TY - GEN
T1 - Performance Analysis of Multi-core ABAC Systems Using an M/G/m Queue Model
AU - Chandra, Hunny
AU - Yelisetty, Karthikeya S.M.
AU - Madkaikar, Gaurav
AU - Sural, Shamik
AU - Vaidya, Jaideep
AU - Atluri, Vijayalakshmi
N1 - Publisher Copyright: © IFIP International Federation for Information Processing 2026.
PY - 2026
Y1 - 2026
N2 - Attribute-Based Access Control (ABAC) systems make access decisions based on a policy in conjunction with a set of attributes of users, resources and environmental conditions, rather than their individual identities. Large and dynamic ABAC systems, typically running on multi-core servers, must continually incorporate authorization updates through a process called policy mining while ensuring low latency in access resolution. We model this integrated access resolution and authorization update process as an M/G/m queue with server vacation, where access requests arrive into a queue served by m parallel cores, and policy mining periods carried out for authorization updates correspond to temporary unavailability of a subset of those cores. The framework identifies stable operating regimes and quantifies how multi-core operations can mitigate mining-induced delays. A client-server prototype has been developed to validate the model under realistic workloads. It simulates Poisson access requests arriving at dynamic multi-core ABAC engine with configurable incremental authorization update rates and vacation policies. Results show close agreement with the theoretical predictions of M/G/m queues, providing pathway towards designing high performance ABAC systems.
AB - Attribute-Based Access Control (ABAC) systems make access decisions based on a policy in conjunction with a set of attributes of users, resources and environmental conditions, rather than their individual identities. Large and dynamic ABAC systems, typically running on multi-core servers, must continually incorporate authorization updates through a process called policy mining while ensuring low latency in access resolution. We model this integrated access resolution and authorization update process as an M/G/m queue with server vacation, where access requests arrive into a queue served by m parallel cores, and policy mining periods carried out for authorization updates correspond to temporary unavailability of a subset of those cores. The framework identifies stable operating regimes and quantifies how multi-core operations can mitigate mining-induced delays. A client-server prototype has been developed to validate the model under realistic workloads. It simulates Poisson access requests arriving at dynamic multi-core ABAC engine with configurable incremental authorization update rates and vacation policies. Results show close agreement with the theoretical predictions of M/G/m queues, providing pathway towards designing high performance ABAC systems.
KW - ABAC
KW - M/G/m queue
KW - Vacation model
UR - https://www.scopus.com/pages/publications/105047027374
UR - https://www.scopus.com/pages/publications/105047027374#tab=citedBy
U2 - 10.1007/978-3-032-33260-8_8
DO - 10.1007/978-3-032-33260-8_8
M3 - Conference contribution
SN - 9783032332592
T3 - Lecture Notes in Computer Science
SP - 142
EP - 161
BT - Data and Applications Security and Privacy XL - 40th IFIP WG 11.3 Annual Conference, DBSec 2026, Proceedings
A2 - Palanisamy, Balaji
A2 - Samarati, Pierangela
PB - Springer Science and Business Media Deutschland GmbH
T2 - 40th IFIP WG 11.3 Annual Conference on Data and Applications Security and Privacy, DBSec 2026
Y2 - 28 July 2026 through 30 July 2026
ER -