se-3910 real-time systems week 9, classes 1 and 2 – announcement* (regexp style) – significance...

30
SE-3910 Real-time Systems • Week 9, Classes 1 and 2 – Announcement* (regexp style) – Significance Testing – Failure statistics – Data flow diagrams SE-3910 - Dr. Josiah Yoder Slide style: Dr. Hornick Much Material: Dr. Schilling, Some from Dr. Hornick, etc. 1

Upload: helen-sherman

Post on 29-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

SE-3910Real-time Systems

• Week 9, Classes 1 and 2– Announcement* (regexp style)– Significance Testing– Failure statistics– Data flow diagrams

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling, Some from Dr. Hornick, etc. 1

3-Question Review Quiz• http://

www.polleverywhere.com/survey/Beo1ZC-L5• (See following slides for source material)

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling2

Normal Distribution

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling3

Normal Distribution (alternate)

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling4

http://education-portal.com/academy/lesson/normal-distribution-of-data-examples-definition-characteristics.html#lesson

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling5

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling6

http://www.stat.ufl.edu/~athienit/Tables/Ztable.pdf

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling7

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling8

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling9

Determining if two processes are “significantly” different

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling10

In-Class Example• Suppose two processes have measured means

of 12s and 13s runtime (measured from 8 samples each), and we know the standard deviation ACTUALLY IS 1 s.– Construct a statistic for determining if this is

significant– Provide a confidence interval on said statistic with

p=0.01– What is the p-value for this result?

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling11

Zoomed-in t table

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling12

Modified process (1)• Suppose we don’t know what the standard

deviation is…– What do we do?

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling13

Modified process (2)• What do we do? (if we don’t know the true std.)

– Estimate it!– But there is a side-effect. Our estimate has some

error as well.– This increases the size of our confidence interval.– By how much? Enter the t-test!

14

Modified process (3)– Use the same confidence approach as before,

except:• Simply replace the “true” with the standard• And use one of the “df” (degrees of freedom) rows

from the table• For difference of means, df = 2(N – 1) = 2N – 2• Where N is the number of samples

– Use exactly the same formula for z:• z =

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling15

Zoomed-in t table

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling16

In-Class Activity• Suppose two processes have a mean of 12.5

ms and 15ms runtime (with 5 samples for each), and we know the standard deviation ACTUALLY IS 2 ms

Is this significant at the p = 0.01 level?

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling17

Example• An airplane software system has a failure

probability of 10-6 failures per hour. What is the probability of reliable operation for a 10 hour flight?

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling18

Example• An airplane software system has a failure

probability of 10-9 per hour. Supposing this failure will cause the plane to crash. What is the chance that the plane will crash within the year if it flies 2400 hours a year?

• Compute assuming both independent and dependent.

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling19

SE-3910 - Dr. Josiah YoderSlide style: Dr. Hornick

Much Material: Dr. Schilling20

SE3910 Real Time Systems

Data flow diagrams• Entity

– An entity is the source or destination of data. – The source in a DFD represents these entities that are outside

the context of the system. – Entities either provide data to the system (referred to as a

source) or receive data from it (referred to as a sink). • Process

– The process is the manipulation or work that transforms data, performing computations, making decisions (logic flow), or directing data flows based on business rules. Data Store

• Data Store– A data store is where a process stores data between processes

for later usage by the same process or another process.• Data Flow

– Data flow is the movement of data between the entity, the process, and the data store. Data flow portrays the interface between the components of the DFD.

SE3910 Real Time Systems

Data Flow Diagram Symbols

SE3910 Real Time Systems

Case study: Traffic Control• We are going to walk through the design

of a Traffic Control System– Starting with the needs of the system.– We want to talk about how data flows

through the system.

SE3910 Real Time Systems

An Intersection System

SE3910 Real Time Systems

Intersection Control Diagram

SE3910 Real Time Systems

Dataflow diagram

SE3910 Real Time Systems

Data Dictionary• An essential aspect of a structured design

– Includes entries for data flows, control flows, data stores, buffers, etc.

SE3910 Real Time Systems

Data Dictionary Example

SE3910 Real Time Systems

Discussion: Which software failure we have talked about should have been caught using this approach?

DFD – Practical ExampleLaunched Dec. 11, 1998, the Climate Orbiter plunged too steeply into the Martian atmosphere Sept. 23, 1999, and either burned up or crashed. In an initial failure report released Oct. 15, 2000 the review board blamed the navigation error on a communications foul-up between NASA's Jet Propulsion Laboratory and prime contractor Lockheed Martin.

Collect,analyze,

generate flightcontrol data

JPL-1

J1 JPL store

Convert datafrom Metric to

English

?

LM1 LM store

Controlspaceflight

LM-1

English data

Transfer of Flight Control DataThis processwas missing

Metric data

Transfer data

?

Who wasresponsible

for this task?