outline introduction to discrete time signals and systems

12
Introduction to Discrete Time Signals and Systems Week 02, INF3190/4190 Andreas Austeng Department of Informatics, University of Oslo August 2019 AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 1 / 46 Outline Outline 1 Discrete-time signals Time-domain representation Some elementary discrete-time signals Classification of sequences Signal manipulations 2 Discrete-time systems Discrete-time systems System properties FIR and IIR LTI Systems Discrete-time systems described by difference equations AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 2 / 46 Discrete-time signals Outline 1 Discrete-time signals Time-domain representation Some elementary discrete-time signals Classification of sequences Signal manipulations 2 Discrete-time systems Discrete-time systems System properties FIR and IIR LTI Systems Discrete-time systems described by difference equations AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 3 / 46 Discrete-time signals Time-domain representation Discrete-time sequences Signals represented as sequence of numbers / samples: {x [n]} = {..., x [-1], x " [0], x [1],...}. x [n] represents the n’th sample of sequence {x [n]} N 2 N 1 . When clear, we let x [n] represent either the n’th sample or entire sequence. x [n] only defined for integer values of n, and undefined for non-integer values of n. AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 4 / 46

Upload: others

Post on 16-Oct-2021

11 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Outline Introduction to Discrete Time Signals and Systems

Introduction toDiscrete Time Signals and Systems

Week 02, INF3190/4190

Andreas Austeng

Department of Informatics, University of Oslo

August 2019

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 1 / 46

Outline

Outline

1 Discrete-time signalsTime-domain representationSome elementary discrete-time signalsClassification of sequencesSignal manipulations

2 Discrete-time systemsDiscrete-time systemsSystem propertiesFIR and IIR LTI SystemsDiscrete-time systems described by difference equations

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 2 / 46

Discrete-time signals

Outline

1 Discrete-time signalsTime-domain representationSome elementary discrete-time signalsClassification of sequencesSignal manipulations

2 Discrete-time systemsDiscrete-time systemsSystem propertiesFIR and IIR LTI SystemsDiscrete-time systems described by difference equations

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 3 / 46

Discrete-time signals Time-domain representation

Discrete-time sequences

Signals represented as sequence of numbers / samples:{x [n]} = {. . . , x [�1], x

"[0], x [1], . . .}.

x [n] represents the n’th sample of sequence {x [n]}N2N1

.When clear, we let x [n] represent either the n’th sample or entiresequence.x [n] only defined for integer values of n, and undefined fornon-integer values of n.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 4 / 46

Page 2: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Time-domain representation

Discrete-time systemsProcesses a given input sequence x [n] to generate anoutput sequence y [n].In most cases; a single-input, single output system

I Characterized through an input-output transformation:

y [n] = H{x [n]}

or x [n] �! H{·} �! y [n]

or x [n] H�! y [n].

Examples: Constant multiplier, unit delay, unit advance.I Visualized with a block diagram

I Example 2-input, 1-output systems: Adder, Signal multiplier(Modulator).

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 5 / 46

Discrete-time signals Time-domain representation

Sampled sequences

x [n] = {. . . , x [�1], x"[0], x [1], . . .}.

x [n] may be generated by periodically samling a continuous-timesignal xa(t) at uniform intervals of time, i.e.x [n] = xa(t)|t=nT = xa(nT ), n = . . . ,�3,�2,�1, 0, 1, 2, . . .T : sampling interval/period, FT = 1

T : sampling frequency,⌦T = 2⇡FT : Sampling angular frequency.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 6 / 46

Discrete-time signals Time-domain representation

Sampled sequences ...

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 7 / 46

Discrete-time signals Time-domain representation

Sampled sequences ...

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 8 / 46

Page 3: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Time-domain representation

Sampled sequences

Example (Sampling of continuous-time signal)Consider the continuous-time signalx(t) = A cos(2⇡f0t + �) = A cos(⌦0t + �)

The corresponding discrete-time signal is

x [n] = A cos[⌦0nT + �]

= A cos[2⇡⌦0⌦T

n + �]

= A cos[!0n + �],

where !0 = 2⇡⌦0/⌦T = ⌦0T is the normalized angular frequencyof x [n].

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 9 / 46

Discrete-time signals Time-domain representation

Example (continues ...)Given three signals g1(t) = cos(6⇡t), g2(t) = cos(14⇡t) andg3(t) = cos(26⇡t) with frequencies 3 Hz, 7 Hz and 13 Hz.When sampled at a samplings rate of 10 Hz, we get the threesequences g1[n] = cos[0.6⇡n], g2[t ] = cos[1.4⇡n] andg3[n] = cos[2.6⇡n].

