# Symmetric Encryption, meet in the middle attack, effective key length

What is the effective key length of a deterministic encryption scheme, given a meet-in-the-middle attack scenario when you encrypt five times (64 bit keys for each)?

.

Say the encryption scheme has |key space K| = |plaintext space M| = |ciphertext space C| , all keys are 64 bit length, and $$k1,k2,k3,k4,k5$$ are all different (no key reuse like in some DES key options).

$$c = E_{k5}(E_{k4}(E_{k3}(E_{k2}(E_{k1}(m))))$$

You know that double encryption would not result in $$|K|^2$$ possibilities, but in fact due to the mitm attack the possibilities the attacker is forced to iterate over is only around ~2*K. You also know that triple encryption is much better, as it forces the attacker to have to iterate a total of ~$$|K|^2$$ even with the mitm tactic. What is the effective key length then when using double encryption, triple encryption, and x5 encryption respectively when using the above scheme?

• Hint: Can you perhaps group the five encryptions into two "abstract encryptions" to get a double encryption to apply MITM? Sep 21, 2020 at 9:09
• My interest really lies in the fourth iteration; does the "effective doubling" of our encryption per iteration (work the attacker has to do for mitm) continue to increase once we pass triple encryption? Such as with 3DES, reaching an effective key length of 2^112, on iteration four does it achieve 2^224 effective key length? Sep 22, 2020 at 22:56
• Or do we need to consider the encryption in bunches of three, given the nature of how the attacker makes the list of pre-processed k values and runs them all through the "left side" of the encryption equation? Sep 22, 2020 at 22:58