Key generation and delivery Sample Clauses

Key generation and delivery. Key generation. Xxxxx, the initiator, is in charge of generating a key with strong randomness, using any existing algorithm. The bit string for the secret key should be (1) sufficient long, i.e., more than 128 bits in common, and (2) statistically random. Feature pairing. After computing the features for 0/1 bits and generating the key, TDS needs to perform feature pairing, i.e., choosing one from features of 0-bit and one from features of 1-bits and making them a pair to represent a bit. The simplest solution is to make the two features computed from a same group of samples to be a pair. However, one disadvantage is that it is possible that in some pairs the two features are close and hence make it easier to produce mismatched bits. Hence this feature pairing step is to find an optimal strategy of making the difference of feature values for each 0/1 pair larger than a certain threshold θ. We take a paring algorithm based on Max-Weighted Bi- partite Matching to solve this problem. We assemble all the ˆ n×n

Key generation and delivery. Key generation. Xxxxx, the initiator, is in charge of generating a key with strong randomness, using any existing algorithm. The bit string for the secret key should be (1) sufficient long, i.e., more than 128 bits in common, and (2) statistically random. Feature pairing. After computing the features for 0/1 bits and generating the key, TDS needs to perform feature pairing, i.e., choosing one from features of 0-bit and one from Key delivery and information reconciliation. To represent the sequence of generated bits Ka whose length is l, Xxxxx selects l features from l pairs in her S-box Sa. She then sends the features to Xxx and Xxxxxx. Xxx and Xxxxxx use their S-boxes S−1 and Sc−1 to decode the key. TDS uses existing information reconciliation method [3] [28] to create consistent keys on difference devices. After information rec- onciliation, Xxx’s key will be corrected to Xxxxx’s. If there are more than two devices, Xxxxx will run information rec- onciliation to every of them in order. Information reconcil- iation includes an information-theoretically secure authen- tication using universal hashing [17]. Secure authentication can also defend against impersonation attack, in which an attacker pretends to be Xxxxx and send a sequence of feature Algorithm 1: KM based feature mapping Input: C0,C1,θ; if the bit error rate is smaller than a reasonable threshold,