[go: up one dir, main page]
Scribd
Upload
776 views18 pages

Maximum Likelihood Decoding Techniques Notes

Maximum likelihood decoding is used to determine the most likely transmitted codeword based on the received signal. It can be applied to both hard decision channels like the binary symmetric channel and soft decision channels like the Gaussian channel. The Viterbi algorithm performs maximum likelihood decoding with reduced complexity by only considering the most likely trellis paths at each step and eliminating unlikely paths. It works by calculating a metric to measure the similarity between received signals and trellis paths, retaining the best path entering each state at each time step.

Uploaded by

Vidhya Nair
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
776 views18 pages

Maximum Likelihood Decoding Techniques Notes

Maximum likelihood decoding is used to determine the most likely transmitted codeword based on the received signal. It can be applied to both hard decision channels like the binary symmetric channel and soft decision channels like the Gaussian channel. The Viterbi algorithm performs maximum likelihood decoding with reduced complexity by only considering the most likely trellis paths at each step and eliminating unlikely paths. It works by calculating a metric to measure the similarity between received signals and trellis paths, retaining the best path entering each state at each time step.

Uploaded by

Vidhya Nair
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 18

Maximum Likelihood Decoding

Unit 4

Channel Models: Hard vs Soft decisions


Binary Symmetric Channel (BSC)
- Discrete memoryless channel, binary i/p & o/p and symmetric transition probabilities
P(0 |1) P(1|1) P(1| 0) p

P(0 | 0) 1 p

- hard decision channel - U(m) is chosen closest in Hamming distance to Z - From U(m) , U(m) is chosen for which distance to Z is minimum

MDCT Unit 4: ML Decoding

Channel Models: Hard vs Soft decisions


Gaussian Channel
-Each demodulator o/p symbol is a value from continuous
alphabet -Symbol cannot be labeled as a correct or incorrect detection decision -Maximizing P(Z|U(m)) is equiv. to maximizing inner product between codeword sequence U(m) and Z
n

-Decoder chooses U(m) if it maximizes


i 1 j 1

z ji u (jim )

-Equivalent to choosing U(m) closest in Euclidean distance to Z -Soft decision channel


MDCT Unit 4: ML Decoding

Viterbi Decoding Algorithm


Performs ML decoding with reduced computational load special structure in code trellis Calculates measure of similarity or distance received signal (ti) and all trellis paths entering each state (ti) Trellis paths which could not be the candidate for ML choices are not considered Surviving path - The path with best metric is chosen when two paths enter the same state, and performed for all states - least likely paths are eliminated Optimum path: expressed choosing codeword with maximum likelihood metric or minimum distance metric Advantage: Complexity is not a function of no. of symbols in codeword sequence
MDCT Unit 4: ML Decoding

Convolutional Encoder (rate K=3)

MDCT Unit 4: ML Decoding

Encoder Trellis Diagram (rate K=3)

MDCT Unit 4: ML Decoding

Decoder Trellis Diagram (rate K=3)

MDCT Unit 4: ML Decoding

The basis
Any two paths merge to a single state, one path is eliminated in search of an optimum path

Path metrics for two merging paths

MDCT Unit 4: ML Decoding

Decoding Principle
At each time ti, 2K-1 states are present in trellis and can be entered by means of two paths Decoding computes the metric for two paths entering each state and eliminating one of them Computation is done for each of the 2K-1 states at time ti and decoder moves to time ti+1 and the process is repeated At any time the winning path metric for each state is termed as state metric for that state at that time
MDCT Unit 4: ML Decoding

Selection of Survivor paths

Survivors at t2
MDCT Unit 4: ML Decoding

Survivors at t3

Selection of Survivor paths (Contd.)

Metric comparisons at t4

Survivors at t4 Only one surviving path between time t1 and t2 and it is termed as common stem Transition occurred between 0010 and since it is due to input bit 1, decoder output is 1
MDCT Unit 4: ML Decoding

Selection of Survivor paths (Contd.)

Metric comparisons at t5
MDCT Unit 4: ML Decoding

Survivors at t5

Selection of Survivor paths (Contd.)

Metric comparisons at t6

Survivors at t6

MDCT Unit 4: ML Decoding

Sequential Decoding
Proposed by Wozencraft and modified by Fano Generates hypothesis about transmitted codeword sequence and computes metric between these hypothesis and received signal Goes forward still metric indicates its choices are likely; else goes backward changing hypothesis still finding a likely one Can work with both hard and soft decisions, however soft decisions are normally avoided since large storage elements are used and also complexity

MDCT Unit 4: ML Decoding

Convolutional Encoder (rate K=3)

MDCT Unit 4: ML Decoding

Tree Diagram (rate K=3)

MDCT Unit 4: ML Decoding

MDCT Unit 4: ML Decoding

Sequential Decoding Example

MDCT Unit 4: ML Decoding

You might also like