Towards Trustworthy Neuromorphic Computing

This book explores the reliability and security challenges of Computing-in-Memory (CIM), a promising approach to overcoming the limitations of traditional computing. It examines the weaknesses of memristive devices, which are key to this technology, and presents practical methods to improve their stability and security. Key topics include a fault injection platform for testing reliability, the NeuroHammer attack, which exposes security risks, and the NeuroBreakoutBoard, a hardware tool for evaluating real devices. By bridging theory and practice, this book is essential for researchers, engineers, and industry professionals working on next-generation computing systems.

In addition, this book: Describes a fault injection platform for evaluating the reliability of Logic-in-Memory operations Discusses NeuroHammer, novel hardware security attack targeting memristive crossbar arrays Includes an instrumentation platform enabling the execution and analysis of CIM operations on real memristive devices