This paper surveys the literature on certificateless encryption schemes. In particular, we examine the large number of security models that have been proposed to prove the security of certificateless encryption schemes and propose a new nomenclature for these models. This allows us to "rank" the notions of security for a certificateless encryption scheme against an outside attacker and a passive key generation centre, and we suggest which of these notions should be regarded as the "correct" model for a secure certificateless encryption scheme. We also examine the security models that aim to provide security against an actively malicious key generation centre and against an outside attacker who attempts to deceive a legitimate sender into using an incorrect public key (with the intention to deny the legitimate receiver that ability to decrypt the ciphertext). We note that the existing malicious key generation centre model fails to capture realistic attacks that a malicious key generation centre might make and propose a new model. Lastly, we survey the existing certificateless encryption schemes and compare their security proofs. We show that few schemes provide the "correct" notion of security without appealing to the random oracle model. The few schemes that do provide sufficient security guarantees are comparatively inefficient. Hence, we conclude that more research is needed before certificateless encryption schemes can be thought to be a practical technology.
