''Seed perso'' means in the context of security tokens the process of storing the secret seed/key on the token. The seed for a Display-TAN card consists of 160 bits.
The following seed perso concept for Display-TAN has the advantage that the bank's card-producer/-personaliser is not involved in the Display-TAN seed perso. This means that no extra machines or extra processes are needed at the bank's card-producer/-personaliser: the only change is that the usual input plastic cards are replaced by Display-TAN cards. In case the EMV chip is contactless and needs an NFC-antenna, the Display-TAN card is delivered with an additional NFC-antenna (same contact points, no connection to the Display-TAN parts).
The main idea is to let the bank customer trigger the final seed-perso process on his smartphone: When the bank customer during a banking session on his smartphone turns on the Display-TAN card the first time a bidirectional communication chain
is set up by the banking app on the bank customer's smartphone. A preliminary seed on the card allows to execute this final seed-perso in an encrypted way, end-to-end, i.e., Bank Server to Card.
The following diagram is showing the seed perso process.
More detailed, the suggested Display-TAN seed perso works the following way:
The only way to compromise the seed perso is by (A) stealing the pre-seeds (in Taiwan, on the CD, at the bank server) and then (B) interfering into the key exchange process as a virus within the Internet or on the smartphone of the bank customer. Managing (A) or (B) is hard, managing both seems to be practically impossible. Note that even simple sabotage (= damage without use for the attacker) is not possible.
It may be an issue that a physical thief or a smartphone trojan may steal the online password and intialize another card which he/it received as a customer of the bank. In other words: the breach of the first factor allows a breach of the second factor - at least for a certain period of time. If this is considered a problem there is the following modification of the solution which prevents this attack: When the letters containing the new bank card are send to the customer, a random OTP for this card is provided within that letter. The personalization of the Display-TAN part of the card proceeds like described above but with the additional step that the bank customer enters this OTP into the smartphone. This extended solution prevents the attack described above. Its disadvantages are (1) the bank card personalizer has this additional step of generating the OTP (but it only requires a software change and does not affect the card production), and (2) the user is explicitly involved into the personalization process whereas without this OTP step the personalization could fully be hidden from him.
This Display-TAN seed perso concept was developed by B. Borchert and K. Reinhardt, Feb. 2016.