Hey it's a me again @drifter1!
Today we continue with my mathematics series about Signals and Systems in order to cover Discrete-Time Fourier Transform Properties.
So, without further ado, let's dive straight into it!
Discrete-Time Fourier Transform Usefulness
The discrete-time Fourier transform has mostly the same properties as the continous-time Fourier transform. A major difference between the two is that the discrete-time Fourier transform always results in a periodic function of Ω. In addition to that, the discrete-time Fourier transform is also specified completely over a range of 2π. In the case of real-valued sequences ranges like (0, π) are also sufficient, because of conjugate symmetry.
Let's recall the two major equations of the Discrete-Time Fourier Transform.
The first one, known as the synthesis equation, is defined as follows:
Because this equations takes in X(Ω) and gives back the original signal x[n], its most commonly referenced as the Inverse Fourier Transform.
The analysis equation or Fourier Transform is defined as follows:
As such, this equation takes in the original signal x[n] and gives the coefficients X(Ω).
Its also common to denote these two transformations as follows:
Let's cover the various properties of the Discrete-Time Fourier Transform.
Because the discrete-time Fourier transform results in a periodic function of Ω, periodicity becomes a very useful property.
As mentioned before, the Fourier transform has a period of 2π, which means that the Fourier transform at Ω is the same as the one at Ω + 2π, Ω + 2⋅2π, etc, or:
Thus, for any integer m the following is true:
Similar to continuous-time, the Fourier Transform of discrete-time also satisfies linearity.
As such, the following is true, for any number of weighted signals:
Time and Frequency Shifting
Using specific substitutions on the analysis and synthesis equations of the discrete-time Fourier transform, its possible to specify time and frequency shifting properties.
More specifically, using x[n - no] instead of x[n] in the analysis equation leads to the time-shifting property:
Similarly, a substitution of X(Ω) with Χ(Ω - Ωo) leads to frequency-shifting, which is defined as
Conjugation and Conjugate Symmetry
As mentioned before, the discrete-time fourier transform also satisfies conjucation and conjugate symmetry.
The result of the Fourier Transform, X(Ω), can be re-written as:
When the signal x[n] is real, conjugate symmetry is satisfied, which means:
because the real part is even and the imaginary part is odd:
Differencing and Accumulation (basically Differentation and Integration in Discrete-Time)
In Discrete-Time there is no differentation and integration, but its possible to define similar things.
Differentiation can be thought of as the difference between a pair of Fourier transforms, basically x[n] - x[n - 1], which leads us to the differencing property:
Accumulation is the analog of Integration and defined as follows:
I'm not quite sure about these two equations, as the notation was quite different in my two references (ejω = Ω in MIT/Openheim for the most part) and the MIT notes skip these two properties.
Time Reversal and Expansion
Next up are the properties for time reversal and expansion.
Similar to continuous-time, the time-reversal of x[n], x[-n], leads to the inversion of the Fourier transform as well, or in maths:
In discrete-time time scaling is known as time expansion. This process is more difficult to replicate in discrete-time. Parallelizing a signal x[n] by a factor k leads to a signal x[n / k]. This signal is non-zero only when n is a multiple of k (or n / k is an integer). Oppenheim denotes this signal as x(k)[n].
Let's summarize the definition of the time-expanded signal:
The corresponding Fourier transform of such a signal is given by the equation:
Differentiation in Frequency
In addition to the accumulation and differencing properties of time, its also possible to define the derivative in frequency.
Differentiating both sides of the analysis equations by Ω, and multiplying by j leads to the following property:
Let's note that integration is not defined in frequency.
Parseval's relation can also be defined for discrete-time.
For any Fourier transform pair x[n] and X(Ω) the following is true:
The square of the Fourier transform X(Ω) is referred to as the energy density spectrum of signal x[n]. Integrating from -π to +π yields finite energy, which tells us that x[n] has finite energy.
Convolution and Multiplication
Exactly like in continuous-time, the Fourier transform of the convolution of two discrete-time signals is equal to the multiplication of their individual Fourier transforms.
