Design of an Instruction-Context-Aware PUF for Secure ALU Operation with ML Attack-Resistant Dynamic Key Generation
This project presents the design, verification, and physical implementation of a secure Arithmetic Logic Unit (ALU) integrated with an instruction-context-aware Physical Unclonable Function (PUF) for secure operation against ML-Attack with the help of dynamic key generation. Traditional PUF challenge generation schemes are vulnerable to Machine Learning (ML) modeling attacks. To address this, we introduce an architecture that generates dynamic, non-linear challenge sequences directly from the specific ALU opcode and operands executing in the pipeline. We evaluated four challenge generation techniques (Concatenation, XOR Fold, Modulo Multiplication, and LFSR Mix) across three PUF architectures (Arbiter, XOR, and Ring Oscillator) to assess security against Regression, Random Forest, and Deep Neural Network (DNN) attacks. The system operates via a 3-stage FSM and achieves 100% functional verification using a UVM-style testbench. Synthesis was performed using the Nangate 45nm Open Cell Library, identifying an optimal 8 ns clock period. A fully DRC-clean layout was achieved with 59.95% core utilization, balancing the area, power, and latency trade-offs inherent in hardware security primitives.
DESIGN AND IMPLEMENTATION OF A PULSE OXIMETER USING PHTOPLETHYSMOGRAPHY
Pulse oximetry is a non-invasive method for determining arterial oxygen saturation. This project presents the design and implementation of a cost-effective pulse oximeter based on the principles of photoplethysmography (PPG). The technique leverages the differing absorption characteristics of oxygenated and deoxygenated hemoglobin to estimate oxygen saturation and heart rate, in accordance with the Beer–Lambert Law. A transmissive optical configuration is employed, where light passes through the vascular region of the fingertip. The hardware setup incorporates two light sources operating at distinct wavelengths—660 nm (red) and 940 nm (infrared). A hybrid detection scheme using a light dependent resistor (LDR) and an infrared photodiode is utilized to accurately capture variations in light intensity. An ESP32 microcontroller serves as the central processing unit for data acquisition and computation. The proposed system demonstrates a scalable and economical solution capable of real-time physiological monitoring with acceptable accuracy, making it suitable for potential applications in healthcare environments.