Parallel mapping implementation of AES for Coarse-Grained Reconfigurable Block Encryption Array

Abstract

In this paper, we explore different parallel mapping implementations of Advanced Encryption Standard (AES) on the Coarse-Grained Reconfigurable Block Encryption Array(RBEA). To improve performance and resource efficiency of the direct mapping implementation, we propose 8 modified implementations based on merger, task duplication and loop unrolling methods. Experimental results show that the improved implementations increase the throughput from 1023.99 Mbps to 13643.96 Mbps and the performance-area ratio from 63.99 Mbps/FB to 309.62 Mbps/FB, developing the superiority of array's computing resources. Compared with other platforms, our work has 1.16–13.34 times higher throughput and 3.55–11.73 times higher throughput per unit of array area.