The Researches on the ASIP of ECC in Embedded Domain
潘振宽,程相国,秦尧,夏辉,于佳,陈仁海
DOI: https://doi.org/10.11897/SP.J.1016.2017.01092
2017-01-01
Abstract:Compared with other public-key cryptographies,elliptic curve cryptography (ECC) possesses the advantages of the good ability of resistance to the attack,reduced calculation and fast processing speed,and it has become the public key cryptosystem standard of the next generation.With the widespread application of ECC in embedded domain,how to increase the effect becomes a hotspot in the current study.This paper proposes a general design scheme for the application specific instruction processor,which is applied to ECC,intending for the obvious promotion of execution efficiency in the embedded environment of limited hardware resources.First of all,optimize the software for the implementation processes of ECC based on the OpenSSL open software optimization method.Considering the characteristics of the processor platform,make further improvements of the filed multiplication arithmetic and the polynomial squaring arithmetic.Secondly,divide the basic instruction blocks (BIBs) for the optimized realization of algorithm and transfer them into the corresponding data flow graphs (DFGs).Locate the instruction blocks that can be optimized in each DFG in accordance with the special instruction design rules,by virtue of the approximate optimal solution method.And then,design the special multiple corresponding instruction for this type of instruction blocks,to realize the extension of the original basic instruction set architecture (ISA) of the processor.Next,design successively and verify the system level model and the Verilog register transfer level model of ECC_ASIP in turn with the electronic system level designing (ESL) approach introduced.In the end,synthesize,place and route the RTL processor model comprehensively,which has passed the verification,and convert it into the corresponding gate-level circuits and calculate out the resource information that the model use.Finally,write the information into FPGA to finish the transplantation and performance verification of ECC_ASIP.Compared with the ARM11 processor model,it has been proved in the experimental results that,after a series of simulations,at the expense of the increase of 9.23x％ hardware extension resources,the ECC_ASIP has increased the speed of algorithm implementation for 2.74x times and saved 59.36x％0 instruction code storage space.