Note: The same sample values for any given n.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 10 / 46

Discrete-time signals Time-domain representation

Example (continues ...)Another example of aliasing ...

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 11 / 46

Discrete-time signals Time-domain representation

Example (continues ...)

Theorem (Shannon Sampling Theorem)A continuous-time signal x(t) with frequencies no higher than fmax canbe reconstructed exactly from its samples x [nT ] = x(nTs), if thesamples are taken at a rage fs = 1/Ts that is greater than 2fmax .

2fmax : The Nyquist rate.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 12 / 46

Page 4: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Some elementary discrete-time signals

General discrete-time signals

x [n] = {. . . , x [�1], x"[0], x [1], . . .}

I May have infinite length.I Infinite-length sequences may further be classified as either being

right-sided, left-sided, or two-sided.A finite-length sequence is equal to zero for all values of n outsidethe interval [n1, n2].

I Length of such a sequence: Lx = n2 � n1 + 1.

In general, a discrete-time signal may be complex-valued.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 13 / 46

Discrete-time signals Some elementary discrete-time signals

Complex sequences

A complex signal may be expressedeither in terms of its real and imaginary parts,

z[n] = a[n] + |b[n] = <{z[n]}+ |={z[n]}or in polar forms in terms of its magnitude and phase,

z[n] = |z[n]|exp{|arg{z[n]}}.The magnitude square may be derived as

|z[n]|2 = z[n]z⇤[n] = <2{z[n]}+ =

2{z[n]}.

The phase may be found usingarg{z[n]} = tan�1 ={z[n]}

<{z[n]} .

z⇤[n] is called the complex conjugate, and formed by changing thesign of the imaginary part of z[n]:

z⇤[n] = <{z[n]}� |={z[n]} = |z[n]|exp{�|arg{z[n]}}.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 14 / 46

Discrete-time signals Some elementary discrete-time signals

Some elementary discrete-time signals

Unit sample sequence, �[n]. Details

I Any arbitrary sequence x [n] can be synthesized as a weighted sumof delayed and scaled unit-sample sequences.

Unit step sequence, u[n]. Details

Unit ramp sequence, ur [n]. Details

Exponential sequences, x [n] = A↵n8n.

I Real-valued exponential sequences. Details

I Complex-valued exponential sequences. Details

Sinusoidal sequences, x [n] = Acos[!0n + �], 8n. Details

Periodic sequences. Details

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 15 / 46

Discrete-time signals Some elementary discrete-time signals

The unit sample sequence

�[n] =

(1, n = 00, otherwise,

or �[n] = u[n]� u[n � 1]

−4 −3 −2 −1 0 1 2 3 4 5−0.5

00.5

11.5

Plot of δ[n]

−4 −3 −2 −1 0 1 2 3 4 5−0.5

00.5

11.5

Plot of δ[n−2]

n

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 16 / 46

Page 5: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Some elementary discrete-time signals

The unit step sequence

u[n] =

(1, n � 00, otherwise,

or u[n] =P1

k=0 �[k ]

−4 −3 −2 −1 0 1 2 3 4 5−0.5

00.5

11.5

Plot of u[n]

n

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 17 / 46

Discrete-time signals Some elementary discrete-time signals

The unit ramp sequence

ur [n] =

(n, n � 00, otherwise.

−4 −3 −2 −1 0 1 2 3 4 50246

Plot of u[n]

n

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 18 / 46

Discrete-time signals Some elementary discrete-time signals

Real valued exponential sequencesx [n] = A↵n, �1 < n < 1,where A and ↵ are real numbers.A one-sided exponential sequence ↵n, n � 0; ↵ 2 < is called ageometric series.

Geometric sequence:PN

n=0 ↵n�!

