Scribd
Upload
361 views37 pages

Signals and Systems: LTI Systems Defined by Difference Equations

This document discusses linear time-invariant (LTI) systems defined by difference equations. It defines difference equations as the discrete-time analog of differential equations for continuous-time systems. Difference equations express the input/output relationship of an LTI system through a constant-coefficient difference equation. The document provides examples of how difference equations can model real-world systems like savings accounts and car loans. It then discusses techniques for solving difference equations non-recursively, including finding the impulse response and using convolution.

Uploaded by

fdkgfdgk lksfldsfk
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
361 views37 pages

Signals and Systems: LTI Systems Defined by Difference Equations

This document discusses linear time-invariant (LTI) systems defined by difference equations. It defines difference equations as the discrete-time analog of differential equations for continuous-time systems. Difference equations express the input/output relationship of an LTI system through a constant-coefficient difference equation. The document provides examples of how difference equations can model real-world systems like savings accounts and car loans. It then discusses techniques for solving difference equations non-recursively, including finding the impulse response and using convolution.

Uploaded by

fdkgfdgk lksfldsfk
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 37

Signals and Systems

LTI Systems Defined by


Difference Equations
Ertem Tuncel
Professor & Chair of Electrical and Computer Engineering
University of CA, Riverside
Difference equations
• The input/output relation of an LTI system can
sometimes be expressed as a constant-
coefficient difference equation.
• Analogous to constant-coefficient differential
equations for continuous-time systems.
• Example:

is very much like


Difference equations
• Example:

• We can solve this recursively if two initial


conditions, y[0] and y[1], are given to us:

CALCULATED GIVEN GIVEN KNOWN KNOWN


Diff. equations in "real" life
• Example: You open a CD account with $1,000.
The account has an annual percentage rate
(APR) of 6%. At the end of each month n, you
are allowed to deposit an additional x[n]
dollars as cash.
• The money in the account at the end of month
n is governed by the difference equation

with the initial condition y[0]=1,000.


Diff. equations in "real" life
• Example: You finance a $20,000 car using a
60-month loan with 3% APR. You pay $5,000
up front as down payment. What is your
monthly payment?
• If you pay P dollars each month, the loan
amount at the end of month n is governed by
the difference equation

with the initial condition y[0]=15,000.


• P must be chosen to satisfy y[60]=0.
Non-recursive solutions
• It seems that we need a technique to solve
difference equations non-recursively.
• Otherwise, it is virtually impossible to solve
the last problem, for instance.
• We might guess a P, solve the difference equation
recursively, and see if y[60]=0. If y[60]>0, increase
P and try again. If y[60]<0, decrease P and try again.
• This is a very painstaking process, and car dealers
certainly do not do this when they are offering you a
loan.
How to solve
• Without the initial conditions, there will be an
infinite family of solutions. Initial conditions
are used to zero in on the unique solution.
• To find that infinite family of solutions, start
with just "a" solution, called the particular
solution.
• For the car loan example

try the solution y[n] = K.


How to solve
• For the car loan example

try the solution y[n] = K.

• This yields the particular solution


• Add onto this the homogeneous solution, i.e.,
the solution to
How to solve

• Why is a solution?
• Because

+
How to solve
• Conversely, if y[n] is a solution,
must be a homogeneous solution:

• Conclusion: all solutions must be of the form


How to solve
• Back to the homogeneous solution

• What function increases with a factor r when


its argument increases by 1?
• No function other than for some arbitrary c.
• Therefore,
How to solve
• Therefore,

which makes

the entire family of solutions.


• Now find c using y[0]=15,000:

meaning .
How to solve
• Recall that the original problem asked for the
monthly payment P in order for y[60] = 0.

• Solving for P, we obtain


Resultant y[n]
General car loan case
• Loan amount: L dollars.
• Interest: I % APR.
• Loan period: M months.
• Monthly payment: P dollars.
• The solution is given by

with .

• The monthly payment is


Causal LTI systems
• If the system is known to be causal, we do not
need explicit initial conditions.
• The system would be in initial rest, i.e.,
y[n] = 0 for all n before the input comes.
• We could use this fact together with the
recursive method to figure out a few y[n]'s
after the input comes (as many as we need to
solve the difference equation).
Causal LTI systems
• Under the initial rest regime, the system
indeed becomes linear and time-invariant.
• If y[n] is the solution to the difference equation

with input x[n], then

• That is, is a solution for input


TIME-INVARIANT
Causal LTI systems
• Similarly, if and are solutions to
the difference equation

with inputs and , respectively, then

• That is, is a solution for input


LINEAR
Causal LTI systems
• Example: Consider the causal LTI system
whose input-output relation is given by

Find y[n] if x[n] = u[n].


• Solution: Since the system is causal, we
immediately have y[n] = 0 for n < 0.
• For , we have the equivalent equation
• For , we have the equivalent equation

• For particular solution, try :

or .
• For homogeneous solution, try :

or
or

• So both and are homogeneous solutions.


• For , we have the equivalent equation

• Combining these together,

becomes the family of all solutions.


• To find and , it suffices to know y[0] and
y[1].
• But since we know that the system is in initial
rest,
• Recall that this was all for , so
Causal LTI systems
• Example: Consider the causal LTI system
whose input-output relation is given by

Find y[n] if .
• Solution: Since the system is causal, we
immediately have y[n] = 0 for n < 0.
• For , we have the equivalent equation
• For , we have the equivalent equation

• For particular solution, try :

or

or K = 1.
• For homogeneous solution, try :

or
or
• For , we have the equivalent equation

• Combining these together,

• But, Im
0.5+0.5j

Re

• Therefore,
• Since we know that the system is in initial rest,

• Use these to find the unknown coefficients:

• In other words,
• Recall that this was all for , so
Finding the impulse response
• For every input, we need to guess another
particular solution. This can be hard for some
inputs.
• Since every difference equation with initial
rest describes an LTI system, why not find the
impulse response instead, and use the
convolution sum to find the output for any
input?
Finding the impulse response
• Example: Consider the first example with

Find the impulse response, and find the output


for the input x[n] = u[n] using convolution.
• Solution: In other words, we are asked to solve

• The important observation is that for n > 0, we


got ourselves a homogeneous equation. So, no
need to guess and try a particular solution.
Finding the impulse response

• Recall that we already found the homogeneous


solution to be of the form .
• As before, and are determined by
and .
Finding the impulse response

• Thus, the impulse response is given by

• Now, when the input is x[n] = u[n],


Finding the impulse response
• Example: Consider the causal LTI system with

Find the impulse response.


• Solution: As before, we will use the
homogeneous solution

we derived earlier, together with the initial


conditions h[0] and h[1] we will derive
recursively.
Finding the impulse response

• Therefore,
Block diagrams
• We can also represent the LTI system for

with a block diagram using basic elements:

D
a
Block diagrams
• Example:
• This is the same as

D
Block diagrams
• Example:

D D

2 2
D

You might also like