An updated security analysis of PFLASH

Abstract

One application in post-quantum cryptography that appears especially difficult is security for low-power or no-power devices. One of the early champions in this arena was SFLASH, which was recommended by NESSIE for implementation in smart cards due to its extreme speed, low power requirements, and the ease of resistance to side-channel attacks. This heroship swiftly ended with the attack on SFLASH by Dubois et al. in 2007. Shortly thereafter, an old suggestion re-emerged: fixing the values of some of the input variables. The resulting scheme known as PFLASH is nearly as fast as the original SFLASH and retains many of its desirable properties but without the differential weakness, at least for some parameters. PFLASH can naturally be considered a form of high degree HFE− scheme, and as such, is subject to any attack exploiting the low rank of the central map in HFE− . Recently, a new attack has been presented that affects HFE− for many practical parameters. This development invites the investigation of the security of PFLASH against these techniques. In this vein, we expand and update the security analysis of PFLASH by proving that the entropy of the key space is not greatly reduced by choosing parameters that are provably secure against differential adversaries. We further compute the complexity of the new HFE− attack on instances of PFLASH and conclude that PFLASH is secure against this avenue of attack as well. Thus PFLASH remains a secure and attractive option for implementation in low power environments.