Modeling: Discrete dynamics

Arijit Mondal
Dept. of Computer Science & Engineering
Indian Institute of Technology Patna
[email protected]

IIT Patna 1
Introduction
• Embedded systems can include both discrete and continuous dynamics
• Continuous dynamics can be modeled by ordinary differential equation
• State machines are used to model discrete behavior of the systems
• A system operates in a sequence of discrete steps
• Example
• Number of cars in a parking area

IIT Patna 2
Car parking
• Arrival detector, departure detector

Arrival
Detector Counter
count
X Display
Departure i
detector

• Similar to integrator
• Input is not continuous u : R → {absent, present}
• Also known as pure signal

IIT Patna 3
Event
• Systems are event triggered
• Sequence of steps known as reaction
• A particular reaction will observe the values of the inputs at a particular time and
calculate output values for the same time
• An actor has input ports P = {p1 , p2 , . . . , pN }
• Vp denotes the type of p (values may be received)
• At each reaction a variable can take p ∈ Vp ∪ {absent}

IIT Patna 4
Notion of state
• State of a system is its condition at a particular point of time
• State affects how the system reacts to inputs
• Integrator : discrete vs continuous
• Discrete modes with finite state space are called finite state machine

IIT Patna 5
Finite State Machine
• A state machine is a model with discrete dynamics that maps valuations of the inputs
to outputs where the map may depend on its current state

State 1 State 2

Initial
guard/action

State 3

IIT Patna 6
Finite State Machine: example

inputs: up, down: pure

outputs: count:{0, 1, . . . , M}

up ∧ ¬down/1 up ∧ ¬down/2 up ∧ ¬down/3 up ∧ ¬down/M

0 1 2 ... M
down ∧ ¬up/0 down ∧ ¬up/1 down ∧ ¬up/2 down ∧ ¬up/M − 1

IIT Patna 7
Transition
• It governs the discrete dynamics of FSM
• Guard/Action
• Guard determines whether the transition may take on a reaction
• Action specifies the output for each reaction
• If p1 and p2 are inputs to FSM
• true — transition is always enabled
• p1 — transition is enabled if p1 is present
• ¬p1 — transition is enabled if p1 is absent
• p1 ∧ p2 — transition is enabled if both p1 and p2 are present
• p1 ∨ p2 — transition is enabled if either p1 or p2 are present

IIT Patna 8
Default transition

up ∧ ¬down/1

0 ...
true/
¬up ∧ down/0

IIT Patna 9
Finite State Machine: example

inputs: temp: R
outputs: on, off : pure
temp ≤ 18/off

Cooling Heating

temp ≥ 22/on

IIT Patna 10
FSM Definition
• It is a tuple hStates, Inputs, Outputs, Update, InitialStatei
• States — finite number of states
• Inputs — set of input valuations
• Outputs — set of output valuations
• Update — States × Inputs → States × Outputs, mapping a state and input valuation
to a next state and a output valuation
• InitialState — start state

IIT Patna 11
FSM example
• States = {0, 1, 2, . . . , M}
• Inputs = {up, down} → {present, absent}
• Outputs = {count} → {0, 1, 2, . . . , M}
• InitialState = 0 
 (s + 1, s + 1) if s < M ∧ up = present ∧ down = absent

• update(s, i) = (s − 1, s − 1) if s > 0 ∧ up = absent ∧ down = present

 (s, absent) otherwise

inputs: up, down: pure

outputs: count:{0, 1, . . . , M}

up ∧ ¬down/1 up ∧ ¬down/2 up ∧ ¬down/3 up ∧ ¬down/M

0 1 2 ... M
down ∧ ¬up/0 down ∧ ¬up/1 down ∧ ¬up/2 down ∧ ¬up/M − 1

IIT Patna 12
A few terminologies
• Determinacy — If for each state there is at most one transition enabled by each
input value
• Update function is not one to many mapping
• Same input will produce same output always
• Receptiveness — If for each state there is at least one transition possible on each
input symbol
• FSM is receptive as ’update’ is a function, not a partial function
• Chattering — A system oscillates between two states rapidly
• Stuttering — A system remains in the state due to absence of inputs and outputs
• Hysteresis — Dependence of the state of a system on its history.

