Design of reliable and secure devices by algebraic manipulation codes
In this thesis, we firstly present the secure multipliers protected by the AMD codes, and demonstrate that the fault masking probabilities are not worse than the results based on the theoretical analysis of error masking probabilities, if the attacker injects faults at outputs of the inside logic gates of the protected devices. Single-errorcorrecting, double-error-detecting (SEC-DED) codes are widely used for the design of errors, thus they are not suitable for memories used in cryptographic devices. Algebraic Manipulation Detection (AMD) codes provide strong protection against fault injection attacks. But traditional AMD codes can not be used for correcting errors. In this thesis, we also present the constructions of the strongly secure algebraic manipulation correction (AMC) codes. The estimation for a probability for miscorrection of multiple errors is given. Hardware implementations of strongly secure SEC-DED memories based on the proposed codes are presented. Comparison with other codes which have been used for SEC-DED memories with security or weak security are given in terms of numbers of undetected errors, sizes of security kernels and miscorrection probabilities as well as latency, area and power consumption for encoders and decoders. An error handling method to distinguish between random errors and fault injection attacks is presented as well. The proposed code can be applied to most secure-critical memories in cryptographic devices. As far as we know, this is the only efficient approach to provide both high reliability for single and double random errors, and high security for strong fault attack when an attacker has a control of both on the messages (outputs of the memories) and the errors.
Thesis (M.Sc.Eng.) PLEASE NOTE: Boston University Libraries did not receive an Authorization To Manage form for this thesis or dissertation. It is therefore not openly accessible, though it may be available by request. If you are the author or principal advisor of this work and would like to request open access for it, please contact us at firstname.lastname@example.org. Thank you.