Visible to the public Biblio

Filters: Author is Rubio-Medrano, Carlos E.  [Clear All Filters]
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 
Rubio-Medrano, Carlos E., Zhao, Ziming, Ahn, Gail-Joon.  2018.  RiskPol : A Risk Assessment Framework for Preventing Attribute-Forgery Attacks to ABAC Policies. Proceedings of the Third ACM Workshop on Attribute-Based Access Control. :54–60.

Recently, attribute-based access control (ABAC) has emerged as a convenient paradigm for specifying, enforcing and maintaining rich and flexible authorization policies, leveraging attributes originated from multiple sources, e.g., operative systems, software modules, remote services, etc. However, attackers may try to bypass ABAC policies by compromising such sources to forge the attributes they provide, e.g., by deliberately manipulating the data contained within those attributes at will, in an effort to gain unintended access to sensitive resources as a result. In such a context, performing a proper risk assessment of ABAC policies, taking into account their enlisted attributes as well as their corresponding sources, becomes highly convenient to overcome zero-day security incidents or vulnerabilities, before they can be later exploited by attackers. With this in mind, we introduce RiskPol, an automated risk assessment framework for ABAC policies based on dynamically combining previously-assigned trust scores for each attribute source, such that overall scores at the policy level can be later obtained and used as a reference for performing a risk assessment on each policy. In this paper, we detail the general intuition behind our approach, its current status, as well as our plans for future work.

Rubio-Medrano, Carlos E., Lamp, Josephine, Doupé, Adam, Zhao, Ziming, Ahn, Gail-Joon.  2017.  Mutated Policies: Towards Proactive Attribute-based Defenses for Access Control. Proceedings of the 2017 Workshop on Moving Target Defense. :39–49.
Recently, both academia and industry have recognized the need for leveraging real-time information for the purposes of specifying, enforcing and maintaining rich and flexible authorization policies. In such a context, security-related properties, a.k.a., attributes, have been recognized as a convenient abstraction for providing a well-defined representation of such information, allowing for them to be created and exchanged by different independently-run organizational domains for authorization purposes. However, attackers may attempt to compromise the way attributes are generated and communicated by recurring to hacking techniques, e.g., forgery, in an effort to bypass authorization policies and their corresponding enforcement mechanisms and gain unintended access to sensitive resources as a result. In this paper, we propose a novel technique that allows for enterprises to pro-actively collect attributes from the different entities involved in the access request process, e.g., users, subjects, protected resources, and running environments. After the collection, we aim to carefully select the attributes that uniquely identify the aforementioned entities, and randomly mutate the original access policies over time by adding additional policy rules constructed from the newly-identified attributes. This way, even when attackers are able to compromise the original attributes, our mutated policies may offer an additional layer of protection to deter ongoing and future attacks. We present the rationale and experimental results supporting our proposal, which provide evidence of its suitability for being deployed in practice.