Zum Hauptinhalt springen

LBAC

What is LBAC?​

LBAC (Lattice-Based Access Control) is a formal model that can enforce both confidentiality and integrity in one framework. This page shows one Casbin formulation: subject and object each have a confidentiality level and an integrity level; the matcher enforces read/write rules on both dimensions.

Model​

[request_definition]
r = sub, subject_confidentiality, subject_integrity, obj, object_confidentiality, object_integrity, act

[policy_definition]
p = sub, obj, act

[role_definition]
g = _, _

[policy_effect]
e = some(where (p.eft == allow))

[matchers]
m = (r.act == "read" && r.subject_confidentiality >= r.object_confidentiality && r.subject_integrity >= r.object_integrity) || (r.act == "write" && r.subject_confidentiality <= r.object_confidentiality && r.subject_integrity <= r.object_integrity)

Read allowed when subject’s confidentiality and integrity are both ≥ object’s; write allowed when both are ≤ object’s. So confidentiality follows a BLP-like rule and integrity a Biba-like rule, in a single matcher.

Example Requests​

The following examples demonstrate how this LBAC implementation evaluates requests:

# Normal read operations (ALLOWED)
admin, 5, 5, file_topsecret, 3, 3, read      # admin (conf:5, int:5) reads file_topsecret (conf:3, int:3) - ALLOWED
manager, 4, 4, file_secret, 4, 2, read       # manager (conf:4, int:4) reads file_secret (conf:4, int:2) - ALLOWED
staff, 3, 3, file_internal, 2, 3, read       # staff (conf:3, int:3) reads file_internal (conf:2, int:3) - ALLOWED
guest, 2, 2, file_public, 2, 2, read         # guest (conf:2, int:2) reads file_public (conf:2, int:2) - ALLOWED

# Read operation violations (DENIED)
staff, 3, 3, file_secret, 4, 2, read         # staff (conf:3, int:3) reads file_secret (conf:4, int:2) - DENIED (conf < obj.conf)
manager, 4, 4, file_sensitive, 3, 5, read    # manager (conf:4, int:4) reads file_sensitive (conf:3, int:5) - DENIED (int < obj.int)
guest, 2, 2, file_internal, 3, 1, read       # guest (conf:2, int:2) reads file_internal (conf:3, int:1) - DENIED (conf < obj.conf)
staff, 3, 3, file_protected, 1, 4, read      # staff (conf:3, int:3) reads file_protected (conf:1, int:4) - DENIED (int < obj.int)

# Normal write operations (ALLOWED)
guest, 2, 2, file_public, 2, 2, write        # guest (conf:2, int:2) writes file_public (conf:2, int:2) - ALLOWED
staff, 3, 3, file_internal, 5, 4, write      # staff (conf:3, int:3) writes file_internal (conf:5, int:4) - ALLOWED
manager, 4, 4, file_secret, 4, 5, write      # manager (conf:4, int:4) writes file_secret (conf:4, int:5) - ALLOWED
admin, 5, 5, file_archive, 5, 5, write       # admin (conf:5, int:5) writes file_archive (conf:5, int:5) - ALLOWED

# Write operation violations (DENIED)
manager, 4, 4, file_internal, 3, 5, write    # manager (conf:4, int:4) writes file_internal (conf:3, int:5) - DENIED (conf > obj.conf)
staff, 3, 3, file_public, 2, 2, write        # staff (conf:3, int:3) writes file_public (conf:2, int:2) - DENIED (both > obj)
admin, 5, 5, file_secret, 5, 4, write        # admin (conf:5, int:5) writes file_secret (conf:5, int:4) - DENIED (int > obj.int)
guest, 2, 2, file_private, 1, 3, write       # guest (conf:2, int:2) writes file_private (conf:1, int:3) - DENIED (conf > obj.conf)

Security Levels​

In this implementation, both confidentiality and integrity are represented as integers where higher values indicate higher security levels:

Confidentiality Levels​

  • Level 1: Public/Unclassified
  • Level 2: Confidential
  • Level 3: Secret
  • Level 4: Top Secret

Integrity Levels​

  • Level 1: Low integrity (e.g., public data, user-generated content)
  • Level 2: Medium integrity (e.g., verified data, trusted sources)
  • Level 3: High integrity (e.g., system data, administrative content)
  • Level 4: Critical integrity (e.g., security policies, system configuration)

Applicable Scenarios​

This LBAC implementation is well-suited for:

  • Multi-level security environments
  • Applications where both data protection and data accuracy matter

Implementation Notes​

  • The model implements mandatory access control (MAC) with dual security properties
  • System administrators assign security levels
  • Access decisions consider both confidentiality and integrity levels
  • The model prevents information leakage and data corruption