Side channel vulnerability assessment method based on causal inference

Abstract

The invention discloses a side channel vulnerability assessment method based on causal inference. The method comprises the following steps: firstly, constructing a structural relationship between an inner side channel of a system and system security, then respectively selecting an adjustment set of each side channel, further calculating unbiased influence of each side channel on the system security, and finally, depicting vulnerability of each side channel based on the unbiased influence, thereby providing a theoretical basis for a defense party to actively counteract side channel attacks. The method is a first side channel security evaluation scheme under a causal perspective, and solves the problem of the existing pain points of the local perspective and the later visibility caused by defending the side channel attack based on an experimental method. According to the embodiment of the invention, the method can guarantee the vulnerability discovery correctness in qualitative and quantitative aspects while achieving the universal and processized side channel vulnerability discovery.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of system security, and specifically to a method for a defender to actively counter an attacker based on a global perspective in a side channel attack scenario, specifically to a side channel vulnerability assessment method based on causal inference. Background Art

[0002] Side-channel attacks involve the attacker exploiting side-channel information obtained during the operation of an electronic system (such as instruction execution timing, power consumption, and sound), rather than relying on brute force searches or theoretical flaws in algorithms, to crack and obtain target information. Due to their stealth and effectiveness, side-channel attacks can incur significant security risks at low cost, undermining the confidentiality and integrity of private information and potentially destroying the beneficial data-based ecosystem.

[0003] Existing mainstream side-channel attack analysis approaches fall into two categories: one is device-specific, customized vulnerability discovery. By exploring feasible attacks against a specific device, these attacks are reported to the community for timely patching and protection against the corresponding attacks. The other focuses on integrated defense research based on common vulnerabilities. This approach focuses on a side channel known to present security risks across different devices and designs unified defense mechanisms for that side channel. However, both approaches suffer from attack vector dependence and a localized perspective. Attack vector dependence refers to the fact that existing work only targets side channels for which specific attack vectors can be identified, while ignoring side channels that present security risks but for which no viable attack vectors have yet been identified. This premature defense strategy puts defenders in a passive position. The localized perspective, on the other hand, means that during the attack and defense process, attackers only need to target a single side channel to launch an attack, while defenders are required to secure all side channels to ensure system security. In other words, to achieve comprehensive defensive effectiveness, defenders must maintain a holistic perspective to protect against all potentially vulnerable side channels. Summary of the Invention

[0004] In order to overcome the shortcomings of the above research, we propose a side channel vulnerability assessment method based on causal inference to provide a solution for proactively countering side channel attacks from a global perspective. Figure 1 As shown, the steps are as follows:

[0005] Side channel relationship building module: This module analyzes the data flow within a specific attack scenario and the mutual influence between side channels to construct the structural relationship of the side channels within the system.

[0006] Side channel unbiased impact estimation module: calculate the specific side channel C i Unbiased effect on system safety T;

[0007] Side channel vulnerability calculation module: Based on the obtained unbiased effect of the side channel on system security, the vulnerability of the side channel in the system is quantified.

[0008] A further technical solution is to use nodes in the side channel relationship construction module to represent components including each side channel and system security indicators, and use edges to represent the information flow relationship between nodes.

[0009] A further technical solution is that the side channel unbiased impact estimation module includes the following steps:

[0010] Adjustment set selection module: used to identify the confounding factors that affect the side channel C i and the node of system security T, if node A is not a side channel C in the established side channel relationship i Descendant nodes of the node, while node A can block all side channels C i Node and system security T node points to C i path, then node A will be selected as side channel C i Elements in the adjustment set;

[0011] Unbiased impact calculation module: for a specific side channel C i , the selected adjustment set is recorded as Then the specific side channel C i The unbiased impact on system security T satisfies Among them, a i express The possible specific values ​​of each element.

[0012] A further technical solution is that the side channel unbiased impact estimation module needs to calculate based on the output result of the side channel unbiased impact estimation module, and for the side channel C under a specific attack scenario i , its vulnerability D(C i )for: in Where P(T=t|do(C i =c)) is the output of the side channel unbiased impact estimation module, indicating C i The unbiased causal effect between and T, and Indicates that C is taken into consideration i The impact of all possible states on system security. BRIEF DESCRIPTION OF THE DRAWINGS In order to more clearly illustrate the implementation of the present invention or the existing technical solutions, the following briefly introduces the drawings required for use in the embodiments or the description of the existing technologies. Figure 1 The figure is a flowchart of the overall framework of the invention. By observing this figure, one can clearly understand the basic implementation method of a side channel vulnerability assessment method based on causal inference. Figure 2 This is a flowchart of the implementation steps in the embodiment of the invention. By observing this figure, one can clearly understand the specific implementation process of a side channel vulnerability assessment method based on causal inference. Figure 3 The results of the side channel structure relationship under the constructed data compression scenario. DETAILED DESCRIPTION

[0013] The following is a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are also within the scope of protection of the present invention.

[0014] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0015] like Figure 2 As shown, an embodiment of the present invention discloses a side channel vulnerability assessment method based on causal inference in a data compression scenario, comprising the following steps:

[0016] S1: Collect the data required for causal inference-based side channel vulnerability assessment in this scenario;

[0017] S2: By analyzing the data flow within the attack scenario and the mutual influence between the side channels, the side channel structure relationship within the scenario is constructed;

[0018] S3: Determine the adjustment set for each side channel;

[0019] S4: Calculate the unbiased impact of each side channel on system security;

[0020] S5: Measuring the vulnerability of each side channel based on unbiased influence.

