Signal Summary
Signal Summary
Time Scaling
Time Reversal
Only the case a > 1, where a is an integer, makes sense, as x[n] is undefined for fractional values of n. In this case, called decimation or downsampling, we not only get a time compression of the signal, but the signal can also lose part of its information; That is, some of its values may disappear in the resulting signal y[n].
EXPONENTIAL SIGNALS
This means that the output of an LTI system subjected to an exponential input signal will also be an exponential with the same exponent, but in general with a different real or complex amplitude.
Real Exponential Signals
1. Continuous Time
2. Discrete Time
2.
Discrete Time
Using Eulers relation, we get the rectangular form of the discrete-time complex exponential:
Clearly, the magnitude r of a determines whether the envelope of x[n] grows, decreases, or remains constant with time.
PERIODIC COMPLEX EXPONENTIAL AND SINUSOIDAL SIGNALS In our study of complex exponential signals so far, we have found that in the cases in continuous time and in discrete time, we obtain signals whose trajectories lie on the unit circle in the complex plane. In particular, their real and imaginary parts are sinusoidal signals. We will see that in the continuous-time case, these signals are always periodic, but that is not necessarily the case in discrete time. Periodic complex exponentials can be used to define sets of harmonically related exponentials that have special properties that will be used later on to define the Fourier series.
Continuous Time
In continuous time, complex exponential and sinusoidal signals of constant amplitude are all periodic. Periodic Complex Exponentials Consider the complex exponential signal periodic with fundamental period complex exponential signals: . We have already shown that this signal is . Now let us consider harmonically related
That is, complex exponentials with fundamental frequencies that are integer multiples of . These harmonically related signals have a very important property: they form an orthogonal set. Two signals inner product is equal to zero: are said to be orthogonal over an interval if their
Where x*(t) is the complex conjugate of x(t). This notion of orthogonality is a generalization of the concept of perpendicular vectors in three-dimensional Euclidean space . Two such
We know that a set of three orthogonal vectors can span the whole space by forming linear combinations and therefore would constitute a basis for this space. It turns out that harmonically related complex exponentials (or complex harmonics) can also be seen as orthogonal vectors forming a basis for a space of vectors that are actually signals over the interval . This space is infinitedimensional, as there are infinitely many complex harmonics of increasing frequencies. It means that infinite linear combinations of the type can basically represent any function of time in the signal space, which is the basis for the Fourier series representation of signals.
Sinusoidal Signals
Discrete Time
Complex Exponential Signals The complex exponential signal seems like it is for any w0. is not periodic in general, although it
are periodic of (not necessarily fundamental) period N. They are also orthogonal, with the integral replaced by a sum in the inner product:
Sinusoidal Signals Discrete-time sinusoidal signals of the type are not always periodic. Mathematically, we saw that x[n] is periodic if there exists an integer N > 0 such that
as well.
The total energy and average power over of an arbitrary integrable continuous-time signal x(t) are defined as though the signal were a voltage across a one-ohm resistor:
And over
And over
The class of continuous-time or discrete-time finite-energy signals is defined as the set of all signals for which The class of continuous-time or discrete-time finite-power signals is defined as the set of all signals for which
The total average power of a periodic signal can be calculated over one period only as:
Note that
Even: x(t) x(-t) symmetric with respect to the vertical axis. Odd: x(t) -x(-t) symmetric with respect to the origin. In odd signals x(0) 0
Any signal can be decomposed into its even part and its odd part as follows:
unit impulse
The unit step is the running sum of the unit impulse: And conversely, the unit impulse is the first-difference of a unit step:
Also, the unit step can be written as an infinite sum of time-delayed unit impulses:
By the Principle of Superposition, the response y[n] of a discrete-time linear system is the sum of the responses to the individual shifted impulses making up the input signal x[n]. Let hk[n] be the response of the LTI system to the shifted impulse then:
If we knew the response of the system to each shifted impulse calculate the response to any input signal x[n] using Equation 2.2.
, we would be able to
It gets better than this: for a linear time-invariant system (the time-invariance property is important here), the impulse responses hk[n] are just shifted versions of the same impulse response for k = 0 :
If we substitute h[n-k] for hk[n], we obtain the convolution sum that gives the response of a discrete-time LTI system to an arbitrary input.
The convolution operation has the following properties. It is Commutative, Associative, Distributive, Commutative with respect to multiplication by a scalar, Time-shifted when one of the two signals is time-shifted, Finally, the convolution of a signal with a unit impulse leaves the signal unchanged.