Rescheduling for optimized SHA-1 calculation

Abstract

This paper proposes the rescheduling of the SHA-1 hash function operations on hardware implementations. The proposal is mapped on the Xilinx Virtex II Pro technology. The proposed rescheduling allows for a manipulation of the critical path in the SHA-1 function computation, facilitating the implementation of a more parallelized structure without an increase on the required hardware resources. Two cores have been developed, one that uses a constant initialization vector and a second one that allows for different Initialization Vectors (IV), in order to be used in HMAC and in the processing of fragmented messages. A hybrid software/hardware implementation is also proposed. Experimental results indicate a throughput of 1.4 Gbits/s requiring only 533 slices for a constant IV and 596 for an imputable IV. Comparisons to SHA-1 related art suggest improvements of the throughput/slice metric of 29% against the most recent commercial cores and 59% to the current academia proposals. © Springer-Verlag Berlin Heidelberg 2006.