Cryptographic hardware accelerator with dynamic reconfigurable redundancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cryptographic hardware accelerators are vulnerable to fault injection attacks, which require significant preparation time and are not always feasible, and side-channel analysis attacks exploit fixed register placements, necessitating improved resistance mechanisms.
Innovation Solution
Implementing a cryptographic hardware accelerator with dynamically reconfigurable redundancy and discrepancy detection, using a controller to selectively enable discrepancy and voting operations, and employing dummy-addressing to camouflage energy consumption patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed register placement is used in cryptographic hardware accelerators, then manufacturing and implementation is simplified, but the device becomes vulnerable to side-channel analysis attacks
Solution Approach 1:
The patent implements dynamic reconfiguration of register placements during cryptographic operations. The register file can be repositioned or reassigned different addresses between operations, preventing attackers from correlating fixed physical locations with logical register addresses. This dynamic behavior eliminates the static patterns that side-channel analysis exploits, while maintaining implementation feasibility through controlled reconfiguration sequences.
2Reliability
If redundancy mechanisms are added to protect against fault injection attacks, then security is improved, but chip area and energy consumption increase
Solution Approach 1:
The patent makes the existing register file serve multiple functions: it acts as both the operational register storage and the redundancy mechanism for fault detection. By dynamically reconfiguring the register placements and using the same physical resources in different configurations, the system achieves redundancy protection without requiring separate dedicated redundancy circuits, thus avoiding additional chip area overhead.
Solution Approach 2:
The system changes the addressing parameters and logical organization of registers dynamically rather than adding redundant physical structures. By varying register addresses, assignments, and configurations between operations, the system creates redundant computational paths using the same hardware resources, providing fault injection resistance without increasing chip area.
3Measurement precision
If discrepancy detection operations are performed continuously, then detection of faults is improved, but energy consumption increases
Solution Approach 1:
The patent implements periodic or selective discrepancy detection rather than continuous monitoring. Discrepancy detection operations are performed at strategically chosen intervals or triggered by specific events during cryptographic operations, balancing fault detection capability with energy conservation. This periodic approach maintains security while significantly reducing the energy overhead compared to continuous detection.
Data Source
Figure 1
Figure 2
Figure 3~5
AI summary
The invention is directed to a cryptographic hardware accelerator (100) for performing a computational operation in a cryptographic application, comprising one or more addressable blocks (102), a databus (106) and a controller (108). When a single target block (102.N) is addressed the controller is configured to selectively enable a discrepancy detection operation, wherein a comparison block (102.4) is selected and the target block (102.N) and the comparison block (102.4) selected are addressed for performing the discrepancy detection operation. The cryptographic hardware accelerator further comprises a discrepancy detection unit (100) configured to provide a discrepancy detected signal (DS) indicative of whether or not respective contents (C) of the target block (102.N) and the comparison block (102.4) differ. When the discrepancy detection operation is enabled, the controller is configured to perform a write-to-bus operation upon determining that the discrepancy detected signal (DS) is indicative of non-differing contents.