Real-Time operating

System
Outline

Real-Time System
Task categories
Why we need scheduling
Real time scheduling taxonomies
Scheduling Periodic Tasks
 Real-Time Scheduling Algorithms
Static scheduling algorithms
Dynamic scheduling algorithms
Hybrid algorithm
references
Real-Time System
• Real-time systems have been defined as: “those systems
in which the correctness of the system depends not only
on the logical result of the computation, but also on the
time at which the results are produced“
• Correct function at correct time
• Usually embedded
• Deadlines
• Hard real-time systems
• Soft real-time systems
Real Time Tasks
• Task in which performance is judged on the basis of time.
• The result of task are good if has produced within specified time
constraints otherwise system fail or reduced value for the quality of
service.

• Process control in industrial plants

• Robotics
• Air Traffic control
• Telecommunications
• Weapon guidance system e.g. Guided missiles
• Medical diagnostic and life support system
• Automatic engine control system
• Real time data base
Real time speech or video processing
• A speech or moving picture sample of 1 second, if processed
in 1 second or in less, makes it real time processing
• if it takes more than 1 sec , it is no longer rel-tine processing
RTOS
Question: Is real time systems and embedded systems are same?
• Embedded systems are designed for specific set of applications.
• When such a system requires ‘time constrained’ operation, it
becomes real-time embedded system
• All embedded system are not real-time system.
• E.g.
• printer is not real-time system
• A robot which counts objects passing through conveyor belt is real-
time system
Terms and definition

• Release time (or ready time): This is the time instant at which a
task(process) is ready or eligible for execution

• Schedule Time: This is the time instant when a task gets its chance
to execute

• Completion time: This is the time instant when task completes its
execution

• Deadline: This is the instant of time by which the execution of task

should get completed.

• Runtime: The time taken without interruption to complete the task,

after the task is released
 Terms and definition

• Tardiness: amount of time by which task misses its deadline

• Laxity : deadline minus the remaining computation time. Max
time task can wait and still meet its deadline.
 Schematic of a Time
Constrained Computation
 Types of Real Time Systems
• Hard real time systems
• Must always meet all deadlines
• System fails if deadline window is missed
• Hard Real time task:
• Task must complete before or at deadline.
• Value of completing the task after deadline is zero.
E.g. automobile braking system
• Air traffic control
• Vehicle subsystems control
• Nuclear power plant control
Soft Real Time System
• Soft real time systems
• Must try to meet all deadlines
• System does not fail if a few deadlines are missed
• Software Real Time
• Missing deadline is penalty
• Penalty increases as tardiness increase

E.g. Weather stations have many sensors for reading temperature,

humidity, wind speed, etc. The readings should be taken and
transmitted at regular intervals, however the sensors are not
synchronized. Even though a sensor reading may be early or late
compared with the others it can still be relevant as long as it is close
enough.
Soft Real Time System
• Example:
• Imagine a game in which scores of the players are displayed
continuously.
• If scores are displayed late , no catastrophe occurs, but
decisions have to be based on scores, if scores are outputted
beyond deadline, we can say that performance of scoring
system is bad.
Firm real time system
• Firm real time systems
• Result has no use outside deadline window
• Tasks that fail are discarded
Firm real time task
• Its value reduces to zero if deadline is not met
• Output of task is discarded if deadline is not met.
• Output of delayed execution is dropped.
• Deadlines are allowed to be missed once in a while , but not too
frequently

E.g. decoding of video frame may occasionally get delayed by reasons like
unexpected interrupts and the like. when such frames are delivered late,
the playback does not look good. Skipping such frames and making sure
that not too much such frames are delayed will be better. Because effect
will be less noticeable to user than playing back the delayed frames.
Real time tasks
• Periodic
Tasks which are activated regularly at fixed rate
Periodic tasks must execute once per period
 Each task is repeated at a regular interval
 Max execution time is the same each period
 Arrival time is usually the start of the period
 Deadline is usually the end
Real time tasks
•Aperiodic
Stream of jobs arriving at irregular intervals
Each task can arrive at any time
• Inter-arrival period between two such tasks can be zero.
• Aperiodic tasks implied to have soft deadlines and aim of
scheduling is to provide fast response time.
Real time tasks
• Sporadic is a an aperiodic task with hard deadline and a
minimum inter-arrival time.
• Without minimum inter-arrival time restriction, it is impossible
to guarantee that a deadline of sporadic task would always be
met.
• E.g. such tasks are emergency conditions like fire
Preemptive and non-Preemptive
tasks
• Preemptive task:
Task which can be preempted if another task of higher priority
becomes ready.

