Blockchain based biometric matching

Abstract

Blockchain technology has become a popular topic, making it possible to remove trusted intermediaries. This is applied to biometric technology where matching operations are executed in a decentralised trust-less environment via smart contracts. Three smart contract based blockchain based models are developed where practical trade-offs amongst them experimentally explored, namely execution time, transaction cost, complexity and the level of decentralisation between them. We find that with the present price of Ether common distance algorithms used for biometric matching such as the Euclidean distance will cost around $8. Costs can be reduced by introducing blockchain oracles, but this reduces the decentralisation level to that of a centralised system. Timings of oracle based transactions settling on the Ethereum main chain will also take twice as much time (70 seconds) of transactions made directly on-chain due to requiring the mining of two blockchain transactions. We find that a proof-of-stake sidechain is the best model tested for this purpose, providing for near instant transaction time (3 seconds), being low cost ($0.001) while not being overly complex and still enjoying a level of execution in a decentralised environment.