Researches On Blind Recognition Of Linear Scramblers

As a key technique of modern communication, scrambling is a bit-levelrandomizing processing applied to the data just prior to transmission for the goal ofimproving the clock recovery capability and data transmission security. Linearscrambler is the only used scrambler in actual communication system, correctly andeffectively recovering a linear scrambler directly influences the obtainment of thetransmission message. The blind recognition of a linear scrambler is a complex anddifficult problem in the field of communication, and it is also what this thesis focuseson.First, for the estimation of the order of scrambler, this thesis improves the orderestimating algorithm of a synchronous scrambler, after theoretical analysis andsimulation of all the existing methods. The improved algorithm reduces the neededamount of scrambled data, and also expands its own application range for the sourcewhich is statistically biased.Second, the thesis researches methods to recognize the generating polynomial of ascrambler. All the existing methods need additional priori knowledge besides thescrambled sequence, and their performances are not so good. On the condition ofbiased information source, this paper lucubrate the imbalance of value states and thedifference of code weight which presented by the scrambled sequence in the positionof the generating polynomial’s nonzero term, and proposes two novel blind recognitionmethods of the generating polynomial. The two proposed algorithms use differentdistance measures between probability density functions to be measure index of theimbalance and the difference. The simulation and contrastive analysis verifies that thealgorithms are practical and effective.Third, based on the theory of fast correlation attacks on stream ciphers, this thesisdeduces error equations which the initial state of a synchronous scrambler satisfied infinite field (Galois field), and uses Walsh-Hadamard transform to recover the initialstate. The algorithm reduces the requirement of source bias; what’s more, the numberof nonzero term of the generating polynomial makes no influence on the algorithm’s performance.Finally, the paper gives complete and detailed scrambler recognizing systemunder two different conditions: the scrambling type known, the scrambling typeunknown. Furthermore, this paper proposes a method to recognize the scrambling typewhen designing the system framework. Theoretical analysis and simulation resultsverify the actual validity of the algorithm and its good performance.