Research - Nanomagnetic Logic, Skyrmion Computing, Edge AI

Research Overview

INTRINSIC Lab advances cyber-physical systems engineering by co-designing sensors, instrumentation, materials and computing architectures that exploit intrinsic device physics (magnetic, electronic, memristive, and quantum effects). Rather than focusing narrowly on electrical or computer engineering abstractions, our work centers on sensing-driven systems where the intrinsic physical properties of materials and devices are leveraged to create energy-efficient computation, robust perception, and secure actuation.

The lab's approach is intentionally broad and multi-inter-disciplinary: we investigate fundamental device physics and materials properties, build instrumentation and automated characterization platforms, design novel compute primitives that exploit those intrinsic properties, and integrate everything into resilient cyber-physical systems for sensing applications in robotics, quantum, space tech, and emerging technologies.

Five Research Pillars

1. Sensors, Instrumentation & Materials-Enabled Devices

Focus: Experimental and theoretical investigation of graphene-based electronics, magnetic skyrmions, spintronic elements, memristive devices, and piezoelectric materials for energy harvesting. Research spans materials synthesis, device characterization, and sensor development for intelligent instrumentation.

Objectives: Device modeling, materials-device co-optimization, compact models for circuit-level integration, and energy harvesting system design using piezoelectric materials and advanced nanomaterials.

Applications: Low-power sensing, non-volatile logic, materials-aware CAD flows, self-powered sensors, and energy-efficient cyber-physical systems.

2. Rebooting Computing & Device-Aware Architectures

Focus: Beyond-CMOS paradigms including nanomagnetic logic, skyrmion-based logic, in-memory computing, hardware-aware neuromorphic accelerators, and digital IC design. Work encompasses chip design, mixed-signal IC design, and indigenous integrated circuits development.

Objectives: Develop energy-efficient arithmetic and control primitives, fault-tolerant mapping techniques, device-to-architecture design methodologies, and digital IC design flows for indigenous integrated chips.

Applications: Edge AI, ultra-low-power inference, approximate computing for resource-constrained systems, and indigenous chip design for defense and critical applications.

3. Sensing-Driven Edge Intelligence & Autonomous Systems

Focus: Sensor fusion, neuromorphic perception, autonomous control tailored for constrained platforms, and system-level integration for intelligent robotics and connected health. Includes physiological sensing via wearables (ECG/PPG/EMG/SpO2), multimodal fusion, and edge health analytics.

Objectives: Co-design of sensors and algorithms, hardware-accelerated perception and biosignal-processing pipelines, artifact-resilient wearable signal processing, on-device ML for arrhythmia/gait/anomaly detection, robust BLE/LoRa connectivity, and real-time edge intelligence deployment.

Applications: Defense robotics, autonomous ground/air vehicles, biomedical wearables for continuous monitoring, secure connected health at the edge, and intelligent cyber-physical systems for mission-critical operations.

4. Secure, Trustworthy Cyber-Physical Systems Engineering

Focus: Hardware security, anti-spoofing sensors, trustworthy IoMT and medical wearables, military systems, and resilient architectures for critical infrastructure protection.

Objectives: Hardware-rooted security primitives, secure sensor architectures and wearable communication (BLE/LoRa), privacy-preserving health data pipelines, methods for provable robustness in CPS deployments, and side-channel attack mitigation.

Applications: Secure positioning, tamper-resistant medical wearables/devices, secure connected health at the edge, resilient critical-infrastructure sensing, and secure indigenous integrated chips for defense systems.

5. AI-Driven Electronics System Design and Electronics Design Driven AI System

Focus: Machine learning for materials and device design, AI-driven automated fabrication and characterization, semiconductor process development using AI automation.

Objectives: AI-driven CAD for device optimization, automated testbeds, physics-informed ML models, hardware-aware AI algorithms, AI-accelerated semiconductor development.

Applications: Digital twin validation, AI-accelerated discovery, algorithm-hardware co-optimization, intelligent semiconductor process for indigenous chip manufacturing.

"Exploiting intrinsic device physics for intelligent cyber-physical systems, from nanomagnetic logic and memristive materials to edge AI and secure autonomy, co-designing sensors, compute architectures, and instrumentation for robust, energy-efficient sensing, and translating device behavior into field-ready engineering solutions."

