Cryptographic Keywords in NVD: Statistics and Visualization
Martin Stanek · 2024
Read Cryptographic Keywords in NVD: Statistics and Visualization by Martin Stanek (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-01942.txt.
Introduction
Martin Stanek presents a preliminary look at cryptographic keywords inside the National Vulnerability Database, the public catalog of software security flaws. The short paper gathers basic statistics on the crypto-related terms that appear across published vulnerability records and shows the results through simple visualizations. It is framed as a first attempt, scoping out what keyword-based views can reveal about the place of cryptography in reported flaws.
Catalog record
| CREATOR | Martin Stanek |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 11 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-01942
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Cryptographic Keywords in NVD: Statistics and Visualization Martin Stanek Abstract A preliminary attempt to use cryptographic keywords and analyze vulnerabilities published in the National Vulnerability Database is presented. Basic statistics and visualizations are included. Keywords: Cryptography, Vulnerability, NVD 1 Introduction Cryptography plays an important role in securing information systems, networks, and other IT services. Similarly to other security controls, the cryptographic controls can fail when ignored or improperly designed, implemented, or operated. The OWASP lists Cryptographic Failures as the second most critical security risk to web applications in their 2021 edition of OWASP Top 10 [ 4 ] . Even though cryptographic vulnerabilities are less frequent than SQL Injections, Cross-site scripting, or Out-of-bound Writes, they still can be potentially exploited with severe consequences. Cryptographic failures were analyzed in general setting using published vulnerabilities [ 2 ] , as well as empirically by analyzing misuses of cryptographic APIs in existing code bases, such as applications written in Java [ 1 ] , C/C++ [ 6 ] , or Python [ 5 ] . The National Vulnerability Database (NVD) [ 3 ] , managed by the National Institute of Standards and Technology (NIST), is the most respected repository of software vulnerabilities, uniquely identified by CVE 1 1 1 Common Vulnerabilities and Exposures ( https://cve.mitre.org/ ) identifiers. Nowadays, it publishes a vast number of vulnerabilities, e.g., more than 8500 CVEs in the third quarter of 2024. The vulnerabilities are classified according to their underlying weakness, and one or multiple CWE 2 2 2 Common Weakness Enumeration ( https://cwe.mitre.org/ ). Note that the NVD uses a selected subset of CWEs. identifiers are assigned to each CVE. There are CWEs that are specifically linked to cryptographic issues, such as lack of cryptography in CWE-312 (Cleartext Storage of Sensitive Information) or CWE-319 (Cleartext Transmission of Sensitive Information), implementation issues in CWE-295 (Improper Certificate Validation), design problems in CWE-321 (Use of Hard-coded Cryptographic Key) etc. However, a significant number of CVEs are assigned special CWE identifiers NVD-CWE-noinfo for vulnerabilities with insufficient information to classify them, and NVD-CWE-Other for weaknesses outside of the NVD selection. For example, 15.1% of CVEs published in the third quarter of 2024 have one of these two assignments. Keyword analysis of CVE descriptions can provide an interesting and potentially useful insight into published CVEs. It is a supplementary analysis that does not rely on classifications already available in the NVD. Keywords The selection of keywords is subjective, and different sets can be used for other analyses. However, the chosen keywords aim to cover basic cryptographic terms, constructions, algorithms, protocols, and attacks. They are specifically selected to indicate with high probability a cryptographic issue or at least the use of cryptographic constructions. Table 1 shows the keywords used in this paper. encrypt plaintext salt elliptic decrypt cleartext password cryptanalysis cipher public key TLS cryptographic hash private key RSA side channel symmetric certificate AES man in the middle asymmetric key exchange ECDSA replay attack