Invariance-Based Concurrent Error Detection for AES
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Concurrent Error Detection (CED) techniques for cryptographic chips, such as those implementing the Advanced Encryption Standard (AES), face challenges in balancing hardware and time overhead while effectively detecting both transient and permanent faults, with existing methods either requiring excessive resources or being limited to detecting only transient faults.
Innovation Solution
The proposed invariance-based CED technique exploits the inherent properties of AES by inserting additional error detection rounds with specific input values to the ByteSub and AddRoundKey functions, enabling detection of faults in the ByteSub, ShiftRow, MixColumn, and AddRoundKey operations with minimal overhead, using two extra rounds to achieve comprehensive fault coverage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Hardware Redundancy is used for CED, then error detection capability is improved, but hardware overhead increases to ≧100%
Solution Approach 1:
The patent applies preliminary action by inserting CED rounds with specific test inputs (0x7E and 0x81) before normal encryption operations to detect faults in advance. These preliminary detection rounds check for permanent faults in the S-boxes and other AES components without requiring full redundant hardware, thus improving error detection capability while limiting hardware overhead to a manageable level.
Solution Approach 2:
The patent changes parameters by using specific input values (0x7E and 0x81) that exploit the invariance properties of the AES S-boxes. When these specific inputs are applied, any permanent fault in the S-box will produce a detectable deviation in the output. This parameter-based approach allows fault detection without requiring duplicate hardware, resolving the contradiction between detection capability and hardware overhead.
2Quantity of substance
If Time Redundancy is used for CED, then hardware overhead is reduced, but time overhead increases to ≧100% and only transient faults are detected
Solution Approach 1:
The patent uses preliminary action by performing detection rounds before normal encryption operations. The CED rounds with inputs 0x7E and 0x81 are executed first to detect permanent faults, and only if these pass does the normal encryption proceed. This preliminary detection approach minimizes time overhead because the expensive detection is only performed when needed, not continuously, and permanently reduces hardware overhead compared to full time redundancy schemes.
Solution Approach 2:
The patent changes the input parameters to specific values (0x7E and 0x81) that reveal permanent faults through the invariance properties of AES. By changing from arbitrary inputs to these specific test inputs, the system can detect permanent faults (not just transient faults) without requiring continuous redundant operation, thus reducing both time and hardware overhead while improving fault detection coverage.
3Quantity of substance
If inverse relationship CED approach is used, then area overhead is reduced to 18.9% at round level, but time overhead increases to 26.55%
Solution Approach 1:
The patent applies preliminary action by inserting CED rounds at specific positions within the AES encryption process rather than performing full re-computation. The detection rounds are performed preliminarily at strategic points (after certain rounds) to catch faults early, reducing the need for complete time redundancy while maintaining area efficiency through the inverse relationship approach.
Solution Approach 2:
The patent uses periodic action by inserting CED rounds at periodic intervals during the encryption process rather than continuously. The detection rounds are performed periodically at specific stages (e.g., after certain AES rounds) to balance detection coverage with time overhead, improving upon the continuous monitoring requirement of full time redundancy while maintaining the area efficiency of the inverse relationship method.
Data Source
AI summary
Advanced Encryption Standard (AES) is an encryption algorithm for securing sensitive unclassified material by U.S. Government agencies and, as a consequence the de facto encryption standard for commercial applications worldwide. Performing concurrent error detection (CED) for protection of such a widely deployed algorithm is an issue of paramount importance. We present a low-cost CED method for AES. In this method, we make use of invariance properties of AES to detect errors. For the first time, the invariance properties of the AES, which are for the most part used to attack the algorithm, are being used to protect it from fault attacks. Our preliminary ASIC synthesis of this architecture resulted in an area overhead of 13.8% and a throughput degradation of 16.67%.