25
1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi Arabia http://faculty.uoh.edu.sa/csse/smisbah

1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

Embed Size (px)

Citation preview

Page 1: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

1

Improving VoIP Transfer Rate over Internet

Syed Misbahuddin, Dr. Engg.

Department of Computer Science and Software Engineering

University of Hail, Saudi Arabia

http://faculty.uoh.edu.sa/csse/smisbah

Page 2: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

2

Presentation outline

• Research Objective• Overview of Voice Over IP (VoIP) • Compression algorithm applied to VoIP• Problems of compression algorithms• The proposed algorithm for Reducing Web

delays in VoIP• Conclusion

Page 3: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

3

Research Objective

To investigate an algorithm to Improve VoIP Traffic rate over Internet

Page 4: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

4

What is Voice Over IP (VoIP)

Using Internet to transfer voice signals after converting them into IP packets

Page 5: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

5

Why VoIP

voice communication with no or minimal cost through the Internet backbone

Page 6: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

6

Digitization of voice signal and 8 bit binary code assignment to each digitized sample

Voice signal

The sampled voice signals are coded into 64000 bits per sec stream

The G.729 compression algorithm compresses 64 kb/s to 8 kb/s

Compressed voice stream is converted into voice frames of 10 mill sec long carrying 80 bits

Generation of Individual Voice Frames in VoIP

Page 7: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

7

Transmission of Voice Frame as IP Packet

Internet

Transmitting end

IP Header(20 ) UDP Header(8) RTP Header(12) Voice Frame(10)

IP Header(20 ) UDP Header(8) RTP Header(12) Voice Frame(10)

IP Header(20 ) UDP Header(8) RTP Header(12) Voice Frame(10)

Note: Numbers in each field shows size of field in bytes

Page 8: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

8

QoS in VoIP

• The Quality of Services (QoS) in VoIP technology is related to the short delivery time of the voice data over the Internet

• To achieve better QoS, data compression algorithms are applied in VoIP systems.

• Standard voice compression algorithms used are: ITU’s G. 723 and G.729

Page 9: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

9

Main Problem with Compression Algorithms

• The compression algorithms reduce voice quality• Better compression algorithms are constantly

being investigated to maintain the voice qualities.

Page 10: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

10

Justification of Proposing a Data Reduction Algorithm for VoIP

• The voice signal is analog signal which varies slowly in time.

• If the voice signal is sampled at relatively high rate, the equivalent digital data will have repeated values in a short time window.

Page 11: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

11

Assumptions

• The voice signals should be sampled at the rate of 16000 samples per second

• each voice sample is assigned a 16 bit code• The sampled voice signals are coded into 64 kb/s

bit stream• A bit stream of 256 kb/s is divided into voice

frames of one milli second duration carrying 256 bits

• Individual voice frames of 1 m sec can be broken into 4 groups of 64 bits of ¼ m sec

Page 12: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

12

Sub grouping of voice Frame

One Voice Frame of 1 m sec duration of 256 bits

64 bits 64 bits 64 bits 64 bits

Page 13: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

13

The Algorithm• Store a copy of recently sent voice frame (of 1 m sec) in a

buffer called VFT_BUFF• Before sending next voice frame, compare contents of

subgroups with the content of most recently sent subgroups

• If in each subgroup some bytes are repeated then produce a 8 bit compression code for each subgroup

• In compression code the ith bit=1 if ith byte in subgroup is repeated otherwise ith bit=0 if ith byte is new

• Include compression code as first byte in each subgroup• Include non-repeated bytes in each subgroup

• Send modified voice frame for further processing.

Page 14: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

14

Subgroup of voice frame of 64 bits

Modified subgroup of voice Frame with compression code (CC)

CC Non Repeated bytes

8 bits 0 to 7 bytes

Size of modified subgroup=1 byte to 8 bytes

Page 15: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

15

Example:

Assume 4 initial bytes in each subgroup are repeated and 4 bytes are new.

1 1 1 1 0 0 0 0 4 Non-Repeated bytes

Size of modified Subgroup=5 bytes

Compression code

Page 16: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

16

Impact of Algorithm on complete voice frame

If all bytes in a voice frame are repeated then a 256 bits long voice frame is represented

by only 64 bits

Page 17: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

17

Voice frame Reconstruction at receiving end of VoIP system

• The receiving end stores a copy of most recently received voice frame in a buffer called VFR_BUFF

• When the receiving system receives another voice frame with compression codes in each subgroup then it retrieve the repeated bytes from VFR_BUFF and non-repeated bytes from received voice frame

Page 18: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

18

Indication of presence of compression codes in voice frame

Two undefined values in Pay Load Type fieldin RTP header may be used to indicate

the presence and absence of compression code in a voice frame

Page 19: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

19

Overview of RTP

• Provides end-to-end delivery services for real-time traffic: interactive audio and video

• Primarily designed to support multiparty multimedia conferences, typically assumes IP multicast.

Page 20: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

20

RTP Header

Page 21: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

21

Pay Load Type Field in RTP Header

Page 22: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

22

PT Values in RTP Header for indicating the presence and absence of compression code in

Voice Frame

• PT =16 Normal Voice Frame

• PT=17 Voice Frame containing compression codes

Page 23: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

23

Summary of Data Reduction AlgorithmBegin

Very first Voice frame

Transmitting End

Yes

Save copy in VF_TBUF

Add RTP, UDP and IP

Send VF to Internet

Obtain VF

No

Compare VF with VF_TBUF

Repetition in VF

NoUpdate

VF_TBUF

Append compression code in each subgroup in VF. Modify PT in RTP

Yes

Update VF_TBUF

Page 24: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

24

Summary of Data Reduction Algorithm

Receiving End Begin

Obtain VF

Very first Voice frame

YesSave copy in VF_RBUF

Process VF

No

CC in VF

Yes

No

Save copy in VF_RBUF

Retrieve repeated bytes in subgroup from VF_RBUF

Retrieve non-repeated bytes form received VF

Page 25: 1 Improving VoIP Transfer Rate over Internet Syed Misbahuddin, Dr. Engg. Department of Computer Science and Software Engineering University of Hail, Saudi

25

Conclusion

• With the application of proposed data reduction algorithm, off the shelf data compression algorithms may not be needed

• Proposed data reduction algorithm may give better Internet bandwidth utilization retaining the quality of voice signals