Secret Key Specification for a Variable-Length Cryptographic Cellular Automata Model

Deeper Investigating Adequate Secret Key Specifications for a Variable Length Cryptographic Cellular Automata Based Model

“…As a pattern associating parameters with the worst performing rules in ciphering was not possible to recognize by a simple visual inspection, it was applied a data mining process (Fidelis et al, 2000). A standard GA was elaborated to mining IF-THEN rules associating the parameters (antecedent) with the performance class (consequent) (Oliveira et al, 2010c). The antecedent part of such rules is composed by a conjunction of conditions of the type: IF <parameter> <operator> <value>, being that the operator can be < (minor), ≥ (larger or equal) or = (different).…”

“…However, we decided to kept Rule ND as one of the best filters because the generalization of Neighborhood Dominance (ND) parameter from radius 2 to bigger radius is better than Absolute Activity (AA) parameter, as discussed in (Oliveira et al, 2000). Some experiments using radius 3 toggle rules specified using a filter rule based on ND are presented in (Oliveira et al, 2010c). The rule based on ND presented here is better than this previous filter rule which was obtained in preliminary data mining analysis.…”

“…The tests performed using radius 3 rules returned adequate mean values. Moreover, no rule returns low entropy in any ciphered plaintext (Oliveira et al, 2010c).…”

Decision Tree-Based Ensembles To Specify Secret Keys For Cellular Automata Cryptographic Model

“…CA rules used as secret keys in VLE must be properly specified to exhibit two characteristics: (i) they must have low probability of failure during pre-image computation to returning a ciphertext length close to the original text; (ii) they must add a high level of entropy during the encryption to ensure a good ciphering quality. In [14], the secret key specification was better investigated. First, a representative rule set formed by all radius 2 right-toggle rules was analyzed.…”

“…The main conclusion is that there are some undesirable behavior rules in the complete set that must be avoided as secret keys. An analysis based on CA static parameters [13] was employed in [14] to capture the pattern associated to underperforming rules. Using nine static parameters and a genetic algorithm (GA) [15] to mine this pattern, it was able to find a good specification of rules to be used as valid secret keys.…”