NOTE: Restrictions are in place to limit access to one or more of the files associated with this item. Authorized users must log in to gain access. Non-authorized users do not have access to these files.
ALU for mbedTLS Diffie-Hellman Parameters Generator on FPGA Embedded Processor System
MetadataShow full item record
Safe prime is a unique subset of the general prime number where both p and (p-1)/2 are primes. Commonly used Public Key encryption scheme Diffie-Hellman key exchange algorithm utilizes ultra large safe primes as the private key. In practice, crypto software libraries implement a specific Diffie-Hellman parameters generator that searches for safe primes with Rabin-Miller probabilistic primality test algorithm. Without any proven theory to predict their occurrences among natural numbers, generator programs generally start at a randomly seeded odd positive integer of a predetermined size; and perform primality tests in iterations over incrementing candidates until success. The staggeringly low density of safe primes causes a prohibitive amount of computing resources to be dedicated in the generation process. As the result, power conscious mobile and embedded devices can no longer compute the standard 2048-bit safe primes without causing prolonged disruption to the overall system performance. Based on the hot path analysis of the generator program, a parallelized and pipelined ALU is proposed and implemented on the FPGA embedded processor system. Utilizing merely 3% of LUT (584/17600) and 20% of DSP (16/80) available from the Xilinx Zynq 7010 All Programmable SoC, the suggested design is theoretically capable of offsetting more than 90% of CPU utilizations needed for the entire safe prime generation process. Such results demonstrate the deficiency of today's general purpose CPU in handling certain complex and resource intensive computations. Such scenarios greatly incentivize the integration of programmable hardware with fixed design CPU. Additional research is suggested to focus in the area of automating the processes of locating the specific CPU intensive task, translating such task onto programmable hardware, and providing software accessible interface to enable fast development and deployment of the hot function based programmable hardware design. From there, programmable hardware assisted computing platforms can be further enhanced to dynamically program hardware modules based on real-time utilizations to achieve even greater overall system performance. A new system design paradigm can potentially be introduced as the result.
Chen, Changning; Dempsey, Brian (2017). ALU for mbedTLS Diffie-Hellman Parameters Generator on FPGA Embedded Processor System. Undergraduate Research Scholars Program. Available electronically from