1 / 33

Resource Management for Real-Time Environments

Resource Management for Real-Time Environments. Instructor: Dr. Subra Ganesan Presented by: Pooja Mehta Date: 10/16/06. Presentation outline. Motivation Problem illustrations of Radar systems Basic Radar model Tasks with Harmonic Periods Offline Template Generation

justine-roy
Télécharger la présentation

Resource Management for Real-Time Environments

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Resource Management for Real-Time Environments Instructor: Dr. Subra Ganesan Presented by: Pooja Mehta Date: 10/16/06

  2. Presentation outline • Motivation • Problem illustrations of Radar systems • Basic Radar model • Tasks with Harmonic Periods • Offline Template Generation • Schedule construction on Hyperperiod • Some Proposed Solutions • Feasible Intervals • Online Template Generation • Finite Horizon Scheduling • Conclusions

  3. TASK 1 0 T1 2T1 3T1 TASK 2 0 T2 2T2 3T2 4T2 Periodic tasks Known periods Known execution times Known deadlines Motivation • The traditional notion of real-time systems • However, many important applications lack this simple structure • Complexity arises because of • Stringent task requirements • Scale of systems

  4. Presentation outline • Motivation • Problem illustrations of Radar systems • Basic Radar model • Tasks with Harmonic Periods • Offline Template Generation • Schedule construction on Hyperperiod • Some Proposed Solutions • Feasible Intervals • Online Template Generation • Finite Horizon Scheduling • Conclusions

  5. Basic Radar Model Ai : Transmit Power txi : Transmit pulse width twi: Wait time tri : Receive time Radar System Model

  6. End-to-end deadline FILTERING CLASSIFICATION COMMAND GENERATION Execution requirements on each node Processing requirements for radar tasks • Signals received at the antenna need to be processed (backend computations) • At multiple stages • Within an end-to-end deadline

  7. Radar dwell scheduling (N+1)th job Nth job Last illumination time Illumination window Processing window Temporal distance

  8. Radar dwell Dwell packing Power (kw) P(t) t Radar dwell scheduling Constraints on power Non-preemptible Reusable Question: How do we schedule many such tasks?

  9. Template-based Schedule

  10. Q-RAM & Scheduler Admission Control • Reduce the resource utilization bounds • Changes at irregular intervals

  11. Offline Template Generation

  12. Offline Template Generation • task types were restricted to a finite set • appropriate templates were chosen during online operation • Resource managers could only pick task types from the finite set.

  13. Presentation outline • Motivation • Problem illustrations of Radar systems • Basic Radar model • Tasks with Harmonic Periods • Offline Template Generation • Schedule construction on Hyperperiod • Some Proposed Solutions • Feasible Intervals • Online Template Generation • Finite Horizon Scheduling • Conclusions

  14. Dynamic Q-RAM Optimization

  15. Online Template Generation Arbitrary tasks can be interleaved or nested on-the-fly.

  16. Online Template Generation • arbitrary task types can be combined on-the-fly to produce a template; • provides greater freedom to a resource manager. • The resource manager can tune the parameters of each task with finer granularity. • Online template generation is carried out using a fast heuristic based on task characteristics.

  17. Resource management framework

  18. Dwell packing Radar dwell scheduling – issues Temporal distance constraints Constraints on power Non-preemptible

  19. Feasible intervals Synthetic period Temporal distance Fixed length templates for packing dwells Heuristics for building templates Template length divides the smallest period Dwell scheduling – solutions

  20. Modular Schedule Updates Without modular schedule update With modular schedule update

  21. Constraints • Temporal Constraints When new tasks are admitted, the schedule changes only within the templates in which new jobs are inserted. • Energy Constraints Since a job is inserted into a template only if it will not cause the energy level to exceed ETH, and since job insertions assume that the energy level at the start of a template is ETH, job insertions are guaranteed to be safe in terms of the energy constraint.

  22. Cool-down duration for Dwell A Cool-down duration for Dwell B Dealing with the energy constraint • Cooldown time ETH L

  23. horizon Finite horizon scheduling Task B arrives; is rejected A A A A A T+H T Task A departs Task B need not have been rejected Feasible intervals for Task B

  24. Scheduling overhead

  25. Reduced task rejection rates

  26. Utilization improvement Maximum achievable with energy bound

  27. Presentation outline • Motivation • Problem illustrations of Radar systems • Basic Radar model • Tasks with Harmonic Periods • Offline Template Generation • Schedule construction on Hyperperiod • Some Proposed Solutions • Feasible Intervals • Online Template Generation • Finite Horizon Scheduling • Conclusions

  28. Conclusions • All Real time systems doesn’t follow Ideal model • Determination of Schedulability Regions • Knowing the Schedule not just the schedulability • Systems should be able to handle unseen tasks, without violating the Temporal and Energy constraints

  29. References [1] C.-S. Shih, S. Gopalakrishnan, P. Ganti, M. Caccamo, L. Sha: “Template-based real-time dwell scheduling with energy constraint,” IEEE Real-Time Technology and Applications Symposium, Washington D.C., USA, May 2003. [2] C.-S. Shih, S. Gopalakrishnan, P. Ganti, M. Caccamo, L.Sha: “Scheduling real-time dwells using tasks withsynthetic periods,” IEEE Real-Time Systems Symposium, Cancun, Mexico, December 2003. [3] C.-G. Lee, P.-S. Kang, C.-S. Shih, L. Sha: “Radar dwell scheduling considering physical characteristics of phased array antenna,” IEEE Real-Time Systems Symposium,Cancun, Mexico, December 2003. [4] J. Hansen, S. Ghosh, R. Rajkumar, J. Lehoczky: “Resource management of highly configurable tasks,” Workshop on Parallel and Distributed Real-Time Systems, Santa Fe, USA, April 2004.

  30. References Contd.. [5] MURI on QoS in Surveillance and Control Radar Dwell Scheduling for Phased-Array Radars PIs Lui Sha Marco Caccamo Chang-Gun Lee [6] GOPALAKRISHNAN, S. Resource Management for Real-Time Environments. PhD thesis, University of Illinois, Urbana, Illinois, Dec. 2005. [7] GOPALAKRISHNAN, S., CACCAMO, M., SHIH, C.-S., SHA, L., AND LEE, C.-G. Finite horizon scheduling of radar dwells with online template construction. Real-Time Systems (2006).

  31. Thank you !!!!

  32. Questions and Answers

More Related