• Non- preemptive:
execution of non-preemptive task should be without
interruption, once started
Preemptive task
Non-preemptive task
Real Time Operating System
• Some embedded applications require only dedicated
hardware and firmware e.g. mp3 player, printer and scanner
etc.
• Such system are called superloop system
• Whole code for systems is written in one loop which executes
continuously.
• When external event come, interrupts are generated to alert
the processor and system respond appropriately.
• There can be no. of inputs and corresponding actuators too.
• code for all these is in flash memory.
Real Time Operating System
• Some embedded system like mobile phone need a manager
• all embedded system need not real –time operating system.
Only where time constraint is a factor real-time OS is required
What does RTOS does?
• RTOS provides abstraction layer between embedded hardware
and application software
• RTOS manages interaction between hardware and
applications.
• RTOS ensures that the multiple tasks that comes in are
managed and done on time.
• OS has a kernel which form core of OS.
Hardware- Software hierarchy
in complex embedded system
Kernel services of RTOS
Real-Time Scheduling Algorithms

Off-line On-line

Static Dynamic
Priortity Priority

Preemptive Non-Preemptive Planning Best

Based Effort

Fixed Priority Algorithms Dynamic Priority Algorithms

Rate Deadline
Monotonic Monotonic Earliest
Least Laxity
scheduling scheduling Deadline
First
First
Off Line Scheduling (Pre Run time
scheduling)
• They generate scheduling information prior to system
execution (Deterministic System Model)
• This scheduling is based on :
• Release time
• Deadlines
• Execution
Scheduling algo can use precise schedule which optimizes several
different measures and optimal algo can be used which
guarantee very good system performance.
E.g. Fixed factory jobs where nothing changes under normal
conditions can use this approach
• Disadvantage: Inflexibility, If any parameter changes, the
policy will have to be redone
On-Line Scheduling
• Number and types of tasks, associated parameters are not
known in advance.
• Scheduling must accommodate
• dynamic changes in user demands
• Availability of resources

• Online Scheduling are of two types:

• Static Priority
• Dynamic Priority
Static Priority
• Tasks with highest priority gets the chance to execute first

• This can be preemptive or Non-preemptive

• Non preemptive: task with highest priority runs till it
completes.
• Preemptive: The execution of task can be pre-empted when
higher priority task appears in ready queue.
Dynamic Priority
• Priority can be allowed to change at run time
• Scheduling needs more computation
• Pre-emption may or may not be used
• Flexibility is quite high in such system
• Two subsets:
1. Planning Based:
• Guarantees deadline for all accepted task.
2. Best effort algorithm does its best to maximize performance.
• Guarantees meeting deadline for hard time task.
• Optimizes the performance of soft time task
Problem: Static Priority

Tasks Priority Period CPU Burst

T1 1 7 2
T2 2 17 4
T3 3 24 8

• Draw the Gantt-chart for scheduling these tasks.

• Assume the all jobs have same release time
• Schedule the process to meet their deadline
• Without Pre-emption
• With Pre-emption
• Note: Deadline of task is the time when its next burst arrive
Without preemption

