Tool Mentor: TU Braunschweig Probabilistic Timing Analysis
Relationships
Main Description

Abstraction level

Implementation

Use Case Tasks

·         Specify Probabilistic Timing Properties / Analyze

Covered aspect

Probabilistic Timing

Algorithm

Typical-case analysis of sporadic overload

Inputs

1)    Based on a trace

1.    system structure: tasks and resources, not necessarily runnables, task chaining, resource mapping, scheduling policy SHOULD BE SPP and we need task priorities

2.    we derive execution times from the trace. Does anyone want to provide their own [BCET;WCET]?

3.    csv trace = activation, termination, preemption, start, resume etc. + indication about tasks and instances

4.    information about what should be approximated: which sources? For each approximated source, we need the approximate activation model.

2)    Based on a model

1.    system structure: tasks and resources, not necessarily runnables, task chaining, resource mapping, scheduling policy SHOULD BE SPP and we need task priorities

2.    for each task, an interval [BCET;WCET]; does anyone want to be more precise?

3.    for each source, an activation model

4.    information about which sources should be approximated. For each approximated source, we need the approximate activation model and the overload model.

Particular constraints on inputs

The scheduling policy should be SPP.

For 1) the approximate model should be periodic (possibly with jitter).

Preparation of input

One must determine where sporadic overload appears in the trace

Invocation of the algorithm

·         To be determined

Outputs

For each task Ti: a safe WCRTi and an approximate WCRTi with an error model. An error model is a function erri such that if we consider k consecutive instances of Ti, their response time  cannot be larger than the approximate WCRTi more than erri(k) times.

Visualization of results

Different ways of visualizing the results to be determined.