Engineering student at Vallurupalli Nageswara Rao Vignana Jyothi Institute of Engineering & Technology (VNR VJIET) (7th Sem, CGPA: 7.52). Focused on building reliable embedded solutions that connect sensors, electronics, and intelligent software. Applying knowledge of C programming, microcontrollers, signal processing, and real-time data acquisition to embedded product development, testing, and system validation.

VNR VJIET

B.Tech EIE (7th Sem)

7.52 CGPA

Academic Standing

1^st Prize

Patent Summit Award

0.028 MSE

SP-GNN Fault Error

In Progress PROJECT ID: STM32-PPG-AF

PPG-Based AF Detection on STM32 — Inter-Patient Validation

Developing an STM32-based Photoplethysmogram (PPG) signal acquisition and Atrial Fibrillation (AF) detection system. Incorporates raw optical signal acquisition, digital bandpass preprocessing to isolate cardiac pulsatile waveforms, HR/HRV feature extraction, and lightweight classification, evaluated using patient-disjoint validation for evaluating generalization to unseen subjects.

Verified Processing Pipeline

PPG Signal → Signal Preprocessing → HR/HRV Analysis → Lightweight Classification → Patient-Disjoint Eval

STM32 (ARM Cortex-M) Embedded C PPG Sensors Digital Bandpass Filtering HR/HRV Metrics STM32CubeIDE

100 Hz

Optical PPG Sampling Rate

Architecture:	ARM Cortex-M (STM32)
Filtering:	0.5 – 5.0 Hz Bandpass
Validation:	Patient-Disjoint Folds
Status:	Active Development

Research Presentation • ICAMSTA-2025 OSMANIA UNIVERSITY

Hybrid Statistical–Physics Informed Graph Neural Network (SP-GNN)

Developed an unsupervised SP-GNN framework for sensor fault detection using the Intel Berkeley Research Lab (IBRL) dataset comprising 54 sensors and approximately 2.3 million readings. Integrated physics-informed message passing, graph-Laplacian regularization, and Monte-Carlo Dropout for uncertainty-aware anomaly detection.

Physical Message-Passing Topology

54 Sensors (IBRL) → Graph Relational Adjacency → Graph-Laplacian Regularization → Monte-Carlo Dropout → Uncertainty Anomaly Score

Python PyTorch / Geometric Graph Neural Networks Statistical Physics Graph-Laplacian Distributed Sensor Systems

0.028

Mean Reconstruction Error (MSE)

Standard GCN:	0.072 MSE
Physics-GNN:	0.049 MSE
SP-GNN (Ours):	0.028 MSE (-61.1%)
Dataset Scale:	54 Sensors • ~2.3M Pts

Hardware Prototype PROJECT ID: ARDUINO-EV-MON

Embedded EV Monitoring System Using Arduino

Developed an Arduino-based embedded monitoring prototype for real-time acquisition and monitoring of key EV parameters, including battery level, vehicle speed, and temperature. Implemented sensor interfacing and data acquisition with a software dashboard for real-time visualization and estimated driving-range monitoring.

Hardware-Software Telemetry Flow

Battery, Speed, Temp Sensors → Signal Conditioning & ADC → Arduino DAQ Firmware → Serial Telemetry → Driving-Range Dashboard

Arduino C / C++ Sensor Interfacing Data Acquisition (DAQ) State of Charge (SOC) Software Dashboard

Real-Time

Multi-Parameter Telemetry

Controller:	Arduino ATmega
Monitored Parameters:	Battery SOC, Speed, Temp
Range Estimation:	Dynamic Calculation
Visualization:	Software Dashboard

1st Prize Winner • Patent Summit NATIONAL WORKSHOP • VNR VJIET

Passive Alcohol Detection System for Vehicle Safety

Awarded 1st Prize in the Patent Summit Competition at the National Workshop on Patent Analytics and Filing Framework, VNR VJIET. Designed a non-intrusive automotive safety prototype that replaces active breathalyzer blowing with continuous, passive ambient air sampling in the steering column/cabin, interfaced with ignition interlock circuitry to prevent vehicle operation under intoxication.

Safety Interlock Mechanism

Passive Ambient Gas Sensing → Signal Conditioning & ADC → Microcontroller Threshold Logic → Ignition Relay Cutoff

Microcontrollers Gas Sensors (MQ Series) Signal Conditioning Relay Actuation Patent Analytics Framework

1^st Prize

Patent Summit Competition

Sensing Mode:	Passive / Non-Intrusive
Interlock Action:	Automated Relay Disconnect
Venue:	National Workshop @ VNR VJIET
Scope:	Automotive Safety Innovation

Connecting Physical Signals to Intelligent Systems.