Lee, Jaewoo
Email Address
ORCID
Disciplines
Search Results
Now showing 1 - 8 of 8
Publication Towards Compositional Mixed-Criticality Real-Time Scheduling in Open Systems(2015-12-01) Lee, Jaewoo; Chwa, Hoon Sung; Easwaran, Arvind; Lee, Insup; Shin, InsikAlthough many cyber-physical systems are both mixed-criticality system and compositional system, there are little work on intersection of mixed-criticality system and compositional system. We propose novel concepts for task-level criticality mode and reconsider temporal isolation in terms of compositional mixed-criticality scheduling.Publication Removing Abstraction Overhead in the Composition of Hierarchical Real-Time System(2011-04-01) Chen, Sanjian; Phan, Linh T.X.; Lee, Jaewoo; Lee, Insup; Sokolsky, OlegThe hierarchical real-time scheduling framework is a widely accepted model to facilitate the design and analysis of the increasingly complex real-time systems. Interface abstraction and composition are the key issues in the hierarchical scheduling framework analysis. Schedulability is essential to guarantee that the timing requirements of all components are satisfied. In order for the design to be resource efficient, the composition must be bandwidth optimal. Associativity is desirable for open systems in which components may be added or deleted at run time. Previous techniques on compositional scheduling are either not resource efficient in some aspects, or cannot achieve optimality and associativity at the same time. In this paper, several important properties regarding the periodic resource model are identified. Based on those properties, we propose a novel interface abstraction and composition framework which achieves schedulability, optimality, and associativity. Our approach eliminates abstraction overhead in the composition.Publication MC-Fluid: Fluid Model-Based Mixed-Criticality Scheduling on Multiprocessors(2014-12-01) Lee, Jaewoo; Phan, Kieu-My; Gu, Xiaozhe; Lee, Jiyeon; Easwaran, Arvind; Lee, Insup; Shin, InsikA mixed-criticality system consists of multiple components with different criticalities. While mixed-criticality scheduling has been extensively studied for the uniprocessor case, the problem of efficient scheduling for the multiprocessor case has largely remained open. We design a fluid model-based multiprocessor mixed-criticality scheduling algorithm, called MC-Fluid in which each task is executed in proportion to its criticality-dependent rate. We propose an exact schedulability condition for MC-Fluid and an optimal assignment algorithm for criticality-dependent execution rates with polynomial-time complexity. Since MC-Fluid cannot be implemented directly on real hardware platforms, we propose another scheduling algorithm, called MC-DP-Fair, which can be implemented while preserving the same schedulability properties as MC-Fluid. We show that MC-Fluid has a speedup factor of (1 + √ 5) /2 (~ 1.618), which is best known in multiprocessor MC scheduling, and simulation results show that MC-DP-Fair outperforms all existing algorithms.Publication Improving Resource Utilization for Compositional Scheduling Using DPRM Interfaces(2010-11-30) Lee, Jaewoo; Phan, Linh T.X.; Chen, Sanjian; Sokolsky, Oleg; Lee, InsupThe paper revisits the generation of interfaces for compositional real-time scheduling. Following an established line of research, we use periodic resource models in component interfaces to describe resource demand of the component. We identify a deficiency of existing interface generation algorithms that may require parameters of the resource model to be infeasibly small. We propose a new algorithm for interface generation that avoids this deficiency. We further demonstrate that resource utilization can be improved by using dual-periodic resource model (DPRM) interfaces that employ two periodic resource models to characterize the resource demand more precisely.Publication Overhead-Aware Compositional Analysis of Real-Time Systems(2013-04-01) Phan, Linh T.X.; Xu, Meng; Lee, Jaewoo; Lee, Insup; Sokolsky, OlegOver the past decade, interface-based compositional schedulability analysis has emerged as an effective method for guaranteeing real-time properties in complex systems. Several interfaces and interface computation methods have been developed, and they offer a range of tradeoffs between the complexity and the accuracy of the analysis. However, none of the existing methods consider platform overheads in the component interfaces. As a result, although the analysis results are sound in theory, the systems may violate their timing constraints when running on realistic platforms. This is due to various overheads, such as task release delays, interrupts, cache effects, and context switches. Simple solutions, such as increasing the interface budget or the tasks’ worst-case execution times by a fixed amount, are either unsafe (because of the overhead accumulation problem) or they waste a lot of resources. In this paper, we present an overhead-aware compositional analysis technique that can account for platform overheads in the representation and computation of component interfaces. Our technique extends