Abstract
In order to understand the development of current channel energy analysis technology and protection strategy, at the same time, we will lay the foundation for establishing safety evaluation system about Encryption device in the next stage. We analyzed the side-channel attacks of the current mainstream encryption algorithms. The side-channel attack is classified by combining the specific operations of the algorithm that are easily attacked by the attacker. Aiming at the more popular DPA attacks, the paper focuses on the corresponding protection strategies, including the first-order DPA attack defense strategy and the second-order DPA attack defense strategy. Finally, the evaluation index of the current safety assessment system is summarized, and the effect of each index is elaborated.
Introduction
Traditional cryptography is built on the assumption that an adversary only has access to inputs (plaintext) and outputs (ciphertext). However, in the real world, cryptographic hardware leaks information through physical "side channels" such as power consumption, execution time, and electromagnetic emissions. This paper addresses the gap between mathematical security and physical implementation. Even if an algorithm is mathematically "unbreakable," its physical execution can reveal secret keys to an attacker using techniques like Differential Power Analysis (DPA). The authors systematize the current understanding of these attacks and propose a rigorous safety evaluation system to measure a device's resistance to them.
Algorithmic Vulnerability Analysis
The core of a side-channel attack (SCA) isn't breaking the math; it's exploiting the intermediate values generated during execution. If an attacker can predict an intermediate state based on a key guess, they can correlate that guess with physical power traces. The paper identifies specific "points of attack" within common algorithms.
Symmetric Ciphers: AES and DES
In AES-128, attackers typically target the first round's S-box output using DPA. By analyzing the power consumption during these specific operations, the secret key can be extracted.
Similarly, DES is highly vulnerable on hardware like IC cards. The 48-bit subkey search space can be drastically reduced by attacking 6-bit groups within the iterative rounds of the Feistel network.
Asymmetric Ciphers: RSA
For RSA, vulnerabilities often lie in modular exponentiation. Timing attacks and cache-hit analysis can successfully crack substantial portions of a key (e.g., 340 bits of a 512-bit key), significantly narrowing the search space for an adversary.
Defense Strategy Classification
Protection methods against SCA are categorized into two main types:
- Hiding: Reducing the signal-to-noise ratio (SNR) by randomizing the timing of operations (e.g., inserting pseudo-operations) or the amplitude of power consumption to mask the useful signal.
- Masking: A mathematical approach where intermediate values are combined with random "masks" (Boolean or arithmetic) so that the physical leakage becomes statistically independent of the secret data.
While masking is a powerful defense, it often fails in practice due to "mask reuse" or "zero-value" vulnerabilities. Furthermore, while first-order masking protects against basic DPA, "second-order" attacks can bypass this by predicting two intermediate values simultaneously and analyzing their joint leakage.
Security Evaluation Framework
The authors argue that security should be measured by the cost to the attacker rather than just a binary status. They propose five critical indices for security assessment:
- Success Rate: The probability of a correct key crack.
- SNR (Signal-to-Noise Ratio): Quantifies how much "useful" leakage exists relative to background noise.
- Sample Size: The number of power traces/measurements required to crack the key (a proxy for attack cost).
- Guess Entropy: The average workload remaining for an attacker after the side-channel information is processed.
- Mutual Information: Uses information theory to quantify the dependency between the secret key and the leaked signal.
Conclusion
While defenses against first and second-order DPA are becoming better understood, high-order DPA remains a significant threat that requires further study. Current research into security assessment is lagging behind the rapid development of new attack methods. Future work should focus on establishing a more robust, standardized evaluation system that can keep pace with increasingly sophisticated "joint leakage" attacks.
Summary updated