Scribd
Upload
116 views

RiceX - Discrete Time Signals and Systems, Part 1 Time Domain

This document provides a summary of course notes for RiceX: ELEC301.1x Discrete Time Signals and Systems, Part 1: Time Domain. Key concepts covered include [1] properties of signals including shifting, [2] important test signals such as delta functions and sinusoids, [3] representing signals as vectors and calculating their inner products, and [4] introducing linear time-invariant systems and computing convolution.

Uploaded by

Cassia Lmt
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
116 views

RiceX - Discrete Time Signals and Systems, Part 1 Time Domain

This document provides a summary of course notes for RiceX: ELEC301.1x Discrete Time Signals and Systems, Part 1: Time Domain. Key concepts covered include [1] properties of signals including shifting, [2] important test signals such as delta functions and sinusoids, [3] representing signals as vectors and calculating their inner products, and [4] introducing linear time-invariant systems and computing convolution.

Uploaded by

Cassia Lmt
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

RiceX: ELEC301.

1x Discrete Time Signals and Systems, Part 1: Time Domain (brief course notes)

Week 1

Signal Properties
1) Modulo Arithmetic y[n]=x[(n)N]
2) Causal (t<0, x[n]=0), anti-causal (t>0,x[n]=0), acausal
3) Any signal can be decomposed into even and odd component: x[n]=e[n]+o[n]

Shifting Signals

4) Periodic Shift/Circular Shift: Assume a finite-length signal as periodic before shifting

Key test signals


1) Delta function (unit impulse): delta[n] = {1 if n= 0} ; {0, if otherwise}
Picks out one signal at m using delta[n-m]
2) Unit step u[n-m] = {1, n-m≥ 0}; {0, n-m<0}
3) Unit Pulse (boxcar): p[n] == {0, n<N1; 1, N1≤n≤N2; 0, n>N2}
4) Real exponential

Sinusoids
Discrete-Time sinusoids are weird

1. Aliases: two sinusoids with diff freq but are identical


2 πk
2. Aperiodicity: for y= sin(ω ( n )+ ϕ ¿ ,if ω ≠ (not harmonic freq), y is not periodic
N

Complex exponentials

Week 2

Signals are vectors – Part 3


1) Energy/Euclidean length/2-norm of vector x: ||x|| 2= square every element in x, sum them up, sqrt ans
2) 1-norm of a vector: ||x||1 = sum the absolute value of all elements
3) Infinity-norm: max[x[n]] (measures peak value of signal)
4) Normalizing a vector x: ||x||2 =1; divide all elements by ||x||2

Signal inner/dot product


1) Conjugate/Hermitian transpose for a complex matrix (‘ in matlab. Use .’ instead for normal transpose)
2) Inner Product: normalized dot product of x and y: <x,y> = cos (angle between x and y)

3) Energy of x = <x,x>
4) Orthogonal: <x,y> = 0

Signal inner product – part 2

5) Harmonic sine waves are orthogonal


6) For dk[n] = ( 2 πkN n ), n,k,N are integers, 0≤n,k≤N-1
e
||dk||2 = sqrt (N)

Cauchy Schwarz Inequality


1. 0 , ≤ ¿
Creates an upper and lower boundary for the inner product of x and y
2. Forms the basis behind a matched filter

Infinite-Length Vectors
1. Hilbert Space/Banach Space: contains all infinite length vectors
2. sup|x[n]| is same as max, for infinite-length vector

Week 3: Discrete Time Systems

Systems
Linear systems

1) Matrix multiplications are all linear systems


2) All linear systems can be expressed as matrix multiplications

Time-invariant systems

3) H does not change with time

LTI (Linear Time-Invariant) systems

4) H is a Toeplitz matrix: entries on matrix diagonals are same


5)
6)

Matlab: circulent(h) = H

Week 4 – Convolution

System Impulse Response


1) Impulse response signal for LTI systems h: n=0 th column vector
Obtained when H acts on a delta function

Computing Convolution
1) Convolve h * x = y: find a vector containing all the inner products of h and x (function conv(h,y) in
matlab)
2) To convolve h with x,
Step one: plot x[-m] i.e. flip along y axis
Step two: shift the plot horizontally right by n>0, left by n<0
Step three: multiply plot with h plot and sum up all entries to obtain convolution result
3) Used in smoothing to average out noise: tradeoff between smoothing & bluring for longer impulse
response moving average filter (e.g. for an impulse response of length 3, take 3 values and average
them to get output)
4) Convolve h with x = convolve x with h i.e. compute output when input h is entered into a system x, or
the other way round aka swop h and x
5) Circular convolution for finite LTI systems: cconv(h,x,N) in matlab; where N=period. This is a zero-
padded convolution

Properties of convolution
1) Commutative
2) Cascade: x -> H1 -> H2 -> y i.e. x*H1 then *H2 to give y == x-> H1*H2->y i.e. H1*H2 then *x to give y
3) Parallel: (x-> H1) + (x->H2) = y == x -> (H1+H2)
4) An LTI system is causal if its impulse response is causal. Toeplitz system matrix is lower triangular
5) Duration
Support interval [N1,N2] : interval where x ≠ 0;
Duration of x: Dx = N2-N1+1
For y = x* h,
Max(Dy) = Dx + Dh – 1
6) An LTI system can have either finite (FIR) or infinite (IIR) impulse response
7) Circular convolution of x & h, both of which have finite duration signals D g & Dh but infinite length

System stability
1) BIBO stability: bounded input via h gives bounded output
An LTI system is BIBO stable if and only if h has finite L-1 norm
e.g. FIR systems

*For a concise revision, refer to revision lecture

You might also like