Configurable FPGA Access Control via Initiator and Target Bridges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Integrated circuit devices face security vulnerabilities due to shared access by multiple tenants to resources, leading to trust boundary violations and privilege challenges, especially in cloud-computing implementations where complex access control is resource-intensive and requires frequent adjustments.
Innovation Solution
The implementation of initiator bridges and target bridges within the integrated circuit device, which communicate via a network on chip (NoC), determine and manage access control by generating agent identifiers based on security attributes and user-defined instructions, allowing for dynamic adjustment of trust boundaries and access permissions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple tenants share access to IC device resources, then resource utilization and productivity are improved, but security vulnerabilities and trust boundary violations increase
Solution Approach 1:
The patent segments the IC device into multiple isolated tenant regions, each with its own trusted execution environment. This segmentation allows multiple tenants to share the physical hardware while maintaining logical isolation through separate memory spaces, register files, and configuration areas, thus enabling high resource utilization without compromising security.
Solution Approach 2:
The patent introduces a security manager as an intermediary component that mediates all access requests between tenants and shared resources. The security manager enforces access control policies, validates tenant identities, and manages cryptographic keys, acting as a trusted mediator that enables secure multi-tenant access without requiring direct tenant-to-resource connections.
2Reliability
If complex access control is implemented to enhance security, then security is improved, but device complexity and resource consumption increase
Solution Approach 1:
The patent implements a universal security manager that handles multiple security functions including authentication, authorization, key management, and access control policy enforcement. This single multi-functional component replaces what would otherwise require multiple separate security mechanisms, reducing overall device complexity while maintaining comprehensive security.
Solution Approach 2:
The patent uses configurable security parameters and policies that can be adjusted without changing the underlying hardware architecture. Access control rules, tenant permissions, and security levels are implemented as programmable parameters that can be modified through software configuration, allowing flexible security management without increasing physical device complexity.
3Adaptability or versatility
If access control instructions are frequently adjusted to meet changing security requirements, then adaptability is improved, but loss of time and reconfiguration overhead increase
Solution Approach 1:
The patent implements dynamic access control where security policies and tenant permissions can be modified in real-time without requiring physical reconfiguration or device shutdown. The security manager allows runtime updates to access control instructions, enabling the system to adapt to changing security requirements while maintaining continuous operation and avoiding reconfiguration delays.
Data Source
AI summary
Systems or methods of the present disclosure may provide systems and techniques for controlling access to components and resources of an IC device by multiple tenants. For example, a method may include: receiving access control instructions defining a first mapping between one or more tenants and respective security attributes and a second mapping between one or more agent identifiers and respective access permissions; receiving a communication intended for a target component of the IC device; determining an origin tenant from which the communication originated; determining a security attribute associated with the origin tenant based on the first mapping; and sending the communication and an agent identifier comprising the security attribute and an initiator bridge identifier to a corresponding target bridge, wherein the corresponding target bridge is configured to grant or deny access of the communication to the target component based on the agent identifier and the second mapping.