Matrix Library with Algorithm-Based Fault Tolerance (ABFT)

Overview

This project develops a matrix library that incorporates ABFT techniques to detect and correct errors in matrix computations, ensuring reliable results even in the presence of hardware faults or data corruption.

ABFT Techniques Implemented

• Checksum Encoding - Adding redundant checksum rows/columns for error detection
• Error Detection - Algorithms to identify corrupted data
• Error Correction - Methods to recover original data from faulty computations

Matrix Operations

• Addition (add_data_2d.c) - Fault-tolerant matrix addition
• Multiplication (mul_data_2d.c) - Fault-tolerant matrix multiplication
• Checksum Operations - Row and column checksum calculations
• Data Corruption Simulation - Testing fault tolerance under artificial faults

Testing & Validation

Includes comprehensive testing scripts and utilities for validating ABFT implementations, including fault injection and correction verification.

Skills Demonstrated

C programming, algorithm design, fault tolerance techniques, numerical computing, error detection and correction, and software reliability engineering.

Project Report

Download the full project report for detailed implementation details, algorithms, and testing results.