What are embedded systems?
application specific
information processing system
which is embedded into
a bigger technical context or technical system
What do embedded systems consist of?
cooperating
optimized
hardware and software components
Are embedded systems majorly different from general purpose computing systems? (servers, pcs)?
yes
-> tasks differ substantially
What points in time are there for a real-time ES?
r_s -> release time
point in time where job becomes available for execution by the ES ( i.e. sensor data arrives…)
t_r -> response time
time from release time until embedded system finishes and responds
r_d -> (relative) deadline
time the ES has left (from the release time on) to respond
t_exe -> execution time (between r_s, r_s+t_r
real time:
t_r < t_d
What does the abbreviation RAS stand for?
reliability
availability
serviceability
How is safety defined?
failure of an embedded system
has no catastrophic consequence
for technical context
in higher-order system
What means and is the difference between fail-safe and fail-operational?
fail safe:
in face of failure or malfunction
-> shut down or go into safe state
fail operaional:
in face of failure or malfunciton
-> is able to continue operation (i.e. safety critical system like aircraft control…)
In what state of the development cycle should one consider RAS?
must not be added as afterthought
-> design it into the system from the very beginning
How is reliability (idea wise) defined?
R(t): Probability that system works correct until time t
! presuming it worked correct at reference time t0 = 0
What different types of changes in failure rate exist?
early “infant mortality” failure
wear out failure
constant (random) failure
=> stacks / adds up to
observed failure rate
What different durations/metrics exist in availability?
MTTR: mean time to repair
how long to go after failure back to operational
MTTF:
how long after last failure till next failure occurs
MTBF:
how long since (initial) occurence of last failure till next occurs?
How is availability defined?
A: fraction of time the system works correct in between two consecutive failures
How is serviceability measured?
S: measure considering the time it takes to repair system after a benign faiulre
-> MTTR
How does reliability of two parts of a system stack when they both need to work correct for the system to be operational?
add both lambda -> lambda_sys = lambda_1 + lambda_2
MTTF_sys = 1/(lambda_1 + lambda_2)
…
=> if vastly different -> reliability gets dominated by lowest MTTF (largest lambda)
How can one calculate the expected time of a system running with a failure probability lower of e.g. 10%?
need to find t until reliability is 90%
-> F(t) = 1 - R(t) (failure probability 10%…)
-> R(t) !>= 90%
-> e^(lambda_sys*t) >= 90%
ln(e^(lambda_sys*t)) >= ln(90%)
-(lambda_sys*t) >= ln(0.9)
t <= ln(0.9) / lambda_sys
How to calc the failure rate in a system with more than one component?
system failure rate is the sum of the individual failure rates
-> assuming all have components have to work in order for the system to work…
How to calc the failure rate in a system with identical, redundant components?
System reliablity = 1 - (1 - component reliability)^3
MTTF = integral from 0 to infinity over system reliability
system failure rate = 1 / MTTF system
What are additional requirements for HW/SW systems?
efficiency
cost, energy, execution time, weitht, area, …
often multi-dimensional optimization targets
real time capability
Last changeda year ago
