Scheduling of Soft Realtime systems for contextaware Applications

Scheduling of Soft Real-time systems for context-aware Applications Presented by, Hemanth Amara 10/24/2021 CSE 666 - Real Time Computer Systems 1

Introduction: Soft real-time systems has been mainly motivated by applications from two domains : Databases and Multimedia. • Context-sensitive applications Mobile applications, Sensor networks, Pervasive and Ubiquitous Computing, Internet browsing, Video Games and virtual reality. • (Context can be defined as the set of environmental states and settings that either determine an application’s behavior or in which an application event occurs and is interesting to the user ) 10/24/2021 CSE 666 - Real Time Computer Systems 2

Statistical Distribution Function: A task is a single program which can be invoked and executed. E. g. single execution of jpeg encoder, audio player. • The statistical distribution function (SDF) for execution time provides information about the statistical likelihood of specific runtimes for a task. • • Dependency: Results of one task needed by another task. Weighted Likelihood relationship reflects the user’s likelihood to request the execution of a task after the completion of another task. • 10/24/2021 CSE 666 - Real Time Computer Systems 3

Benchmarks: e. g. SDFs of Location Discovery, g 721 encode tasks. 10/24/2021 CSE 666 - Real Time Computer Systems 4

Statistical. SRTS: • Non-parametric statistical techniques are used to model the runtime of each task. • The resulting output is the statistical distribution function (SDF) of execution times for each task. • Analysis of each of the tasks on different simulation platforms is performed in order to gather statistical runtime data for each of the tasks. 10/24/2021 CSE 666 - Real Time Computer Systems 5

Global Flow for Scheduling SSRTS: 10/24/2021 CSE 666 - Real Time Computer Systems 6

Statistical SSRTS: • Once the user profile and task profile have been established the user can begin to use the system. • When the user invokes a task, one of two situations may occur: 1)The scheduler had pre-fetched the invoked task and the data is available. 2)If the scheduler had not pre-fetched the task, the task may still be finished prior to the deadline placed according to user's profile. User is benefited in either case. 10/24/2021 CSE 666 - Real Time Computer Systems 7

Task Model: • The purpose of the task model is to provide a compact and accurate representation of soft realtime tasks with respect to their execution time. • Consists of three hierarchical layers: Window of consideration, Intertask and Intratask. 10/24/2021 CSE 666 - Real Time Computer Systems 8

Algorithm: 10/24/2021 CSE 666 - Real Time Computer Systems 9

Conclusions: • Uncertainties are addressed using statistical techniques. • The use of the SDF in the task model is more effective than task model with SDF replaced with average run-time or worst case run -time. • Non parametrical statistical techniques can be used for modeling the runtime of tasks. • Cumulative utilization. 10/24/2021 potential based scheduling can be used to maximize CSE 666 - Real Time Computer Systems 10

Questions ? “It is important that students bring a certain ragamuffin barefoot irreverence to their studies, they are here not to worship the known but to question it” Jacob Bronowski 10/24/2021 CSE 666 - Real Time Computer Systems 11

10/24/2021 CSE 666 - Real Time Computer Systems 12

10/24/2021 CSE 666 - Real Time Computer Systems 13

10/24/2021 CSE 666 - Real Time Computer Systems 14

10/24/2021 CSE 666 - Real Time Computer Systems 15

10/24/2021 CSE 666 - Real Time Computer Systems 16

10/24/2021 CSE 666 - Real Time Computer Systems 17

10/24/2021 CSE 666 - Real Time Computer Systems 18