1�↵N

1�↵ , 8↵

0 10 20 30 400

10

20

30

40

50α = 1.1

Time index n

Am

plitu

de

0 10 20 30 400

0.2

0.4

0.6

0.8

1α = 0.9

Time index n

Am

plitu

de

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 19 / 46

Discrete-time signals Some elementary discrete-time signals

Complex valued exponential sequencesx [n] = A↵n, �1 < n < 1,where one or both of A and ↵ are complex numbers.Rewriteing: ↵ = e(⇢0+|!0) and A = |A|e|�

=) x [n] = |A|e|�e(⇢0+|!0)n = x<[n] + |x=[n], wherex<[n] = |A|e⇢0n cos[!0n + �]

x=[n] = |A|e⇢0n sin[!0n + �]

0 10 20 30 40−1

−0.5

0

0.5

1Plot of ’real(exp(−1/10 + j*π/5)*n)’

Time index n

Am

plitu

de

0 10 20 30 40−1

−0.5

0

0.5

1Plot of ’imag(exp(−1/10 + j*π/5)*n)’

Time index n

Am

plitu

de

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 20 / 46

Page 6: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Some elementary discrete-time signals

Sinusoidal sequencesx [n] = Acos[!0n + �], 8n,where A is the amplitude, !0 the angular frequency, and � thephase of x [n]

0 5 10 15 20 25 30 35 40−1

0

1x[n] = cos[0.1 π n]

Time index n

Am

plitu

de

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 21 / 46

Discrete-time signals Some elementary discrete-time signals

Periodic sequencesx [n] periodic iff x [n] = x [n + N] 8nFundamental period: Smallest positive integer N that satisfies therelation.Sinusoidal sequences (A cos[!0n + �]) and complex exponentialsequences (B exp[|!0n]) are periodic sequences of period N if!0N = 2⇡k , where N and k are positive intergers.

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 22 / 46

Discrete-time signals Classification of sequences

Classification of sequences

Symmetric sequences. Details

Signal energy. Details

Signal power. Details

Other types of classification Details

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 23 / 46

Discrete-time signals Classification of sequences

Symmetric sequencesA signal is conjugate symmetric (even if real) if, for all n,x [n] = x⇤[�n]A signal is conjugate antisymmetric (odd if real) if, for all n,x [n] = �x⇤[�n]Any signal can be decomposed into a sum of a conjugatesymmetric signal and a conjugate antisymmetric signal:

x [n] = xcs[n] + xca[n]where

xcs[n] =12(x [n] + x⇤[�n])

xca[n] =12(x [n]� x⇤[�n])

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 24 / 46

Page 7: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Classification of sequences

Energy signals

Signal energy, Ex =P1

�1 x [n]x⇤[n] =P1

�1 |x [n]|2.

DefinitionSignal with finite energy, Ex < 1 is called a Energy signal.

A finite sample valued, infinite length sequence may or may nothave finite energy!A finite sample valued, finite length sequence have finite energy,but zero power.

Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 25 / 46

Discrete-time signals Classification of sequences

Power signals

Average signal power: Px = limL!11

2L+1PL

n=�L |x [n]|2.

DefinitionSignal with nonzero, finite average power is called a Power signal

Periodic signals are power signals (P = 1NPN�1

n=0 |x [n]|2).Return

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 26 / 46

Discrete-time signals Classification of sequences

Other types of classification

A sequence x [n] is said to be bounded if|x [n]| Bx < 1

A sequence x [n] is said to be absolutely summable ifP1n=�1 |x [n]| < 1

A sequence x [n] is said to be square-summable ifP1n=�1 |x [n]|2 < 1

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 27 / 46

Discrete-time signals Signal manipulations

Signal manipulations

Transformation of the independent variable:The index is n modified; y [n] = x [f [n]]where f [n] is some function of n.

I Time shifting: f [n] = n � n0, i.e. y [n] = x [n � n0].I Time reversal: f [n] = �n, i.e. y [n] = x [�n].I Time scaling: f [n] = Mn or f [n] = n/N, M,N 2 N .

F down-sampling: f [n] = Mn.Then y [n] = x [Mn], i.e. every Mth sample of x [n].

F up-sampling: f [n] = n/N.Then y [n] = x [f [n]] is defined as

y [n] =

(x⇥ n

N

⇤n = 0,±N,±2N, · · ·

0 otherwise.

Time shifting, Time reversal and Time scaling operations areorder-dependent, i.e. not commutative.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 28 / 46

Page 8: Outline Introduction to Discrete Time Signals and Systems

Discrete-time signals Signal manipulations

Signal manipulations ...

The most common types of amplitude transformations are straitforward and involves only point wise operations on the signal:

Addition (and in the same way subtraction)The sum of two signals is formed by pointwise addition of thesignal values: y [n] = x1[n] + x2[n], �1 < n < 1.Multiplication (and in the same way division)The multiplication of two signals is formed by point wise product ofthe signal values: y [n] = x1[n]x2[n], �1 < n < 1.ScalingAmplitude scaling of a signal by a constant c is accomplished bymultiplying every signal value by c: y [n] = cx [n], �1 < n < 1.