Prof. Santhosh Sivasubramani

Interconnected Research Framework

These five pillars are not isolated research areas but represent interconnected domains that complement and reinforce each other. The synergy between these pillars enables comprehensive problem-solving approaches and fosters innovation that bridges traditional disciplinary boundaries. This integrated approach allows us to tackle complex real-world challenges with holistic solutions.

Cross-Cutting Research Themes

Within the scope of the five pillars, our research spans cross-cutting themes that integrate multiple technology domains and application areas, enabling comprehensive solutions from fundamental device physics to real-world deployment.

Vertical Technology Domains

Quantum Information Processing & Cryptography

Research explores quantum information processing spanning quantum sensing, quantum computing, quantum cryptography, and quantum-classical hybrid systems. Building on Magnetic Quantum-dot Cellular Automata foundations, work extends to quantum-enabled sensing components, quantum metrology for precision instrumentation, and quantum cryptography for secure communication in cyber-physical systems. Active areas include quantum key distribution, quantum-inspired algorithms, and establishing experimental testbeds for quantum sensing to validate quantum-classical interfaces and demonstrate quantum-enabled edge intelligence for defense and healthcare applications.

Semiconductor Nanofabrication & Process Integration

Work emphasizes semiconductor nanofabrication, semiconductor process development, nanoscale device fabrication, and materials synthesis with focus on process-device co-optimization. Research spans nanoscale lithography, plasma etching, thin-film deposition, and cleanroom-based semiconductor nano fabrication for beyond-CMOS devices including spintronics device fabrication, memristor fabrication, graphene interconnect processing, and skyrmion device prototyping. Semiconductor process development for indigenous integrated chips leverages AI-driven fabrication and characterization pipelines enabling automated materials characterization, device-level measurements, and reproducible nanofabrication workflows for chip design and manufacturing.

Hardware Security: Side-Channel Attacks & Resilient Architectures

Security research addresses side-channel attacks, power analysis attacks, fault injection attacks, and hardware Trojans. Work on secure architectures includes physically unclonable functions, true random number generators, secure boot mechanisms, and tamper-resistant designs for critical infrastructure. Research extends to resilient architectures that maintain functionality under adversarial conditions, Byzantine-resilient protocols, and provably secure hardware-software interfaces. Security solutions leverage spintronics-based security primitives, quantum cryptography integration, and hardware-rooted security for trustworthy cyber-physical systems including anti-spoofing sensors for GPS-denied navigation and secure Internet of Medical Things.

Horizontal Integration Areas

Technology Translation: Research to Deployment Roadmap

A structured technology translation roadmap guides research from laboratory to real-world impact through five stages: fundamental research with device physics and proof-of-concept demonstrations, prototype development with testbed validation and performance benchmarking, system integration with existing technologies and interoperability testing, field deployment through pilot programs and real-world validation, and commercialization pathways via intellectual property protection, industry partnerships, startup incubation, and technology licensing. This roadmap ensures technologies move from academic prototypes to deployable solutions with engineering confidence, addressing technical readiness levels, manufacturability, cost-effectiveness, and long-term sustainability through reproducible fabrication, automated testing, and rigorous instrumentation for indigenous technology development.

System-Level Design & Hardware-Software Co-Design

Research advances system-level design methodologies bridging device physics to deployed systems through hardware-software co-design, FPGA prototyping, ASIC design flows, System-on-Chip integration, and digital IC design. Work encompasses RTL design, electronic design automation tools, reconfigurable computing platforms, chip design for indigenous integrated circuits, mixed-signal IC design, and digital twin frameworks for virtual prototyping. System-level design extends to approximate computing for edge systems, AI-hardware co-optimization, and end-to-end design flows from device models to system deployment demonstrating complete semiconductor process development to chip design pipeline for avionics, defense System-on-Chip applications, and cyber-physical systems integration.

IoT-Edge-Cloud Convergence for Pervasive Sensing

Research integrates IoT architectures, edge computing, fog computing, and sensor networks to enable pervasive sensing for defense, healthcare, and industrial applications. Edge AI inference on resource-constrained IoT devices leverages approximate computing, neuromorphic accelerators, energy-efficient hardware, and energy harvesting solutions using piezoelectric materials for self-powered sensor nodes. IoT security addresses secure communication protocols, authenticated sensor networks, and privacy-preserving edge intelligence. Applications span smart healthcare, industrial IoT monitoring, and defense-grade sensor networks. Edge-cloud convergence enables real-time decision-making, reduced latency, bandwidth efficiency, and privacy-by-design through distributed processing across IoT endpoints, edge gateways, and cloud infrastructure.