IIT Patna 13
Mealy vs Moore machine
• Mealy machine
• Named after George Mealy
• Characterized by producing outputs when a transition is taken
• Moore machine
• Named after Edward Moore
• Produces the output when the machine is in a state
• Output is function of state only
• Strictly causal

• A Mealy machine can be converted into Moore machine

• A Moore machine can be converted into Mealy machine
• Mealy machine is preferred because of compactness and output is instantaneous with
respect to inputs

IIT Patna 14
Moore machine: example

inputs: up, down: pure

outputs: count:{0, 1, . . . , M}

up ∧ ¬down up ∧ ¬down up ∧ ¬down up ∧ ¬down

0/0 1/1 2/2 ... M/M

down ∧ ¬up down ∧ ¬up down ∧ ¬up down ∧ ¬up

IIT Patna 15
Extended FSM

¬up ∧ down ∧ c > 0/c − 1

c := c − 1

variable: c : {0, 1, . . . , M}
inputs: up, down: pure counting
outputs: count:{0, 1, . . . , M} c := 0

up ∧ ¬down ∧ c < M/c + 1

c := c + 1

IIT Patna 16
Extended FSM

guard/output action
set action

variable declaration
input declaration State 1 State 2
output declaration Initial set
condition
guard/output action
set action

IIT Patna 17
Example: pedestrian crosswalk
• It starts with red
• It moves to green after 60 seconds
• It will remain in green if there is no pedestrian
• If the light goes to green, then it remains there at least for 60 seconds
• If there is a pedestrian, light becomes yellow if it has been green for more than 60
seconds
• The yellow light will remain for 5 seconds before it turns to red

IIT Patna 18
Example: pedestrian crosswalk

green

red pending

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
output: sigY , sigG , sigR : pure
green

red pending

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
output: sigY , sigG , sigR : pure
green

red pending

count := 0

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
output: sigY , sigG , sigR : pure
green

count :=
count + 1 red pending

count := 0

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
output: sigY , sigG , sigR : pure

count ≥ 60/sigG green

count := 0

count :=
count + 1 red pending

count := 0

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green
count := 0

count :=
count + 1 red pending

count := 0

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count :=
count + 1 red pending

count := 0

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count :=
count + 1 red pending count :=
count + 1
count := 0

yellow

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count :=
count + 1 red pending count :=
count + 1
count := 0 count ≥ 60/sigY
yellow count := 0

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count := pedestrian ∧ count ≥ 60/sigY

count + 1 red count := 0 pending count :=
count + 1
count := 0 count ≥ 60/sigY
yellow count := 0

IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count := pedestrian ∧ count ≥ 60/sigY

count + 1 red count := 0 pending count :=
count + 1
count := 0 count ≥ 60/sigY
yellow count := 0

count := count + 1
IIT Patna 19
Example: pedestrian crosswalk
variable: count : {0, 1, . . . , 60}
input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count := pedestrian ∧ count ≥ 60/sigY

count + 1 red count := 0 pending count :=
count + 1
count := 0 count ≥ 60/sigY
count ≥ 5/sigR
count := 0 yellow count := 0

count := count + 1
IIT Patna 19
Extended FSM
• The state of an extended machine includes not only the information about which
discrete state the machine is in, but also what values any variables have.
• Suppose there is, n discrete states, m variables each of which can take one of p possible
values
• Size of the state space will be |States| = np m

