Reliable and High-Performance Hardware Architectures for the Advanced Encryption Standard/Galois Counter Mode

The high level of security and the fast hardware and software implementations of the Advanced Encryption Standard (AES) have made it the first choice for many critical applications. Since its acceptance as the adopted symmetric-key algorithm, the AES has been utilized in various security-constrained applications, many of which are power and resource constrained and require reliable and efficient hardware implementations. In this thesis, first, we investigate the AES algorithm from the concurrent fault detection point of view. We note that in addition to the efficiency requirements of the AES, it must be reliable against transient and permanent internal faults or malicious faults aiming at revealing the secret key. This reliability analysis and proposing efficient and effective fault detection schemes are essential because fault attacks have become a serious concern in cryptographic applications. Therefore, we propose, design, and implement various novel concurrent fault detection schemes for different AES hardware architectures. These include different structure-dependent and independent approaches for detecting single and multiple stuck-at faults using single and multi-bit signatures. The recently standardized authentication mode of the AES, i.e., Galois/Counter Mode (GCM), is also considered in this thesis. We propose efficient architectures for the AESGCM algorithm. In this regard, we investigate the AES algorithm and we propose lowcomplexity and low-power hardware implementations for it, emphasizing on its nonlinear transformation, i.e., SubByes (S-boxes). We present new formulations for this transformation and through exhaustive hardware implementations, we show that the proposed architectures outperform their counterparts in terms of efficiency. Moreover, we present parallel, high-performance new schemes for the hardware implementations of the GCM to improve its throughput and reduce its latency. The performance of the proposed efficient architectures for the AES-GCM and their fault detection approaches are benchmarked using application-specific integrated circuit (ASIC) and field-programmable gate array (FPGA) hardware platforms. Our comparison results show that the proposed hardware architectures outperform their existing counterparts in terms of efficiency and fault detection capability.