previous overhead accounting methods, but it additionally addresses the new challenges that are specific to the compositional scheduling setting. To demonstrate that our technique is practical, we report results from an extensive evaluation on a realistic platform.Publication Realizing Compositional Scheduling Through Virtualization(2012-04-01) Lee, Jaewoo; Xi, Sisu; Chen, Sanjian; Phan, Linh T.X.; Gill, Chris; Lee, Insup; Lu, Chenyang; Sokolsky, OlegWe present a co-designed scheduling framework and platform architecture that together support compositional scheduling of real-time systems. The architecture is built on the Xen virtualization platform, and relies on compositional scheduling theory that uses periodic resource models as component interfaces.We implement resource models as periodic servers and consider enhancements to periodic server design that significantly improve response times of tasks and resource utilization in the system while preserving theoretical schedulability results. We present an extensive evaluation of our implementation using workloads from an avionics case study as well as synthetic ones.Publication Resource-Efficient Scheduling Of Multiprocessor Mixed-Criticality Real-Time Systems(2017-01-01) Lee, JaewooTiming guarantee is critical to ensure the correctness of embedded software systems that interact with the physical environment. As modern embedded real-time systems evolves, they face three challenges: resource constraints, mixed-criticality, and multiprocessors. This dissertation focuses on resource-efficient scheduling techniques for mixed-criticality systems on multiprocessor platforms. While Mixed-Criticality (MC) scheduling has been extensively studied on uniprocessor plat- forms, the problem on multiprocessor platforms has been largely open. Multiprocessor al- gorithms are broadly classified into two categories: global and partitioned. Global schedul- ing approaches use a global run-queue and migrate tasks among processors for improved schedulability. Partitioned scheduling approaches use per processor run-queues and can reduce preemption/migration overheads in real implementation. Existing global scheduling schemes for MC systems have suffered from low schedulability. Our goal in the first work is to improve the schedulability of MC scheduling algorithms. Inspired by the fluid scheduling model in a regular (non-MC) domain, we have developed the MC-Fluid scheduling algo- rithm that executes a task with criticality-dependent rates. We have evaluated MC-Fluid in terms of the processor speedup factor: MC-Fluid is a multiprocessor MC scheduling algo- rithm with a speed factor of 4/3, which is known to be optimal. In other words, MC-Fluid can schedule any feasible mixed-criticality task system if each processor is sped up by a factor of 4/3. Although MC-Fluid is speedup-optimal, it is not directly implementable on multiprocessor platforms of real processors due to the fractional processor assumption where multiple task can be executed on one processor at the same time. In the second work, we have considered the characteristic of a real processor (executing only one task at a time) and have developed the MC-Discrete scheduling algorithm for regular (non-fluid) scheduling platforms. We have shown that MC-Discrete is also speedup-optimal. While our previous two works consider global scheduling approaches, our last work con- siders partitioned scheduling approaches, which are widely used in practice because of low implementation overheads. In addition to partitioned scheduling, the work consid- ers the limitation of conventional MC scheduling algorithms that drops all low-criticality tasks when violating a certain threshold of actual execution times. In practice, the system designer wants to execute the tasks as much as possible. To address the issue, we have de- veloped the MC-ADAPT scheduling framework under uniprocessor platforms to drop as few low-criticality tasks as possible. Extending the framework with partitioned multiprocessor platforms, we further reduce the dropping of low-criticality tasks by allowing migration of low-criticality tasks at the moment of a criticality switch. We have evaluated the quality of task dropping solution in terms of speedup factor. In existing work, the speedup factor has been used to evaluate MC scheduling algorithms in terms of schedulability under the worst-case scheduling scenario. In this work, we apply the speedup factor to evaluate MC scheduling algorithms in terms of the quality of their task dropping solution under various MC scheduling scenarios. We have derived that MC-ADAPT has a speedup factor of 1.618 for task dropping solution.Publication CARTS: A Tool for Compositional Analysis of Real-Time Systems(2010-11-01) Phan, Linh T.X.; Lee, Jaewoo; Easwaran, Arvind; Ramaswamy, Vinay; Chen, Sanjian; Lee, Insup; Sokolsky, OlegThis paper demonstrates CARTS, a compositional analysis tool for real-time systems. We presented an overview of the underlying theoretical foundation and the architecture design of the tool. CARTS is open source and available for free download at http://rtg.cis.upenn.edu/carts/.