Data Compression Using Elias Delta Code

DOI: 10.23883/IJRTER.2017.3406.TEGS6 210
Data Compression Using Elias Delta Code
Leni Marlina1
, Andysah Putera Utama Siahaan2
, Heri Kurniawan3
, Indri Sulistianingsih4
Faculty of Computer Science, Universitas Pembangunan Panca Budi, Medan, Indonesia
2Ph.D. Student of School of Computer and Communication Engineering, Universiti Malaysia Perlis, Kangar, Malaysia
Abstract — Compression is an activity performed to reduce its size into smaller than earlier.
Compression is created since lack of adequate storage capacity. Data compression is also needed to
speed up data transmission activity between computer networks. Compression has the different rule
between speed and density. Compressed compression will take longer than compression that relies on
speed. Elias Delta is one of the lossless compression techniques that can compress the characters. This
compression is created based on the frequency of the character of a character on a document to be
compressed. It works based on bit deductions on seven or eight bits. The most common characters will
have the least number of bits, while the fewest characters will have the longest number of bits. The
formation of character sets serves to eliminate double characters in the calculation of the number of
each character as well as for the compression table storage. It has a good level of comparison between
before and after compression. The speed of compression and decompression process possessed by this
method is outstanding and fast.
Keywords — Compression, Elias Delta
I. INTRODUCTION
The development of information at this time trigger the creators of storage media to develop
technology on a harddisk, flash, CD, and others. It is due to the lack of capacity provided by certain
media. As well as dual-screen DVDs that can only hold data of approximately 9 GB. Thus, created a
new storage media such as blue ray disk that can store data larger than 30 GB. A common problem
with data compression is how to optimally compress the data, especially text data that does not have a
header file. It aims to obtain the compression results have a size that is much smaller than the original
size. However, to make the compression process is not easy. Compression must be done correctly to
the original information. If there is an error in the preparation of the bit sequence, then during the
decompression process will fail. To support all that, it has to be seen the kind of compression that is
suitable to apply to certain data.
Compression has two types, lossy and lossless techniques [2]. Lossy compression is a non-
returnable compression to all forms. The discharges that occur in this compression can not return to
the original data or in other words; this compression process has no decompression. Processes like this
are often done to shrink the size of video, audio, images. This compression is done because of
constrained storage media, especially on mobile phones or constrained in mobile phone resolution to
display or listen to the media. The second technique is lossless compression. It is a common data
compression for archiving where compression results can be returned to the decompression process to
get the original data. The Elias Delta method is a lossless compression in the process of compacting
the data. By applying this method to shrink the data, it is expected that the burden of storage media
can be reduced.

