Scribd Logo
Upload

Welcome to Scribd!

  • 14 views

    Lecture 2

    Uploaded by

    elmzyonasara
    The document discusses different techniques for decoding received data in communication systems, including hard decision, soft decision, and maximum likelihood decoding. It covers topics like Hamming distance, Euclidean distance, decision variables, error models, and performance measures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Lecture 2

    Uploaded by

    elmzyonasara
    14 views19 pages
    The document discusses different techniques for decoding received data in communication systems, including hard decision, soft decision, and maximum likelihood decoding. It covers topics like Hamming distance, Euclidean distance, decision variables, error models, and performance measures.

    Original Description:

    Error Control Coding

    Original Title

    Lecture2

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses different techniques for decoding received data in communication systems, including hard decision, soft decision, and maximum likelihood decoding. It covers topics like Hamming distance, Euclidean distance, decision variables, error models, and performance measures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    14 views19 pages

    Lecture 2

    Uploaded by

    elmzyonasara
    The document discusses different techniques for decoding received data in communication systems, including hard decision, soft decision, and maximum likelihood decoding. It covers topics like Hamming distance, Euclidean distance, decision variables, error models, and performance measures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 19

    Decoding Techniques

    „ Hard Decision
    Receiver detects data before decoding
    „ Soft Decision
    Receiver quantizes received data and decoder uses
    likelihood information to decode.
    Magnitude of decision variable usually indicates
    the likelihood that the detection is correct.
    Maximum Likelihood Decoding (MLD).
    Hard Decision vs Soft Decision
    0’
    0 0 0
    0
    1 1 1
    1
    1’

    Hard Decision Soft Decision (2 bit quantization)


    Maximum Likelihood Decoding
    „ Transmit codeword c
    „ Receive word r.
    „ Decode c’.
    „ Decoding error occurs if c’ ≠ c.
    Maximum Likelihood Decoding (2)
    P( E | r ) = P(c' ≠ c | r )

    P( E ) = ∑ P (c' ≠ c | r ) P (r )
    r

    The optimum decoding rule is one that minimizes P(E).


    P(r) is independent of the decoding rule, therefore we must
    minimize P(c’≠c|r), which is equivalent to maximizing
    P(c’=c|r).
    Maximum Likelihood Decoding (3)
    P(c | r ) = P (r | c) P (c) / P(r )
    Assuming all codewords are equally likely, maximizing P(c|r)
    is the same as maximizing P(r|c). Assuming hard decision
    with discrete memoryless channel (DMC):

    P (r | c ) = ∏ P ( ri | c i ) = p d ( r , c ) (1 − p ) n − d ( r , c ) (1)
    i
    log P ( r | c ) = d ( r , c ) log p + ( n − d ( r , c )) log( 1 − p ) (2)

    Since p < (1-p), P(r|c) is maximized by the codeword for


    which d(r,c) is minimized. This is known as a minimum
    distance decoding rule.
    Hamming Distance vs Euclidean
    Distance
    „ Hamming Distance = number of positions in which
    two vectors differ.
    „ Hard decision decoding decodes using minimum
    Hamming distance rule previously shown
    „ Euclidean distance between r and c is ||r-c||
    „ Soft decision decoding uses minimum Euclidean
    distance (approximately)
    Decision Variables

    fri(ri|-1) fri(ri|1)

    -1 0 +1 ri

    P(ci=1)/P(ci=-1)=fri(ri|ci=1)/fri(ri|ci=-1)
    Example: HD vs SD
    „ Consider Hamming (7,4) code used with the
    following channels
    0’
    0 P(0’|0)=P(1’|1)=0.6
    0 0
    0 P(0|0)=P(1|1)=0.3
    1 P(1|0)=P(0|1)=0.099
    1 1
    1 P(1’|0)=P(0’|1)=0.001
    P(0|0)=P(1|1)=0.9
    1’
    P(1|0)=P(0|1)=0.1

    (a) HD (b) SD
    Example: HD vs SD
    „ Suppose we receive r = 1’ 0 0’ 0 0’ 0’ 0’
    „ For HD, there is no quantization, so r = 1000000 and
    will be decoded as 0000000 using minimum
    Hamming distance rule.
    „ In the SD case, using (1) with c = 0000000, we get
    P(r|c) = 0.000117
    „ However, for c=1101000, we get P(r|c) = 0.000762
    „ This means that for the given r, it is almost 7 times
    more probable that 1101000 was transmitted than
    0000000.
    Errors and Channel Models
    „ Memoryless channels: Noise affects each transmitted
    symbol independently.
    Tx symbol has probability p of being received incorrectly
    and probability 1-p of being received correctly.
    Transmission errors occur randomly in the received
    sequence.
    Memoryless channels are often referred to as random-error
    channels.
    Errors and Channel Models (2)
    „ Examples
    AWGN: ri = si+ni, E[ni]=0, E[ni2]=σn2 and
    E[ninj]=0 for i≠j.
    DMC: P[0|0]
    0 0
    P[1|0]

    P[0|1]

    1 P[1|1]
    1
    Errors and Channel Models (3)
    „ Channels with memory
    Errors do not occur randomly.
    Either noise is not independent from
    transmission to transmission (coloured noise)
    Or slow time varying signal to noise ratio
    causes time dependent error rates (fading
    channels).
    Errors and Channel Models (4)
    „ Gilbert and Fritchman model
    q1

    Good Bad
    1-q1 state State 1-q2

    q2
    Errors and Channel Models (5)
    „ Channels with memory lead to error
    bursts.
    Burst-error correcting codes
    Random error correcting codes with
    interleaving-deinterleaving to randomize
    errors.
    Performance Measures
    „ Probability of decoding error P(E).
    Probability that codeword at output of decoder is not the
    transmitted one.
    Also referred to as word error rate (WER) or Block error
    rate (BLER).
    „ Bit error rate (BER) Pb
    Probability that message bit at output of decoder is
    incorrect.
    „ Coding Gain (measured in dB)
    Savings in transmitted power to achieve a specific BER
    using coding compared to uncoded case
    Performance Measures 2
    „ Asymptotic coding gain
    Coding gain when Eb/No → ∞
    Performance Measures 3
    1.00E-02

    1.00E-03

    1.00E-04

    1.00E-05
    BER

    1.00E-06

    1.00E-07 uncoded

    coded
    1.00E-08

    1.00E-09

    1.00E-10
    6 7 8 9 10 11 12 13
    Eb/No (in dB)
    Coded Modulation
    „ Use of ECC creates bandwidth expansion due to
    redundant symbols.
    „ Combining ECC and modulation allows the
    redundancy to be contained in the modulation
    1011
    s1(t)+s0(t-T)+s1(t-2T)+s1(t-3T)
    s1(t)+s2(t-T)+s1(t-2T)+s3(t-3T)
    „ Memory is created without adding redundant bits
    by using a higher order modulation scheme and
    using a bit in two successive symbols.
    Trellis Coded Modulation
    „ State machine adds redundant bits and
    creates memory
    „ State change is encoded by selecting a
    symbol from a larger than needed
    constellation, thus no bandwidth
    expansion occurs and significant coding
    gains are achieved.

    You might also like