1 / 17

Quality of Service for In-Home Digital Networks PROGRESS PROJECT EES.5653

Quality of Service for In-Home Digital Networks PROGRESS PROJECT EES.5653. Terminal QoS Alina Weffers-Albu. Contents. Context Project definition – Goals, Approach Previous results, progress Results Future work. Network. Context - QoS in IN-Home Digital Networks.

ivory
Download Presentation

Quality of Service for In-Home Digital Networks PROGRESS PROJECT EES.5653

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. Quality of Service for In-Home Digital NetworksPROGRESS PROJECT EES.5653 Terminal QoS Alina Weffers-Albu Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  2. Contents • Context • Project definition – Goals, Approach • Previous results, progress • Results • Future work Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  3. Network Context - QoS in IN-Home Digital Networks Aim:provide guaranteed and optimised Quality of Service (QoS) for interconnected real-time embedded systems. Network QoS: Dealing with Artefacts Adaptiveness • Terminal QoS: • Performance • RU CPU,mem Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  4. Get Full Packet Put Full Packet …   Full Queue Full Queue Physical Platform Empty Queue Empty Queue   Put Empty Packet Get Empty Packet Component Processing code  FQ FQ FQ C2 CN … C1 EQ EQ EQ Context - Description of Analyzed Systems Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  5. Context - Description of Analyzed Systems Components • Data driven. Execution determined by: • Availability of necessary input • Priority of component task • Time driven. Execution determined by: • Availability of necessary input. (Or NOT) • Priority • Periodicity. Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  6. Context - Description of Analyzed Systems Components Both types. Execution determined by: • Execution scenario of component (fixed/variable) • Computation time. • Nature of input stream. • Suspension time (if task with execution deferral due to cooperation with hardware). Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  7. Goals Performance System Predictability & Optimization Goals 1. Prediction of resource use through a set of performance quality parameters for a given system. 2. Control performance quality parameters - find good practices of design for the system. Approach Study and model the dynamic behavior of a given system => prediction & control of performance quality parameters Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  8. Performance Quality Parameters Performance Quality Parameters. Buffer size Packet size Activation Times Priority setting Resource Utilization (RU) for CPU, memory, bus – feasibility check on the physical platform at hand. Activation Times (AT) – cost of context switches (CS). Response Times (RT) – prediction/control of deadline misses. Number of Context Switches (NCS) – overhead induced by the composed execution of components. Required buffer space Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  9. Pattern Initialization Phase Stable Phase Finalization Phase Stable Phase Approach. Hypothesis Let C1, C2, C3, …, CN be a chain of components communicating through a set of queues. The execution of the chain, after an initialization phase adopts a repetitive pattern of execution. Conditions under which the above statement holds in progress was supposed to be explored.(slides 5,6) Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  10. Previous results • Expanded approach previously tested on particular case to a more general context - tests on other types of components, different priorities assignment. • Formulate “Stable Phase Theorem”, distinguished cases of interest for analysis, proof. => Approach for control and optimization of performance parameters by formulating corollaries deduced from the proof. • Analysis of first case - lemmas, corollaries. • Studied influence of input on the execution pattern of a streaming chain. • Defined goals and approach for PhD project. Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  11. Stable Phase Theorem.Cases of interest for analysis. C1, C2, C3, …, CNchain of components communicating through a set of queues: • N data-driven components (1-1). • C1 data-driven component with execution deferral (1-1),C2, C3, …, CNdata-driven components (1-1). • C1, C2, C3, …, CN-1data-driven components (1-1), CN time-driven component. • C1 data-driven component with execution deferral (1-1),C2, C3, …, CN-1data-driven components (1-1), CN time-driven component. • C1time-driven component,C2, C3, …, CNdata-driven components (1-1). • C1 time-driven component, C2, C3, …, CN-1data-driven components (1-1), CN time-driven component. • C1 data-driven component with execution deferral (1-1), C2 data-driven component (n-m), C3, …, CN-1data-driven components (1-1), CN time-driven component. Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  12. Progress • All cases afore-mentioned have been analyzed • technical note, full paper containing mathematical modeling / argumentation of the results in writing. • Results exposure: • Presented results for industry clients of MRM: • Ruud Derwig (Philips Semiconductors) • Presented results in other groups of Natlab, applicability in other projects: • S. Balakrishnan (CoolCat – Performance Analysis Based Power Management) • M. Bekooij (Hydra) • Liesbeth Steffens (Betsy) • Papers: • M.A. Weffers-Albu, J.J. Lukkien, P.D.V. v.d. Stok, "A Characterization of Streaming Applications Execution", Proceedings RM4NES 2005, Eindhoven • M.A. Weffers-Albu, J.J. Lukkien, P.D.V. v.d. Stok, "Towards A Characterization of Real-Time Streaming Systems", ECRTS WIP 2005, Palma de Mallorca • Posters: • M.A. Weffers-Albu, J.J. Lukkien, P.D.V. v.d. Stok, "Towards A Characterization of Real-Time Streaming Systems", ECRTS WIP 2005 • M.A. Weffers-Albu, J.J. Lukkien, P.D.V. v.d. Stok, "Towards A Characterization of Real-Time Streaming Systems", SEES Workshop 2005 Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  13. Data-driven vs. Time-driven • N data-driven components (1-1) • Stable phase theorem, corollaries: - driving component minimum in priority - shortening initialization phase - minimum necessary of memory - prediction, optimization of NCS, RT, AT • Chain including time-driven component(s): • Stable phase theorem, corollaries – driving component(s) time-driven component(s) regardless of the priority assignment to components in the chain. - prediction, optimization of NCS, RT Time beats priority Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  14. FQ FQ C3 C2 EQ EQ Video decoding chain. Data driven, Execution deferral(1-1) • Calculated parameterized pattern taking into account: • chain execution dependency on the input stream - variation in execution scenario of C2, • timing properties of some components w.r.t. • Periodicity (ex: video renderer) • Variable computation time (ex: video decoder) • Applicability of results in practice: • power saving – prediction of idle time at run time. • knowledge on when/how to change the quality level of decoding for a video decoder. Data driven (n-m) Data driven (1-1) Time driven (1-1) FQ FQ CN … C1 EQ EQ Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  15. m = 2 m = 7 m = 3 m = 5 Video decoding chain.Calculated patterns, validation. 1. C2“1-m” data-driven component, variable m. • E(CN), E(CN-1), …, E(C2, C1), (idle time until the end of 2T(CN)), (m-1) *{ E(CN), E(CN-1), …, E(C2), (idle time until the end of 2T(CN))} 2. C2“n-1” data-driven component, variable n. • E(CN), E(CN-1), …, n * E(C2, C1), (idle time until the end of 2T(CN)). n = 3 n = 2 n = 4 Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  16. Surveillance application. • C1 time-driven component, C2, C3, …, CN-1 data-driven components (1-1), CN time-driven component. • Previous approach: • introduce in the middle of the chain a long buffer that would split the chain into 2 sub-chains with different dependencies between tasks: in the first sub-chain all components would be driven by C1 while all components in the second sub-chain would be driven by CN. • Our result: • Pattern anyway – inherent to this type of chains – middle buffer not needed for predictability. • Optimization of performance parameters achievable by: • controlling the phasing between the activations of the two time driven components. • imposing at design time that sum of computation times of data-drive components smaller than period of time-driven components. • Advantage: • Memory use much diminished – no middle buffer is used. Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

  17. Future work. References. Work in progress: • Technical note - mathematical modeling of results • Full paper ECRTS 2006 to be submitted in December 2005. • Chains with branches Future work: • Multiple chains • Extend analysis to multi-processor platforms References: John T. H. Lan. Philips Research USA - TN-2001-027. Date of Issue: 2001. Scalable MPEG Decoding with Graceful Degradation, and Its Applications and Implementation on TriMedia. Sharon Peng. Technical Report. Scalable MPEG2 Decoder with Graceful Degradation. Alina Weffers-Albu, m.a.albu@tue.nl TU/e Computer Science, System Architecture and Networking Philips Research Laboratories Eindhoven

More Related