International Journal of Recent Trends in Engineering & Research (IJRTER)
Volume 03, Issue 08; August - 2017 [ISSN: 2455-1457]
@IJRTER-2017, All Rights Reserved 211
II. THEORIES
2.1 Compression
The rapid development of information requires that technology in all fields should flourish as
well. It continually seeks to evolve to meet the needs of the community for information that can be
quickly and easily accessed. To do this requires a large database. It deals with storage media. It also
relates to data access process. If the data has a large capacity, it will slow down the access speed and
reduce the capacity of the storage media quickly [7][8]. This problem makes developers of information
technology perform data compression. The original data will be converted into smaller data but does
not reduce the value contained in the information.
Compression is a technique of information reduction so that new data is obtained with a size
smaller than the size before it is compressed [8]. It can be done on some types of data, video, images,
audio, and others. The data in the image is a file with BMP, JPG, PNG, TIFF, BMP extension. Audio
data is in the form of MP3, AAC, RMA, WMA and video in the form of MP4, WMV, AVI and so on.
The compression process is the process of encoding information using bits or other information-
bearing units lower than the data representation that is not encoded in the encoding system [3].
The compression technique has two ways:
 Lossy Compression. This technique is used to degrade the quality of the original media.
Compressed data will remain equally visible or enjoyed like the original but with lower quality. It
is more often used for streaming media, audio, images and video. The resulting file size will be
smaller than the original but still be eligible to apply. This process works by removing unused
portions of data. These data are not necessary for the sight of the human eye because it is not
directly related to the user. So when the data has been compressed, the data can still be used even
though it is not like the original. For example, a video that has a full HD 1920 x 1650 resolution.
For this size, it will affect the speed at the time of video playback on the old computer, so that
requires the reduction of resolution to HD 1280 x 720 or VGA 640 x 480.
 Lossless Compression. This compression is often done to save data storage or to combine some
data into one data so as not to be scattered between one another. It is often done when sending
emails with many attachments that minimize the attachment process. The compressed data can be
returned to all forms. The result of the decompression process is no different from the original. An
example can be seen in the WinRAR app. File compression results in the form of ZIP, RAR, GZIP,
7-Zip and so forth. The images, audio, video files can also be compressed with this method just
like on lossy compression. However, the purpose is different, not to lower the quality or resolution,
but to perform data archiving. The file size after the compression process is not necessarily smaller
than the original file if the data is saved optimally.
2.2 Elias Delta
Peter Elias invented Elias Delta. This code applies methods such as Elias Gamma, especially
in the head [1][4]. The technique is as follows:
 Find the highest binary rank, for example, 11 which has a binary value of 1011 where the highest
rank is 3. So N '= 3.
 Use Gamma Coding to encode N numbers where N = N '+ 1. So for the case of decimal 11 then
we have to make a Gamma Coding of 4 that is 00100.
 Add the remaining N 'binary in result no.2. So obtained answer 00100011.

 Next is the Elias Delta Coding method, the principle is the opposite of steps one through three
above. Suppose we will decode 00100011.
 Find the number of zeros before it finds the number one, which is 00, amounts to two. Means there
are (2 + 1) numbers to watch after these two zeros are 100 which in decimals mean 4, so we get N
'with N-1 = 4-1 = 3.
 If N 'is known, i.e. 3 then three remaining bits are part of that number i.e. 011. So obtained answer
1011 which means 11.
Table 1. Elias Delta encoding system
Number N N+1 Encoding Probability
1 = 20
0 1 1 1/2
2 = 21
+ 0 1 2 0 1 0 0 1/16
3 = 21
+ 1 1 2 0 1 0 1 1/16
4 = 22
+ 0 2 3 0 1 1 00 1/32
5 = 22
+ 1 2 3 0 1 1 01 1/32
6 = 22
+ 2 2 3 0 1 1 10 1/32
7 = 22
+ 3 2 3 0 1 1 11 1/32
8 = 23
+ 0 3 4 00 1 00 000 1/256
9 = 23
+ 1 3 4 00 1 00 001 1/256
10 = 23
+ 2 3 4 00 1 00 010 1/256
11 = 23
+ 3 3 4 00 1 00 011 1/256
12 = 23
+ 4 3 4 00 1 00 100 1/256
13 = 23
+ 5 3 4 00 1 00 101 1/256
14 = 23
+ 6 3 4 00 1 00 110 1/256
15 = 23
+ 7 3 4 00 1 00 111 1/256
16 = 24
+ 0 4 5 00 1 01 0000 1/512
17 = 24
+ 1 4 5 00 1 01 0001 1/512
18 = 24
+ 2 4 5 00 1 01 0010 1/512
Table 1 shows the encoding system of Elias Delta [5][6]. As an understanding, it can be seen
an example, if n = 18 produces the code delta 001010010. It is obtained from the following
calculations:
 Take the closest the power of two, 24
= 16
 Add the power value 4 and 1 = 5 so that the number of characters in the first part is 00101
 The rest of the earlier power is 18 - 16 = 2
 Add the beginning part with bit value as much as four characters = 0010
 The result of the Elias Delta Code is 001010010

