What can be improved so that the the susceptibility to frequency analysis attack problems of alphabetic subsitution methods like Caesar Cipher and Monalphabetic Substituion can be removed?

We can use a different set of encryption mapping in the order of each alphabet encryption. Suppose we have the sentence again

ATTACK STARTS ON SEVEN TONIGHT

We now have three different set of mappings for:

  • ABCDEFGHIJKLMNOPQRSTUVEXYZ

===================================

  • IXSYJECTFHRVBZUAKQWDNLOPMG
  • NFMXGLOYCAPRWQSVBDEHTZJIUK
  • ABGQHRLSCDMOTNUZEFIPVYJWXK

We map first of the above message’s alphabet to mapping 1, then second alphabet to mapping 2 and then the third one to mapping 3 and the fourth one back to mapping 1 and so on and so forth…..

So the encrypted message is

IHPIMM QHPIDPW ……………….

this is called Polyalphabetic Substituion Encryption

This encryption defeats the problem of a constant mapping of a particular alphabet to another alphabet as in the monoalphabetic and caesar cipher. And hence renders the Frequency Analysis Attack to unusable.
If we have, say 9 different mapping systems. Then if we encrypt a message using a key of say 34135, that means we encrypt the message using the third mapping for first alphabet and then the fourth one for second, the first one for the third alphabet, the third one again for the fourth alphabet and the fifth one for the fifth alphabet. The third one would be used again for the sixth alphabet.

So the decryption is easy if we have the key ‘34135’ and the nine set of mapping systems that should be shared among the sender and the receiver.