Research Platforms & Infrastructure

Our research leverages industry-standard platforms and processor architectures in collaboration with leading technology partners including ARM, NVIDIA, Intel, AMD, Google Cloud, and AWS. These partnerships enable cutting-edge exploration across AI/ML systems, wireless networks, and embedded computing.

🚀 AI/ML Development Ecosystem

Utilizing NVIDIA NGC Catalog - a comprehensive AI platform providing GPU-optimized containers, pre-trained models, and SDKs for research and deployment:

AI Frameworks: PyTorch, TensorFlow containers for training and inference
Inference Optimization: Triton Inference Server, TensorRT for deployment
NIM Microservices: Foundation models (LLMs, vision models) via OpenAI-compatible APIs
Specialized Models: DeepStream for video analytics, NeMo for speech/NLP

Enabling research in autonomous agents, multi-modal AI, edge deployment, and production ML pipelines across cloud and edge infrastructure.

📡 Next-Gen Wireless & 6G Research

Research access to NVIDIA AI Aerial platform for software-defined wireless and intelligent networks:

AI-Accelerated RAN: GPU-accelerated Radio Access Networks
Aerial Framework: Software-defined radio architectures
Omniverse Digital Twin: Wireless infrastructure simulation and testing
CUDA Signal Processing: Real-time SDR pipelines

Research directions: intelligent spectrum management, edge-cloud coordination for 6G, AI-driven network optimization.

⚡ Processor Architecture & Embedded Systems

Hardware-software co-design research leveraging ARM processor IP and industry partnerships with Intel, AMD:

ARM Cortex IP: Cortex-A5, Cortex-M0, Cortex-M3 processors (full RTL access)
SoC Design: Custom chip development, FPGA prototyping, ASIC flows
Edge AI Hardware: TinyML on resource-constrained ARM platforms
RTOS Integration: Real-time operating systems for embedded control

Enabling custom processor architectures, low-power embedded AI, and indigenous chip development for defense and aerospace applications.

Emerging Research Directions

🤖 Autonomous AI Systems & Agentic Intelligence

Investigating architectures for autonomous AI agents that can perceive, reason, plan, and act with tool augmentation:

Tool-Augmented Language Models: Function-calling paradigms, API integration, external knowledge access
Multi-Agent Systems: Coordination frameworks, collaborative problem-solving, agent-to-agent communication
Autonomous Decision-Making: Planning under uncertainty, goal-directed behavior, reinforcement learning
Human-AI Collaboration: Interactive workflows, explainable reasoning, verification mechanisms

Technical Framework & Implementation

Current research utilizes industry-standard frameworks and platforms:

LangChain/LangGraph: Agent orchestration, state management, and workflow coordination
Foundation Model APIs: OpenAI GPT-4/4o, Anthropic Claude, Google Gemini for tool integration
Retrieval-Augmented Generation: Vector databases (Chroma, Pinecone), embedding models, document retrieval strategies
NVIDIA NGC NIM: Containerized foundation model inference with OpenAI-compatible endpoints
Memory Architectures: Conversational memory, entity tracking, semantic caching for context persistence

Research Applications & Student Projects

Active investigations span multiple application domains:

Autonomous Research Assistants: Web search integration, paper retrieval, code execution, and summarization for literature analysis
Multi-Agent Development Systems: Collaborative agents for requirements analysis, code generation, review, testing, and documentation
Domain-Specific AI: Context-aware systems for technical documentation, course materials, and laboratory protocols
IoT Intelligence: Autonomous monitoring systems analyzing sensor streams, anomaly detection, and automated response triggering
Hardware Design Automation: Tool-augmented agents for RTL generation, testbench creation, and synthesis optimization

Research Approach & Methodology

This rapidly evolving field benefits from collaborative research with OpenAI, Anthropic, Google DeepMind, and academic communities worldwide. Our methodology emphasizes:

Systematic experimentation with production frameworks and established best practices
Continuous integration of advances in agent architectures, reasoning patterns, and tool integration
Rigorous evaluation addressing reliability, safety, hallucination mitigation, and computational efficiency
Development of reproducible workflows suitable for research and educational deployment

Research focuses on practical deployment patterns and educational integration within the broader academic research community.

Current Status: Active research area with integration into advanced graduate courses and ongoing collaborations with industry partners including NVIDIA NGC and cloud infrastructure providers.

Research Areas