An Energy Efficient Datapath for Asymmetric Cryptography

Public key encryption on resource constrained devices can be a burden in terms of latencies per operation but also in terms of energy consumed. For certain applications, such as WSN and RFID tags, the energy budget is quite small. Understanding the energy cost of each public key operation is important. This paper explores the energy consumption of public key cryptography by presenting a parameterizable microarchitecture for accelerating public key operations and demonstrating how the datapath width and key size effects the energy consumption per Montgomery multiplication, a key component of public key encryption algorithms. We find that for O(n) algorithms, like Montgomery multiplication, a larger datapath provides optimal energy efficiency. Moreover, our microarchitecture is capable of reducing the energy consumption of Montgomery multiplication by a factor of 50 while providing a speed up of 10x for a key size of 192-bits over the traditional software implementation on a low-power embedded processor.