I This operation may also be considered to be the product of twosignals, x [n] and f [n] = c.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 29 / 46

Discrete-time systems

Outline

1 Discrete-time signalsTime-domain representationSome elementary discrete-time signalsClassification of sequencesSignal manipulations

2 Discrete-time systemsDiscrete-time systemsSystem propertiesFIR and IIR LTI SystemsDiscrete-time systems described by difference equations

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 30 / 46

Discrete-time systems Discrete-time systems

Discrete-time systems (again)Processes a given input sequence x [n] to generate anoutput sequence y [n].In most cases; a single-input, single output system

I Characterized through an input-output transformation:

y [n] = H{x [n]}

or x [n] �! H{·} �! y [n]

or x [n] H�! y [n].

Examples: Constant multiplier, unit delay, unit advance.I Visualized with a block diagram

I Example 2-input, 1-output systems: Adder, Signal multiplier(Modulator).

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 31 / 46

Discrete-time systems Discrete-time systems

Example (M-point moving-average system)

y [n] = 1MPM�1

k=0 x [n � k ]Used to smoothing random variations in data

Example (Accumulator)

y [n] =Xn

k=�1x [k ]

=Xn�1

k=�1x [k ] + x [n] = y [n � 1] + x [n]

(or) =X�1

k=�1x [k ] +

Xn

k=0x [n] = y [�1] +

Xn

k=0x [n], n � 0.

The term y [�1] is called the initial condition.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 32 / 46

Page 9: Outline Introduction to Discrete Time Signals and Systems

Discrete-time systems System properties

Classification of discrete-time systems

Classified intoI linear and nonlinear systems, orI time-varying and time-invariant systems, orI static and dynamic systems, orI causal and non-causal systems, orI stable and unstable systems, orI passive and lossless systems.

These are system properties

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 33 / 46

Discrete-time systems System properties

Causal, stable, static and passive systems

Causal system; the output at index n0 depends only in the input upto and including the index n0, and not on future values of the input.A system is bounded-input bounded-output (BIBO) stable if, forany input that is bounded, |x [n]| Mx < 1, the output will bebounded, |y [n]| My < 1.A system is static or memoryless if the output at any time n = n0depends only on the input at time n = n0.A system is passive if, for every finite-energy input x [n], the outputy [n] has, at most, the same enery, i.e.P1

n=�1 |y [n]|2 P1

n=�1 |x [n]|2 < 1.Lossless system iff above inequality is satisfied with an equal signfor every input.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 34 / 46

Discrete-time systems System properties

System properties

AdditivityA system is said to be additive if

H{x1[n] + x2[n]} = H{x1[n]}+H{x2[n]}

for any signals x1[n] and x2[n].HomogeneityA system is said to be homogeneous if

H{cx [n]} = cH{x [n]}

for any complex constant c and for any input sequence x [n].I ) x [n] = 0 H

�! y [n] = 0, i.e. no new signals generated!

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 35 / 46

Discrete-time systems System properties

Linear systems

A discrete system is linear iffH{·} is both additive andhomogeneous, i.e. satisfiesthe general superpositionprinciple.i.e. H{a1x1[n] + a2x2[n]} =a1H{x1[n]}+ a2H{x2[n]}

To test if a system is linear, we test if it is both additive andhomogeneous.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 36 / 46

Page 10: Outline Introduction to Discrete Time Signals and Systems

Discrete-time systems System properties

Output of a linear systemAdditivity gives

I y [n] = H{x [n]} = H�P

1

k=�1x [k ]�[n � k ]

=P

1

k=�1H{x [k ]�[n � k ]}

Homogeneity givesI y [n] =

P1

k=�1H{x [k ]�[n � k ]} =

P1

k=�1x [k ]H{�[n � k ]}

If we define hk [n] to be the response to the system to a unitsample at time n = k ,

I hk [n] = H{�[n � k ]},we get that

I y [n] =P

1

k=�1x [k ]hk [n],

which is known as the superposition summation.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 37 / 46

Discrete-time systems System properties

Time-invariant systems

A linear system T is time-invariant or shift-invariant iff the followingis true:x(n) �! H{·} �! y [n] �! Shift by k �!y [n � k ]

x(n)�! Shift by k �!x [n � k ] �! H{·} �! y [n � k ].

