L19 - Reliability Testing


RELIABILITY TESTING
- input --> collection of soft. test results
- goal --> fault-free performance

- Quanifiable Goals

Testing Objectives
- achieve reliability
- reliability evaluation

Testing Steps
1) Analysis
	- model specifying level of reliability
	- document
2) Design
	- when soft. fails, fault documented
	- testing done till sufficient data for prediciton
3) Evaluation

Reliability Models
- time-dependent
	- time b/w failures
		- independent times b/w failures
		- equal prob. of exposure
		- indepenent faults
		- faults removed after detection
	- failure count per time
		- intervals treated as independent variables
- time-independent
	- Fault Seeding
		- estimate number of faults (N), known outcomes
		- seeded faults randomly distributed
		- indegenous + seeded faults equal prob. of being detected
		1. insert (A) # of faults
		2. test
		3. count # of failures due to injection (f)
		4. count # of faulures due to inherited faults (i)
		4. estimate number of remaining inherited faults
			- A/f = N/i
		- Confidence Level of estimate
			- not all artifical faults found (f<A)
				= (A | f-1) / (E+A+1 | E + f)
			- all aritifical faults found (f=A)
				= A / (E+A+1)
			- A > (b/1-b) where b = faults/time
	- Equivalence Classes
		- input profile distribution known
		- random testing
		- input domain 
- 