Networked Real Time and Embedded Systems Laboratory

Department of Computer Science

The University of Illinois at Urbana Champaign

Adaptive Management

Feasible Region Calculus

Real-time Xen

FPGAs

 

 

Real-Time Resource Management

 

The Networked Real Time Embedded Systems Laboratory at UIUC has a history of great accomplishments in the field of real time computing. All these works have driven major changes in the way real-time systems have been built and analyzed, transforming real-time computing practice from an ad-hoc process to engineering practice based on analytic methods.

Nowadays, real-time resource management technology is experiencing new formidable challenges due to a broader applicability of real-time and embedded systems to futuristic applications (including nationwide medical device and health management networks, worldwide web of wired and wireless sensor networks, real-time transportation networks, tele-presence, etc.) and due to advancements in hardware technology (communication and computing devices). From a resource management point of view, the key challenges include:  

  • Unpredictable behaviors of COTS-based real-time systems: the current generation of real-time resource virtualization, including the current version of avionics standard ARINC 653, is insufficient for providing the required level of temporal protection for safety-critical applications. Hidden channels introduce dependencies across different temporal partitions, not accounted for in current models, thereby invalidating the temporal isolation property. To provide true temporal partitioning, enforceable specifications must address the complex dependencies among all interacting resources.
  • Lack of robustness and temporal QoS in wireless embedded systems: real-time wireless theory is still at an early stage; in fact, strong assumptions are often made in terms of network topology, operating environments, and channel quality. Temporal predictability cannot be achieved except under ideal network conditions. Lack of robustness is another serious concern especially in large scale deployments.
  • Complexity and lack of models: real-time cyber-physical systems are becoming larger and more complex. Assuming the availability of exact task and resource models for worst-case behavior is no longer practical. Instead, a theory has to be established to account for uncertainty and lack of precise knowledge. Composability of temporal (and functional) behavior of systems emerges as a great challenge. Namely, it becomes important to design systems in which the behavior of the whole can be accurately predicted from the composition of component behaviors. 

Research Areas

  • Hardware/software co-design: we aim at developing the scientific foundations for software/hardware co-design practices, which will allow us to analyze and verify the specified temporal and performance requirements of a cyber-physical system architecture before deployment time. For example, we are investigating how FPGA technology can be exploited for reducing hardware unpredictability.
  • Real-time wireless infrastructures: we aim at devising robust and temporally predictable wireless infrastructures. Robustness and quality of service of the proposed wireless-enabled architectures should be validated and enforced in large scale deployments. In addition, safety critical systems should be built in such a way that core components are formally verifiable, and use (but do not depend on) wireless links.
  • Feasible Region Calculus: we develop a theory for composition of temporal behavior of large systems from the behavior of their components. The calculus includes operators that compute feasible regions of composed systems from those of their subsystems. The feasible regions describe system states (such as constraints on resource utilization) for which timing constraints are met.