Tasks Priority Period CPU Burst

T1 1 7 2
T2 2 17 4
T3 3 24 8
With Preemption

Tasks Priority Period CPU Burst

T1 1 7 2
T2 2 17 4
T3 3 24 8
 T1 T1 T2 T1

T1 T2 T3 T1 T3 T1 T3 T2 T1 T3
0 2 6 7 9 14 16 17 21 23 24

Tasks Priority Period CPU Burst

T1 1 7 2
T2 2 17 4
T3 3 24 8
Rate Monotonic Algorithm
• Introduced by Liu and Layland in 1973
• Assigning priorities as a monotonic function of the rate of a
(periodic) process.
• (Monotonic means either increasing or decreasing)
• Priorities are assigned according to increased period of a
process.
• As period increases the priority decreases.
• The process with lowest period will have highest priority
• A set of tasks is said to be schedulable if all of the tasks can
meet their deadlines.
RM Algorithm
• RM provides simple inequality, to verify the sufficient
condition for RM algorithm

• LHS of inequality is total CPU utilization for n tasks

• Where C is CPU Burst , P or T is Period of task
• If the condition satisfied, RM will schedule tasks within their
respective deadline
• This is sufficient, but not necessary condition i.e. Set of task
which satisfy this condition will definitely be schedulable, but
there can be set of tasks which do not satisfy this condition,
and still are schedulable.
RHS of expression
RMA algo
• If a task set passes the Liu and Layland test, then it is
guaranteed to be RMA schedulable. On the other hand, even if
a task set fails the Liu and Layland test, it may still be RMA
schedulable.
Example 1
Tasks Priority Period (T) CPU Burst(C)
T1 1 7 2
T2 2 17 4
T3 3 24 8

• CPU utilization for the set of task:

• LHS : 2/7 + 4/17 + 8/24 = 0.812
• RHS : = 3 X (2 1/3 -1)= 3 x .26 = .78
• LHS> RHS :RM does not get satisfied
• But the task still might get schedulable by RM technique
• In last example Static priority with pre-emption was implementation
of RM algorithm because priorities are ordered in decreasing order of
their priorities
Example2
Tasks Period CPU Burst
T1 12 5
T2 7 3
• CPU utilization= 5/12+3/7=0.845
• For two tasks RHS of inequality is 0.828
• Since LHS>RHS sufficient condition for schedulability is not
satisfied.
• Task T2 has higher periority (lower period) and so its burst
completed first
Example2
Example2

• Both tasks able to finish before their deadline, but processor is

idle for certain periods
Example3
Tasks Period (P or T) CPU Burst(C)
T1 15 4
T2 12 2
T3 20 5

• CPU utilization =4/15+2/12+5/20=0.684

• Bound for scheduling for three tasks is 0.782
• Since LHS <RHS
• Sufficient condition is satisfied, this task set is definitely
schedulable using RM algorithm
Example 3

Tasks Period (P or T) CPU Burst(C)

T1 15 4
T2 12 2
T3 20 5
Example 4

Tasks Period CPU Burst Release Time

T1 3 1 0
T2 10 3 1
T3 15 4 3
Earliest deadline First
• Belong to class of dynamic priority allocation method
• Priority of task changes at run time
• At any instant, priority task is one that has the closest
deadline.
• Tasks that cannot be scheduled by RM can be scheduled by
this method.

Necessary condition for Schedulability:

CPU utilization must be less than 1.
example
Tasks Period CPU Burst
T1 4 1
T2 6 2
T3 9 4

• CPU utilization=1/4+2/6+4/9=1.027
Using RM technique

Tasks Period CPU Burst

T1 4 1
T2 6 2
T3 9 4
Using EDF

Tasks Period CPU Burst

T1 8-12 1
T2 12 2
T3 9-18 4
Example 2
Disadvantages of EDF
• It is dynamic priority algorithm and therefore requires
dynamic determination of priorities.
• It is not as controllable as static priority algorithms.
• More overheads are required to implement this
• Are not usually used in systems which require absolute
predictability
• Slightly greater overheads than fixed priority
• Dynamic priorities must also be calculated at each decision
point, where as static never change and never have to be
recalculated.
Qualities of RTOS
• Performance: fast and high throughput
• Reliability: should not fail or mean time between failure very
high
• Compactness: software and OS ported to memory , size of OS
as small as possible
• Scalability: unnecessary services are removed, OS must be
scale down. Modularity in design of OS, to allow to be scaled
down or up , as necessary.