Lecture 1

Introduction
z Shannon demonstrated that by proper

encoding of information, errors introduced
by a noisy environment can be reduced to
any desired level without sacrificing
transmission rate, as long as transmission
rate is below capacity of channel.
z Since Shannon’s work, much research has
been done to find efficient encoding and
decoding methods.
Introduction (2)
z Transmission and storage of digital

information are two processes that transfer
data from an information source to a
destination.
Digital Source
m c
Information Source Channel Modulator/
Equivalent channel
source coder coder writer
Channel/
memory
Source m’ Channel c’ Demod./
Sink
Destination
decoder decoder reader
Types of Codes
z Two structurally different types of codes are

typically used:
z Block Codes
z Hamming
z BCH, RS
z LDPC
z Convolutional Codes
z Turbo codes typically use convolutional codes as
constituent codes
z TCM based on convolutional codes
Block Codes
z A block of k digital symbols is input to the encoder
from which n digital symbols are output (typically
n > k).
Block
encoder
k symbols n symbols
z Each k bit message sequence is mapped to one of

Mk possible codewords. Since there are Mn
possible M-ary sequences of length n, errors can be
detected if an invalid codeword is received.
Block Codes (2)
z Code rate R = k/n.

z The message sequence carries k symbols of
information.
z The codeword, which carries k symbols of
information, is made up of n symbols.
z There are (n-k) redundant symbols.
Convolutional Codes
zA convolutional code also produces n

symbols for k input symbols.
z However, output not only depends on current
k inputs but also on km previous inputs,
where m is the encoder memory.
z Encoder is implemented by a sequential
logic circuit.
Modulation and Coding
z Symbol rate = Rs, signaling interval = T = 1/Rs.

z For each symbol, the modulator selects a
waveform of duration T to represent the symbol to
be transmitted.
z Example BPSK:
2 Eb
s0 (t ) = cos(2πf c t + π ), 0 ≤ t ≤ T
T
2 Eb
s1 (t ) = cos(2πf c t ), 0 ≤ t ≤ T
T
Modulation and Coding (2)
z Transmitted signal is:
N
s (t ) = ∑ si (t − nT )
n =0
where i = 0,1,…,M-1 and is random (i = 0 or 1
for binary case).
z The received signal is:
r (t ) = a (t ) s (t − τ (t )) + n(t )
where a(t) is the time varying channel gain, τ(t) is the
delay introduced by the channel and n(t) is additive noise.
z AWGN Channel
z a(t) = a and τ(t) = τ.
z Flat Rayleigh Fading
z a(t) = time varying Rayleigh envelope
z τ(t) introduces time varying phase shift.
z Noise introduces detection errors at the

receiver.
z Error rate is function of Es/No.
z BER for BPSK in AWGN is:

⎛ 2 Eb ⎞
Pb = Q⎜ ⎟
⎝ No
⎠
z BER for BPSK in slow flat Rayleigh fading with

ideal channel phase compensation:
1⎡ γb ⎤
Pb = ⎢1 − ⎥
2 ⎢⎣ 1+ γ b ⎥⎦
z Coding increases symbol rate (k info bits

without coding, n code bits after coding).
z For same transmitted power, the code bit
energy is less than the uncoded bit energy Ec
= REb = (k/n)Eb.
z Therefore, the probability that a code bit is
incorrectly detected is higher than the
probability that an uncoded bit is incorrectly
detected.
z Coded data streams provide improved bit

error rates after decoding due to the error
correction capabilities of the code.
Example Hamming (7,4)
z 0000 0000000 1000 1000110

z 0001 0001101 1001 1001011
z 0010 0010111 1010 1010001
z 0011 0011010 1011 1011100
z 0100 0100011 1100 1100101
z 0101 0101110 1101 1101000
z 0110 0110100 1110 1110010
z 0111 0111001 1111 1111111
z Assume that we transmit 0000 in the uncoded case.

z If the first bit is incorrectly detected, we receive 1000,
which is a valid message.
z Assume that we transmit 0000000 in the uncoded
case
z If the first bit is detected in error, we receive 1000000,
which is not a valid codeword.
z Error has been detected.
z Codeword 0000000 differs from 1000000 in only one bit
position. All other codewords differ from 1000000 in at
least two positions.
z Assuming independent errors

z P(uncoded word error) = 1-(1-Pu)4.
z P(coded word error) = 1-(1-Pc)7-7Pc(1-Pc)6.
z In AWGN channel:
⎛ 2 Eb ⎞ ⎛ 2 ( 4 / 7 ) Eb ⎞
Pu = Q⎜ ⎟, Pc = Q⎜ ⎟
⎝ N0
⎠ ⎝ N0
⎠
Example Hamming (7,4) WER
Word Error Rate
1.00E+00
1.00E-01
WER
1.00E-02 uncoded
coded
1.00E-03
1.00E-04
0 1 2 3 4 5 6 7 8 9
Eb/No (dB)
Example Hamming (7,4) BER
z BER
z Uncoded Pb = Pu.
z Coded
Pb = 9 Pc2 (1 − Pc ) 5 + 19 Pc3 (1 − Pc ) 4 + 16 Pc4 (1 − Pc ) 3 +

12 Pc5 (1 − Pc ) 2 + 7 Pc6 (1 − Pc )1 + Pc7
Example Hamming (7,4) BER
Bit Error Rate
1.00E+00
1.00E-01
BER
1.00E-02
uncoded
coded
1.00E-03
1.00E-04
0 1 2 3 4 5 6 7 8 9
Eb/No (in dB)

Lecture 1

Uploaded by

Copyright:

Available Formats

Lecture 1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture 1

Uploaded by

Copyright:

Available Formats

Introduction

z Shannon demonstrated that by proper

z Transmission and storage of digital

z Two structurally different types of codes are

z Each k bit message sequence is mapped to one of

z Code rate R = k/n.

zA convolutional code also produces n

z Symbol rate = Rs, signaling interval = T = 1/Rs.

z Noise introduces detection errors at the

z BER for BPSK in AWGN is:

z BER for BPSK in slow flat Rayleigh fading with

z Coding increases symbol rate (k info bits

z Coded data streams provide improved bit

z 0000 0000000 1000 1000110

z Assume that we transmit 0000 in the uncoded case.

z Assuming independent errors

Pb = 9 Pc2 (1 − Pc ) 5 + 19 Pc3 (1 − Pc ) 4 + 16 Pc4 (1 − Pc ) 3 +

You might also like