Mathematically, in the case of a generic LTI system with input x[n] and impulse response h[n], its possible to define the output y[n] and its Fourier Transform Y[Ω] as follows:
Similar to continuous-time, multiplication is again known as modulation, with the result being the convolution of the individual Fourier transforms.
The main difference is that the convolution is now periodic, and not aperiodic.
Mathematically, the modulation of the signals x[n] and y[n] is defined as follows:
Duality in Discrete-Time
There is no duality between the discrete-time Fourier transform synthesis and analysis equations, but there is one amongst the Fourier series equations of discrete-time. A duality relation also exists between the discrete-time Fourier transform and continuous-time Fourier series.
Duality between the Discrete-Time Fourier Series equations
In the case of continuous-time periodic signals there is an inherent asymmetry which yields a lack of duality between the time and frequency domains. On the other hand, aperiodic signals show a strong duality, with the Fourier transform of the Fourier transform returning the original signal, but time-reversed.
Let's now recall the discrete-time Fourier series equations:
In the case of the discrete-time domain, the situation is exactly opposite. The Fourier series represents a periodic sequence using a periodic sequence of Fourier series coefficients. In contrast to that, the Fourier transform represents an aperiodic sequence using a continuous periodic function, its Fourier transform. Thus, only the Fourier series equations show a strong duality, which means that every property of the discrete-time Fourier series has a dual.
For example, one pair is time-shifting and frequency-shifting:
Duality between the Discrete-Time Fourier Transform and Continuous-Time Fourier Series
There also exists a strong duality between the continuous-time Fourier series and the discrete-time Fourier transform.
Its easy to notice this relationship by comparing the analysis equation of the Fourier series and the synthesis equation of the Fourier transform, or vise versa:
In conclusion, there are three dualities in total:
- Between the continous-time Fourier transform analysis and synthesis equations
- Between the discrete-time Fourier series analysis and synthesis equations
- Between the continous-time Fourier series and discrete-time Fourier transform equations (synthesis to analysis)
- Alan Oppenheim. RES.6-007 Signals and Systems. Spring 2011. Massachusetts Institute of Technology: MIT OpenCourseWare, License: Creative Commons BY-NC-SA.
Mathematical equations used in this article were made using quicklatex.
Block diagrams and other visualizations were made using draw.io
Previous articles of the series
- Introduction → Signals, Systems
- Signal Basics → Signal Categorization, Basic Signal Types
- Signal Operations with Examples → Amplitude and Time Operations, Examples
- System Classification with Examples → System Classifications and Properties, Examples
- Sinusoidal and Complex Exponential Signals → Sinusoidal and Exponential Signals in Continuous and Discrete Time
- LTI System Response and Convolution → Linear System Interconnection (Cascade, Parallel, Feedback), Delayed Impulses, Convolution Sum and Integral
- LTI Convolution Properties → Commutative, Associative and Distributive Properties of LTI Convolution
- System Representation in Discrete-Time using Difference Equations → Linear Constant-Coefficient Difference Equations, Block Diagram Representation (Direct Form I and II)
- System Representation in Continuous-Time using Differential Equations → Linear Constant-Coefficient Differential Equations, Block Diagram Representation (Direct Form I and II)
- Continuous-Time Periodic Signals & Fourier Series → Input Decomposition, Fourier Series, Analysis and Synthesis
- Continuous-Time Aperiodic Signals & Fourier Transform → Aperiodic Signals, Envelope Representation, Fourier and Inverse Fourier Transforms, Fourier Transform for Periodic Signals
- Continuous-Time Fourier Transform Properties → Linearity, Time-Shifting (Translation), Conjugate Symmetry, Time and Frequency Scaling, Duality, Differentiation and Integration, Parseval's Relation, Convolution and Multiplication Properties
- Discrete-Time Fourier Series & Transform → Getting into Discrete-Time, Fourier Series and Transform, Synthesis and Analysis Equations
Final words | Next up
And this is actually it for today's post!
I'm thinking of getting into examples/exercises of some sort for the topics that we covered!
Keep on drifting!