IIT Patna 20
Example
inputs: up, down: pure
outputs: count:{0, 1, . . . , M}

up ∧ ¬down/1 up ∧ ¬down/2 up ∧ ¬down/3 up ∧ ¬down/M

0 1 2 ... M
down ∧ ¬up/0 down ∧ ¬up/1 down ∧ ¬up/2 down ∧ ¬up/M − 1

¬up ∧ down ∧ c > 0/c − 1

c := c − 1

variable: c : {0, 1, . . . , M}
inputs: up, down: pure counting
outputs: count:{0, 1, . . . , M} c := 0

up ∧ ¬down ∧ c < M/c + 1

c := c + 1

IIT Patna 21
Example: infinite states

¬up ∧ down ∧ c > 0/c − 1

c := c − 1

variable: c : {0, 1, . . . , M}
inputs: up, down: pure counting
outputs: count:{0, 1, . . . , M} c := 0

up ∧ ¬down∧c < M/c + 1

c := c + 1

IIT Patna 22
Pedestrian crosswalk: state count

variable: count : {0, 1, . . . , M}

input: pedestrian : pure
count < 60/
output: sigY , sigG , sigR : pure
count := count + 1
count ≥ 60/sigG green pedestrian ∧ count < 60/
count := 0 count := count + 1

count := pedestrian ∧ count ≥ 60/sigY

count + 1 red count := 0 pending count :=
count + 1
count := 0 count ≥ 60/sigY
count ≥ 5/sigR
count := 0 yellow count := 0

count := count + 1

IIT Patna 23
Nondeterminism
• A state machine interacts with the environment
• Modeling of pedestrian

• If for any state, two distinct transitions with guards that can evaluate to true in the
same reaction, then the machine is nondeterministic
true/pedestrian
inputs: sigR, sigG , sigY : pure
output: pedestrian : pure

none waiting
true/

sigG / sigR/
crossing

IIT Patna 24
Nondeterministic FSM
• It is a tuple hStates, Inputs, Outputs, possibleUpdates, InitialStatesi
• States — finite number of states
• Inputs — set of input valuations
• Outputs — set of output valuations
• possibleUpdates — States × Inputs → 2States×Outputs , mapping a state and input
valuation to a next state and a set of possible (next state, output) pairs. Also known
as Transition Relation
• InitialStates — start states

IIT Patna 25
Nondeterministic FSM

output: sigR, sigG , sigY : pure

input: pedestrian : pure green true/sigG
true/sigG

true/sigR red true/sigY

true/sigR
yellow true/sigY

IIT Patna 26
Uses of nondeterminism
• Environment modeling — to hide irrelevant details
• Specifications — system requirements imposes constraints on some features while the
others are unconstrained

• Probabilistic FSM is different from Non-deterministic FSM

• In probabilistic FSM, every transition is associated with some probability

IIT Patna 27
Behavior & Traces
• Behavior of state machine is an assignment of such signals to each port such that the
signal on any output port is the output sequence produced by the input signals
• Example: garage counter
sup = {absent, absent, present, absent, present, present, . . .}
sdown = {absent, absent, absent, present, absent, absent, . . .}
scount = {absent, absent, 1, 0, 1, 2, . . .}
• sup , sdown , scount together form the behavior
• For deterministic FSM if input sequence is known the output sequence can be deter-
mined
• Set of all behaviors of a state machine M is called its language L(M)

IIT Patna 28
Behavior & Traces (contd.)
• A behavior may be more conveniently represented as a sequence of valuations called
observable trace
• If xi is input and Yi is output then following is an observable sequence
((x0 , y0 ), (x1 , y1 ), . . .)

• An execution trace may be defined as

((x0 , s0 , y0 ), (x1 , s1 , y1 ), . . .)

s0 x0 /y0→ s1 x1 /y1→ s2 . . .

IIT Patna 29
Computation trees
• For nondeterministic machine, it may be useful to represent all possible traces

red
true/sigR true/sigY

red green
true/sigY
true/sigY true/sigR
true/sigG
yellow yellow green red

.. .. .. .. .. .. .. ..
. . . . . . . .

IIT Patna 30