Scribd
Upload
319 views

Problem Source Coding...

1. To achieve 100% coding efficiency, the code word length must satisfy the Kraft inequality for a discrete memoryless source with K symbols. 2. Codes I and III are prefix codes according to the Kraft-McMillan inequality. 3. The Huffman code for a 7-symbol source with specified probabilities achieves 100% efficiency by moving combined symbols higher in the code tree.

Uploaded by

Manoj Kumar
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
319 views

Problem Source Coding...

1. To achieve 100% coding efficiency, the code word length must satisfy the Kraft inequality for a discrete memoryless source with K symbols. 2. Codes I and III are prefix codes according to the Kraft-McMillan inequality. 3. The Huffman code for a 7-symbol source with specified probabilities achieves 100% efficiency by moving combined symbols higher in the code tree.

Uploaded by

Manoj Kumar
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Advanced Digital Communication Source Coding 1. Consider a discrete memoryless source whose alphabet consists of K equiprobable symbols.

What conditions have to be satisfied by K and the code-word length for the coding to be 100 percent? [Haykin 9.9b] 2. Consider the four codes listed below: Symbol s0
s1 s2

Code I 0 10 110 1110 1111

Code II 0 01 001 0010 0011

Code III 0 01 011 110 111

Code IV 00 01 10 110 111

s3
s4

(a) Two of these four codes are prefix codes. Identify them, and construct their individual decision trees. (b) Apply the Kraft-McMillan inequality to codes I, II, III, and IV. Discuss your results in light of those obtained in part (a). [Haykin 9.10] 3. A discrete memoryless source has an alphabet of seven symbols whose probabilities of occurrence are as described here: Symbol Probability s0 0.25
s1 s2

s3 0.125

s4

s5 0.0625

s6 0.0625

0.25

0.125

0.125

Compute the Huffman code for this source, moving a "combined" symbol as high as possible. Explain why the computed source code has an efficiency of 100 percent. [Haykin 9.12] 4. Consider a discrete memoryless source with alphabet {so , s1, s2 } and statistics {0.7,0.15,0.15} for its output. (a) Apply the Huffman algorithm to this source. Hence, show that the average code-word length of the Huffman code equals 1.3 bits/symbol. (b) Let the source be extended to order two. Apply the Huffman algorithm to the resulting extended source, and show that the average code-word length of the new code equals 1.1975 bits/symbol. (c) Compare the average code-word length calculated in part (b) with the entropy of the original source. [Haykin 9.13]

5. A computer executes four instructions that are designated by the code words (00, 01, 10, 11). Assuming that the instructions are used independently with probabilities (1/2. 1/8, 1/8. 1/4), calculate the percentage by which the number of bits used for the instructions may be reduced by use of an optimum source code. Construct a Huffman code to realize the reduction. [Haykin 9.15] 6. Consider the following binary sequence 11101001100010110100 Use the Lempel-Ziv algorithm to encode this sequence. Assume that the binary symbols 0 and 1 are already in the codebook.

Solution 1. Source entropy is

For 100% efficiency, we have

To satisfy this equation, we choose

2. (a)

3.

4. (a)

The average codeword length Lav = pi Li = 0.7 1 + ... = 1.3 bits/symbol


i =0

(b) For the extended source, we have p ( s 0 s 0 ) = 0.7 0.7 = 0.49, p ( s 0 s1 ) = 0.105, p ( s 0 s 2 ) = 0.105, p ( s1 s 0 ) = 0.105, p( s1 s1 ) = 0.0225, p( s1 s 2 ) = 0.0225, p ( s 2 s 0 ) = 0.105, p( s 2 s1 ) = 0.0225, p( s 2 s 2 ) = 0.0225 Applying the Huffman algorithm to the extended source, we obtain the following source code:

s0 s0 : 1 s 0 s1 : 001 s 0 s 2 : 010 s1 s 0 : 011 s1 s1 : 000100 s1 s 2 : 000101 s 2 s 0 : 0000 s 2 s1 : 000110 s 2 s 2 : 000111


The average codeword length Lav = 0.49 1 + ... = 2.395 bits/symbol Lav / 2 = 1.1975 bits/symbol (c) source entropy H ( X ) = 0.7 log 2 0.7 0.15 log 2 0.15 0.15 log 2 0.15 = 1.18 bits/symbol
Lav 1 H(X ) + n n

Therefore, H ( X )

5.

6.

You might also like