[0021] In the step S1, we collect the data required for the experiment. In order to quantitatively study the relationship between the compression ratio side channel, compression time side channel, decompression time side channel and system security in the data compression scenario, we need to collect relevant data including system security, compression ratio, compression time, and decompression time. We first randomly generated a file of 16KB as the secret data to be cracked. The file consists only of uppercase and lowercase letters. In the process of attacking and cracking the secret data, the attacker will speculate on the secret data to form speculated data. In the subsequent data collection and simulated attack process, as in previous studies, we assume that the attacker can compress the speculated data together with the secret data. We define the measurement indicators used in this scenario as follows:

[0022] System security: We use the degree of similarity between the inferred data and the secret data to characterize system security. The greater the similarity between the inferred data and the secret data, the worse the system security; conversely, the lower the similarity, the better the system security. Therefore, we construct inferred data with a degree of similarity η (0 ≤ η ≤ 0.9) to the generated secret data. When the degree of similarity between the constructed inferred data and the secret data is η, the proportion of the number of characters in the inferred data and the secret data to the total number of characters is η, and the system security at this point is recorded as 1-η;

[0023] Compression ratio: compress each inferred data together with the secret data, and record the ratio of the compressed file size to the original file size;

[0024] Compression time: Due to the jitter in compression time, the guessed data and the secret data are compressed together for 10 5 times and record 10 5 The average time taken for compression;

[0025] Decompression time: Due to the jitter in decompression time, each guessed data is compressed together with the secret data, and the single compressed data is repeated 10 times. 5 Decompress and record 10 5 The average time it takes to decompress the files.

[0026] In step S2, the side channel structure relationship within the data compression scenario is constructed by analyzing the data flow within the data compression scenario and the mutual influence between the side channels. The analysis shows that there are compression ratio side channels, compression time side channels and decompression time side channels in the scenario. The structural relationship construction result is as follows: Figure 3 shown.

[0027] In step S3, the respective adjustment sets are selected for the compression ratio side channel, the compression time side channel and the decompression time side channel. Specifically, for the side channel C iIf node A is not a side channel C in the established side channel relationship i Descendant nodes of the node, while node A can block all side channels C i Node and system security T node points to C i path, then node A will be selected as side channel C i Thus, the adjustment set of the compression ratio side channel is empty, the adjustment set of the compression time side channel is the compression ratio side channel, and the adjustment set of the decompression time side channel is the compression ratio side channel and the compression time side channel.

[0028] In step S4, the unbiased impact of the compression ratio side channel, the compression time side channel, and the decompression time side channel on the system security is calculated respectively. The specific calculation method is as follows: for a specific side channel C i , the selected adjustment set is recorded as Then the specific side channel C i The unbiased impact on system security T satisfies Among them, a i express The possible specific values ​​of each element.

[0029] In step S5: further, calculate the specific side channel C i Unbiased impact on system security T. For side channel C under specific attack scenarios i , its vulnerability D(C i )for: in Where P(T=t|do(C i =c)) is the output of the side channel unbiased impact estimation module, indicating C i The unbiased causal effect between and T, and Indicates that C is taken into consideration i The impact of all possible states on system security. The calculated compression ratio side channel, compression time side channel, and decompression time side channel vulnerabilities are shown in the following table: Side Channel Compression ratio side channel Compressed time side channel Decompression time side channel Vulnerability 2.303 0.698 0.329

Claims

1. A side channel vulnerability assessment method based on causal inference, characterized in that: The steps include: Side channel relationship building module: This module analyzes the data flow within a specific attack scenario and the mutual influence between side channels to construct the structural relationship of the side channels within the system. Side channel unbiased impact estimation module: calculate the specific side channel C i Unbiased effect on system safety T; Side channel vulnerability calculation module: Based on the obtained unbiased effect of the side channel on system security, the vulnerability of the side channel in the system is quantified.

2. A side channel vulnerability assessment method based on causal inference according to claim 1, characterized in that: In the side channel relationship construction module, nodes are used to represent components including various side channels and system security indicators, and edges are used to represent the information flow relationship between nodes.

3. A side channel vulnerability assessment method based on causal inference according to claim 1, characterized in that: The side channel unbiased impact estimation module includes the following steps: Adjustment set selection module: used to identify the confounding factors that affect the side channel C i and the node of system security T, if node A is not a side channel C in the established side channel relationship i Descendant nodes of the node, while node A can block all side channels C i Node and system security T node points to C i path, then node A will be selected as side channel C i Elements in the adjustment set; Unbiased impact calculation module: For a specific side channel C i , the selected adjustment set is recorded as Then the specific side channel C i The unbiased impact on system security T satisfies: Among them, a i express The possible specific values ​​of each element.

4. The side channel vulnerability assessment method based on causal inference according to claim 1 is characterized in that: The side channel unbiased impact estimation module needs to be calculated based on the output results of the side channel unbiased impact estimation module. For the side channel C under a specific attack scenario, i , its vulnerability D(C i )for: in, Where P(T=t|do(C i =c)) is the output of the side channel unbiased impact estimation module, indicating C i The unbiased causal effect between and T, and Indicates that C is taken into consideration i The impact of all possible states on system security.

Images