III. RESULT AND DISCUSSION
At this stage, compression testing is performed for a simple string. It is "THE QUICK
BROWN FOX JUMPS OVER THE LAZY DOG." Stages in this section include:
 Calculate the length of the text
 Create and calculate the length of the character set
 Calculating bit length
 Specifies padding bits if necessary
Table 2 shows the original string before the compression process. The string has 43 characters.
Tabel 2. Original string
No. Char ASCII Binary
1 T 84 01010100
2 H 72 01001000
3 E 69 01000101
4 32 00100000
5 Q 81 01010001
6 U 85 01010101
7 I 73 01001001
8 C 67 01000011
9 K 75 01001011
10 32 00100000
11 B 66 01000010
12 R 82 01010010
13 O 79 01001111
14 W 87 01010111
15 N 78 01001110
16 32 00100000
17 F 70 01000110
18 O 79 01001111
19 X 88 01011000
20 32 00100000
21 J 74 01001010
22 U 85 01010101
23 M 77 01001101
24 P 80 01010000
25 S 83 01010011
26 32 00100000
27 O 79 01001111
28 V 86 01010110
29 E 69 01000101
30 R 82 01010010
31 32 00100000
32 T 84 01010100
33 H 72 01001000

34 E 69 01000101
35 32 00100000
36 L 76 01001100
37 A 65 01000001
38 Z 90 01011010
39 Y 89 01011001
40 32 00100000
41 D 68 01000100
42 O 79 01001111
43 G 71 01000111
The next stage is the formation of character sets. Repeated characters will be eliminated so that
only single characters live. After this process, there are only 27 characters left with different frequency
of occurrences. The number of bits in this string is 344. The result of the character set process can be
seen in Table 3.
Table 3. Elias Delta characters set
No. Char ASCII Binary Freq. Bits
1 T 84 01010100 2 16
2 H 72 01001000 2 16
3 E 69 01000101 3 24
4 32 00100000 8 64
5 Q 81 01010001 1 8
6 U 85 01010101 2 16
7 I 73 01001001 1 8
8 C 67 01000011 1 8
9 K 75 01001011 1 8
10 B 66 01000010 1 8
11 R 82 01010010 2 16
12 O 79 01001111 4 32
13 W 87 01010111 1 8
14 N 78 01001110 1 8
15 F 70 01000110 1 8
16 X 88 01011000 1 8
17 J 74 01001010 1 8
18 M 77 01001101 1 8
19 P 80 01010000 1 8
20 S 83 01010011 1 8
21 V 86 01010110 1 8
22 L 76 01001100 1 8
23 A 65 01000001 1 8
24 Z 90 01011010 1 8
25 Y 89 01011001 1 8
26 D 68 01000100 1 8
27 G 71 01000111 1 8
344

According to the arrangement of bits of the characters in the Elias Delta table, the result of
compression of the previous string is 248 as seen in Tabel 4. This result is obtained based on the use
of bits according to the frequency of appearance of the characters. The space character is the one that
most often appears in the string; There are eight occasions. There is a character that appears three and
four times. Four characters appear twice and 20 characters each appearing just once.
Table 4. Elias Delta result
No. Char Freq. Binary Length Bits
1 8 1 1 8
2 O 4 0100 4 16
3 E 3 0101 4 12
4 H 2 01100 5 10
5 U 2 01101 5 10
6 R 2 01110 5 10
7 T 2 01111 5 10
8 C 1 00100000 8 8
9 K 1 00100001 8 8
10 B 1 00100010 8 8
11 Q 1 00100011 8 8
12 I 1 00100100 8 8
13 W 1 00100101 8 8
14 N 1 00100110 8 8
15 F 1 00100111 8 8
16 X 1 001010000 9 9
17 J 1 001010001 9 9
18 M 1 001010010 9 9
19 P 1 001010011 9 9
20 S 1 001010100 9 9
21 V 1 001010101 9 9
22 L 1 001010110 9 9
23 A 1 001010111 9 9
24 Z 1 001011000 9 9
25 Y 1 001011001 9 9
26 D 1 001011010 9 9
27 G 1 001011011 9 9
248
Bit Sequence:
011110110001011001000110110100100100001000000010000110010001001110010000100101001
001101001001110100001010000100101000101101001010010001010011001010100101000010101
010101011101011110110001011001010110001010111001011000001011001100101101001000010
1101100110000