signing Diffie ECDH brute force signature random HMAC Table 1: Cryptographic keywords 2 The dataset and results The dataset consists of CVEs published between 2023-01-01 and 2024-09-30 (i.e. the entire year 2023 and three quarters of 2024), downloaded using NVD’s public API. The CVEs with status Rejected were filtered out. This results in 57\,773 CVEs in non-rejected state, with an average CVSS 3 3 3 Common Vulnerability Scoring System, version 3.1 base severity score 7.10 . Matching a keyword in the description is performed as a simple substring search, with some customization: case non-sensitive search, hyphens replaced by spaces, special treatment of some keywords 4 4 4 For example, RSA was treated separately to avoid matching with “path trave rsa l” issues. . Substring search was chosen to match encrypt with encrypt , encryption , encrypted , etc. There are 3914 CVEs (6.8%) containing at least one keyword in their description. The average base severity score of these CVEs is 7.18, slightly higher than the score for the entire dataset. 2.1 Keywords statistics Table 2 on page 2 displays for each keyword: • CVE – the count of CVE’s in which the description includes the keyword, referred to as related CVEs. Unsurprisingly, password is the most frequently mentioned term. Interestingly, some keywords with narrow semantics like man in the middle are mentioned more frequently than some generic terms like cipher or public key . • Unique CWE – the number of unique CWEs identified in the related CVEs. The values are relatively high for the majority of keywords. It indicates that even specific issues (e.g., man in the middle attacks) cannot be analyzed just by focusing on a few CWEs. • Avg Score – average base severity score of related CVEs. Red color indicates entries with scores higher than the dataset’s average score. Keyword CVE Unique CWE Avg Score password 1610 164 7.54 encrypt 530 127 6.73 hash 381 100 6.75 certificate 348 92 6.97 TLS 290 100 7.02 cryptographic 241 72 7.36 signature 215 79 7.14 decrypt 206 69 6.93 random 187 76 7.33 man in the middle 185 59 6.29 brute force 179 48 7.73 cleartext 164 33 6.50 plaintext 146 51 6.69 side channel 92 13 6.19 signing 88 54 7.22 cipher 79 39 6.51 private key 79 39 6.63 AES 65 39 6.28 public key 59 38 7.05 RSA 56 26 6.64 HMAC 33 30 7.62 salt 28 24 6.89 ECDSA 22 21 7.29 replay attack 22 9 7.40 symmetric 21 20 6.65 key exchange 17 12 7.04 ECDH 12 9 6.21 elliptic 12 11 7.83 asymmetric 6 8 7.07 Diffie 5 9 6.93 cryptanalysis 1 2 6.80 Table 2: Keywords statistics CWE statistics Table 3 on page 3 shows top 25 CWEs sorted by a keyword count. A CVE c is assigned to one or multiple CWEs w(c)=\{w_{1},\ldots,w_{c_{n}}\} , and its description contains a set of keywords k(c)=\{k_{1},\ldots,k_{c_{m}}\} . The keyword count for a weakness \mathbf{w} is defined as \text{kc}({\mathbf{w}})=\sum_{c;\;\mathbf{w}\in w(c)}|k(c)| , with the understanding that multiple occurrences of a keyword in the same description are counted as a single occurrence. Note that there is only a weak correlation between keyword count and individual CWE frequency. An alternative method, which is not used in this analysis, would involve scaling the keyword count by the number of CVEs: \text{kc}({\mathbf{w}})/|\{c;{\mathbf{w}}\in w(c)| . Table 3 includes the following columns: • Keywords – the keyword count associated with a particular CWE. • Avg Score – the average base severity score for the CVEs whose descriptions contain at least one keyword and are associated with the CWE. Entries in red color indicate scores higher than the the score in the last column. • Dataset – the average base severity score for all CVEs assigned to the CWE, calculated from the full dataset. CWE Title Keywords Avg Score Dataset NVD-CWE-noinfo 413 6.86 6.84 CWE-295 Improper Certificate Validation 298 6.97 7.02 CWE-798 Use of Hard-coded Credentials 235 8.43 8.46 CWE-203 Observable Discrepancy 225 6.11 6.00 CWE-200 Exposure of Sensitive Information to an Unauthorized Actor 224 6.56 6.36 CWE-312 Cleartext Storage of Sensitive Information 212 6.27 6.25 CWE-327 Use of a Broken or Risky Cryptographic Algorithm 180 7.33 7.28 CWE-287 Improper Authentication 156 7.62 7.89 CWE-89 Improper Neutralization of Special Elements used in an