Pairings are cryptographic primitives which are now used in numerous protocols. Computing and implementing them efficiently is then an interesting challenge relying on an algorithmic and arithmetic study of those mathematical functions. More precisely, pairings are bilinear maps defined over elliptic and hyperelliptic curves. Among those, we restrict our study to supersingular curves, as they allow both symmetric pairings and efficient algorithm for pairing computation. We propose an unified framework for the construction of algorithms computing pairings and we apply it to the design of a novel algorithm for a pairing over a genus-2 characteristic-2 hyperelliptic curve. The computations involved in our algorithms require the implementation of rapid arithmetic for finite fields of small characteristic. Since multiplication is the critical operation, we present an algorithm for the exhaustive search of multiplication formulae. Finally, we apply all the previous methods to the design and implementation of different hardware accelerators for the computation of cryptographic pairings over various curves.
