$\text{ACAC}_{\text{ABCD}}$ Model: Implementation and Comparative Performance Study

The rapid advancements in networking, communication, automation engineering, and artificial intelligence (AI) are shaping society into a highly interconnected and intelligent ecosystem. In smart environments such as smart farming and manufacturing, devices autonomously perform long-lived operations with minimal human intervention, adapting dynamically to system requirements and environmental conditions. To regulate these long-lived operations, referred to as activities (e.g., spraying water, plowing field), an Activity-Centric Access Control (ACAC) model has been proposed. While previous works have incrementally introduced ACAC's decision parameters Authorizations (A), Obligations (B), Conditions (C), and Dependencies (D) and presented policy specifications and proof-of-concept implementation for dependencies (D), a comprehensive implementation of the consolidated <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{ACAC}_{\text{ABCD}}$</tex> model remains unexplored. Validating ACAC's feasibility in highly dynamic environments is essential, particularly where security and safety must be ensured to protect both device operations and the system from potential insider and external threats. Additionally, the absence of constraints across various life-cycle stages undermines the model's comprehensiveness. In this paper, we take a significant step toward implementing the consolidated <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{ACAC}_{\text{ABCD}}$</tex> model in a smart farming use case, incorporating additional constraints such as usage count, cardinality, dynamic separation of duties, and activity duration. We first provided the formal model definitions for authorizations (A), obligations (B) and conditions (C), referred to as <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$A C A C_{A}, A C A C_{B}$</tex>, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$A C A C_{C}$</tex> respectively. Furthermore, to demonstrate proof-of-concept, we utilize AWS cloud infrastructure. The Activity Control Decision <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\left(\text{ACD}_{\text{ACAC}_{\text{ABCD}}}\right)$</tex> policy engine is implemented using AWS Lambda, with AWS S3 storing policies and real-time data. An AWS EC2 instance manages interactions with Lambda, processing requests that specify a source and activity. IoT devices are configured via AWS IoT Core Things to dynamically reflect activity state changes. We simulated the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{ACAC}_{\text{ABCD}}$</tex> model on AWS, conducted a performance analysis based on multiple metrics and compared ACAC's efficiency against the widely used Attribute-Based Access Control (ABAC) model. Our results demonstrate that the ACAC model is practical, scalable, and well-suited for dynamic smart systems where device operations are interconnected and continuously evolving.