Cryptology ePrint Archive: Report 2015/763
Highly Efficient GF(2^8) Inversion Circuit Based on Redundant GF Arithmetic and Its Application to AES Design
Rei Ueno, Naofumi Homma, Yukihiro Sugawara, Yasuyuki Nogami, and Takafumi Aoki
Abstract: This paper proposes a compact and efficient GF(2^8) inversion circuit design based on a combination of non-redundant and redundant Galois Field (GF) arithmetic. The proposed design utilizes redundant GF representations, called Polynomial Ring Representation (PRR) and Redundantly Represented Basis (RRB), to implement GF(2^8) inversion using a tower field GF((2^4)^2). In addition to the redundant representations, we introduce a specific normal basis that makes it possible to map the former components for the 16th and 17th powers of input onto logic gates in an efficient manner. The latter components for GF(2^4) inversion and GF(2^4) multiplication are then implemented by PRR and RRB, respectively. The flexibility of the redundant representations provides efficient mappings from/to the GF(2^8). This paper also evaluates the efficacy of the proposed circuit by means of gate counts and logic synthesis with a 65 nm CMOS standard cell library and comparisons with conventional circuits, including those with tower fields GF(((2^2)^2)^2). Consequently, we show that the proposed circuit achieves approximately 40% higher efficiency in terms of area-time product than the conventional best GF(((2^2)^2)^2) circuit excluding isomorphic mappings. We also demonstrate that the proposed circuit achieves the best efficiency (i.e., area-time product) for an AES encryption S-Box circuit including isomorphic mappings.
Category / Keywords: Compact hardware implementation, GF(2^8) inversion, S-Box, AES.
Original Publication (in the same form): IACR-CHES-2015
Date: received 30 Jul 2015, last revised 1 Aug 2015
Contact author: ueno at aoki ecei tohoku ac jp
Available format(s): PDF | BibTeX Citation
Version: 20150802:054034 (All versions of this report)
Short URL: ia.cr/2015/763
Discussion forum: Show discussion | Start new discussion
[ Cryptology ePrint archive ]