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 38 / 46

Discrete-time systems System properties

Linear time-invariant systems

A system that is both linear and shift-invariant is referred to as alinear time-invariant systemIt is denoted an LTI-system.

Output of a linear time-invariant systemIf h[n] is the response of a LTI system to the unit sample �[n], itsresponse to �[n � k ] will be h[n � k ].The superposition summation then becomes

I y [n] =P

1

k=�1x [k ]h[n � k ] = x [n] ⇤ h[n],

where ⇤ indicates the convolution operation.

The above equation is known as the convolution sum.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 39 / 46

Discrete-time systems System properties

Impulse response and convolution sumThe response of a discrete-time system to a unit samplesequence, �[n], is denoted h[n] and called the unit sampleresponse or simply the impulse response.An LTI system is completely characterized in the time-domain byh[n], i.e. the response to the system to any input x [n] may befound once h[n] is known.Since the response of an LTI system to an input x [k ]�[n � k ] willbe x [k ]h[n � k ], the response, y [n] to an input

x [n] =X1

k=�1x [k ]�[n � k ]

will be

y [n] =X1

k=�1x [k ]h[n � k ]

=X1

k=�1x [n � k ]h[k ]

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 40 / 46

Page 11: Outline Introduction to Discrete Time Signals and Systems

Discrete-time systems System properties

Convolution sum

The summationy [n] =

P1k=�1 x [k ]h[n � k ] =

P1k=�1 x [n � k ]h[k ]

is called the convolution sum of the sequences x [n] and h[n].Compact representation: y [n] = x [n] ⇤ h[n].Properties

I Commutative: x [n] ⇤ h[n] = h[n] ⇤ x [n]I Assosiative: {x [n] ⇤ h1[n]} ⇤ h2[n] = x [n] ⇤ {h1[n] ⇤ h2[n]}I Distributive: x [n] ⇤ {h1[n] + h2[n]} = x [n] ⇤ h1[n] + x [n] ⇤ h2[n]

Performing convolution:I Direct evaluationI Graphical approachI Slide rule method

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 41 / 46

Discrete-time systems System properties

Convolution sum; properties

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 42 / 46

Discrete-time systems System properties

Performing convolution; graphical approach

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 43 / 46

Discrete-time systems System properties

Stability and causuality of LTI systems

An LTI system is BIBO stable iff h[n] is absolutely summable, i.e.P1k=�1 |h[n]| < 1.

ExampleConsider h[n] = ↵nu[n].Then

P1k=�1 |↵nu[n]| =

P1k=0 |↵

n| = 1

1�|↵| if |↵| < 1Therefore, the system is BIBO stable for |↵| < 1.

An LTI system is causal iff the impulse response h[n] = 0, n < 0.

ExampleThe discrete-time accumulator is a causal system since it has a causalimpulse response given as h[n] =

Pnk=�1 �[k ] = u[n].

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 44 / 46

Page 12: Outline Introduction to Discrete Time Signals and Systems

Discrete-time systems FIR and IIR LTI Systems

FIR and IIR LTI Systems

Causal FIR:h[n] = 0, n < 0 & n � M

Conv. formula:y [n] =

PM�1k=0 h[k ]x [n � k ]

FIR always stable since Mfinite.Causal IIR:y [n] =

P1k=0 h[k ]x [n � k ]

If h[n] = banu[n], |a| < 1,theny [n] =

P1k=0 banx [n � k ]

= ay [n � 1] + bx [n]IIR may be unstable.

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 45 / 46

Discrete-time systems Discrete-time systems described by difference equations

Linear coefficient difference-equation representation

An important subclass of LTI discrete-time systems ischaracterized by a linear constant coefficient difference equation.These are realizable in practice.h[n] = ↵nu[n] ) y [n] =

P1k=0 ↵

kx [n � k ] (1)(1) may be written asy [n] = ↵y [n � 1] + x [n] (2)(2) is a special case of a linear constant coefficient differenceequationThe general form:

y [n] =Pq

k=0 b[k ]x [n � k ]�Pp

k=1 a[k ]y [n � k ],where a[k ] and b[k ] are constants that defines the system.If one or more terms a(k) are nonzero; recursive systemIf all coefficients a(k) equal to zero; non-recursive system

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 46 / 46

End of show ...

AA, IN3190/4190 (Ifi/UiO) Signals and Systems Aug. 2019 47 / 46