TB = 248
TC =
𝑇𝐵
8
=
248
8
= 31
Padding = 𝑇𝐵 𝑚𝑜𝑑 8
= 248 𝑚𝑜𝑑 8
= 0
Total Bits = 248 + 8
= 256
The bold character at the end of the bit sequence is the number of padding bits. For the above
calculation, it is obtained that Padding = 0. Character "0" is on the order of 48 in ASCII code which
after converted to binary will produce 00110000. This binary number will be added to the compression
result, so the binary number becomes 248 + 8 = 256 bits. After the 256 bit is converted to a sequence
of characters, this will generate the string {? FÒB ????????? N ?? E ¥ "?? R UuìYX®X, ËH [0 as
many as 32 characters.
Compression Ratio =
𝑎𝑓𝑡𝑒𝑟 𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛
𝑏𝑒𝑓𝑜𝑟𝑒 𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛
∗ 100%
=
256
344
∗ 100%
= 74.41860465116279%
Redudancy =
𝑏𝑒𝑓𝑜𝑟𝑒 −𝑎𝑓𝑡𝑒𝑟 𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛
𝑏𝑒𝑓𝑜𝑟𝑒 𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛
∗ 100%
=
344−256
344
∗ 100%
=
88
344
∗ 100%
= 25.58139534883721%
The compression process has saved data of 25.58139534883721% of the original data. The
savings rate depends on the order and character pattern of the original message.
IV. CONCLUSION
The above calculations prove that the compression made by Elias Delta has been very good at
saving storage capacity. The Elias Delta method can compress to all data types. The data will be
converted into rows of character sets and then sorted by the frequency of appearance of characters.
The most frequently displayed characters will vary by a document. The language also determines the
number of most frequently used characters. Application of this compression is very well done to save
storage media and speed up the process of sending data on a computer network. The downside of this
compression is the decompression process. This compression requires additional tables to match the
order of characters that often appear. Returning characters will take a long time because it must match
the pieces of bits one by one as much as the data stored in the compression table.
REFERENCES
[1] R. T. Handayanto, “Elias Gamma & Delta Coding,” 15 September 2014. [Online]. Available:
https://rahmadya.com/2014/09/15/elias-gamma-delta-coding/. [Diakses 24 August 2017].
[2] S. D. Nasution dan Mesran, “Goldbach Codes Algorithm for Text Compression,” International Journal of Software
& Hardware Research in Engineering, vol. 4, no. 11, pp. 43-46, 2016.

[3] Ameliachy, “Teknik Kompresi Data,” 20 April 2012. [Online]. Available: https://ameliachy.wordpress.com/
2012/04/20/teknik-kompresi-data/. [Diakses 24 August 2017].
[4] Antoni, E. B. Nababan dan M. Zarlis, “Results Analysis of Text Data Compression on Elias Gamma Code, Elias Delta
Code and Levenstein Code,” International Journal of Science and Advanced Technology, vol. 4, no. 9, pp. 17-24,
2014.
[5] Wikipedia, “Elias Delta Coding,” [Online]. Available: https://en.wikipedia.org/wiki/Elias_delta_coding.
[6] P. Elias, “Universal Codeword Sets and Representations of the Integers,” IEEE Transactions on Information Theory,
vol. 21, no. 2, p. 194–203, 1975.
[7] Suherman dan A. P. U. Siahaan, “Huffman Text Compression Technique,” SSRG International Journal of Computer
Science and Enginee ring, vol. 3, no. 8, pp. 103-108, 2016.
[8] R. Aarthi, D. Muralidharan dan P. Swanminathan, “Doubel Compression of Test Data Using Huffman Code,” Journal
of Theoretical and Applied Information Technology, vol. 39, no. 2, pp. 104-113, 2012.

Data Compression Using Elias Delta Code

Related slideshows

More Related Content

Data Compression Using Elias Delta Code