I have following requirement for software I'm working on:
Use strong asymmetric key exchange with public key size <16 bytes.
I know it is pretty hard for this amount of bytes, but I've came up with following solution and wanted some feedback:
Use Diffie Hellman Key Exchange algorithm with P (prime modulus) parameter <2^128, and then use some key derivation algorithm like pbkdf2 on generated shared secret to generate much stronger symmetric key, and use this key to encrypt any data passed.
I want to know if this will improve safety of such protocol compared to simple Diffie Hellman over 2^128.