Electronic Thesis and Dissertation Repository

Degree

Master of Engineering Science

Program

Electrical and Computer Engineering

Supervisor

Dr. Arash Reyhani Masoleh

Abstract

The Advanced Encryption Standard (AES) is a symmetric-key block cipher for electronic data announced by the U.S. National Institute of Standards and Technology (NIST) in 2001. The encryption process is based on symmetric key (using the same key for both encryption and decryption) for block encryption of 128, 192, and 256 bits in size. AES and its standardized authentication Galois/Counter Mode (GCM) have been adopted in numerous security-based applications. GCM is a mode of operation for AES symmetric key cryptographic block ciphers, which has been selected for its high throughput rates in high speed communication channels. The GCM is an algorithm for authenticated encryption to provide both data authenticity and confidentiality that can be achieved with reasonable hardware resources. The hardware implementation of the AES-GCM demands tremendous amount of logic blocks and gates. Due to natural faults or intrusion attacks, faulty outputs in different logic blocks of the AES-GCM module results in erroneous output. There exist plenty of specific literature on methods of fault detection in the AES section of the AES-GCM. In this thesis, we consider a novel fault detection of the GCM section using parity prediction. For the purpose of fault detection in GCM, two independent methods are proposed. First, a new technique of fault detection using parity prediction for the entire GCM loop is presented. Then, matrix based CRC multiple-bit parity prediction schemes are developed and implemented. As a result, we achieve the fault coverage of about 99% with the longest path delay and area overhead of 23% and 10.9% respectively. The false alarm is 0.12% which can be ignored based on the number of injected faults.


Share

COinS