Journals Proceedings

Abstract

In this paper we have proposed three sub-pipelined architectures for Encryption, Decryption and Joint Encryption and Decryption (E/D). These architectures were implemented on Vertex-4 device. The use of Block RAM available in the device for key expansion as well as for the S-Boxes resulted in utilizing less slices and getting higher throughput in all three cases compared to the literature available till date. The encryption architecture clocked a throughput of 35.65Gbps using only 4823 slices while the decryption architecture achieved 33.73Gbps using 6847 slices only. The device used is XC4VLX60. The joint E/D architecture achieved a throughput of 31.62Gbps. Retiming techniques used to balance the computational path delays of encryption and decryption data paths.