Printer Friendly

A Survey of packet loss in voip.

Introduction

Voice Over Internet Protocol (VOIP) is one of the fastest growing applications for the internet today. Many users expect high quality telecommunication services. Voip is internet telephony. It is a category of hardware and software. It enables people to use the internet as the transmission medium for telephone calls by sending voice data in packets using IP rather than by traditional circuit transmissions of the PSTN (Public Switched Telephone Network). Voice over IP networks differ from conventional telephone networks. VoIP technology has allowed phone calls to be routed over Internet infrastructure rather than the traditional Public Switched Telephone Network (PSTN) infrastructure. The technology, called Voice over Internet Protocol (VoIP), uses the Internet Protocol (IP) to route packets containing small portions of voice conversations between the callers.

The transmission technology of VOIP must be in digital. Hence the caller's voice is digitized. The digitized voice is compressed and then separated into packets using complex algorithms. These packets are addressed and sent across the network which is to be reassembled in the proper order at the destination. Again, this reassembly can be done by a carrier, and Internet Service Provider, or by PC. During transmission on the Internet, packets may be lost or delayed, or errors may damage the packets. Conventional error correction techniques would request the retransmission of unusable or lost packets, but if the transmission is a real-time voice communication this technique obviously would not work, so sophisticated error detection and correction systems are used to create sound to fill in the gaps. After the packets are transmitted and arrive at the destination, the transmission is assembled and decompressed to restore the data to an approximation of the original form.

In Voice over IP (VoIP) applications, delay, jitter (13) and packet loss are the main network impairments that affect voice quality. Packet loss occurs when packets are lost during transmission or simply arrive too late to be used.

Packet loss can occur for a number of reasons

(1) Congestion of routers and gateways, which lead to packet being discarded

(2) Delays in packet transmission, with packet arriving too late at the receiver to be played back.

(3) Heavy loading of workstations, leading to scheduling difficulties in multitasking operating system

Transmission of data makes use of the TCP/IP protocol suite which allows for retransmission of missing packets, but VoIP, which uses UDP, does not allow retransmission and the missing packets are simply left out of the call. Such loss causes voice clipping and skips (3). One of the frequently used methods was retransmission. Since retransmission mechanisms are often unacceptable for interactive real-time audio applications such as Internet phone, because of the increased end-to-end delay. Applications that run over UDP do not retransmit lost packets. There is a need for error recovery before transmitting the data. When using network services that do not guarantee the Quality of Service (QoS) required by audio-visual applications, the recovery from losses due to congestion in the network is a key problem that must be solved.

MOS (Mean Opinion Score) is the most well-known measure of voice quality. It is a subjective method of quality assessment. Upto 1% is usually undetectable, more than 3% is the maximum permitted within industry standards. Test subjects judge the quality of the voice transmission system either by carrying on a conversation or by listening to speech samples. They then rank the voice quality using the following scale: 5-Excellent, 4-Good, 3-Fair, 2-Poor, 1-Bad (21).

MOS is then computed by averaging the scores of the test subjects. Using this scale, an average score of 4 and above is considered as toll-quality. MOS was originally designed to assess the quality of different coding standards.

History

The summary of survey was tabulated for a decade. The table 1shows the concepts used by researchers. The results obtained by them and the drawback they faced are also listed.

(17) In this paper, a new front end speech recognition over IP networks were proposed. They extracted the recognition feature vectors directly from the encoded speech instead of decoding it. They considered the ITU G.723.1 standard codec. The benefits quoted by authors are, this approach is very effective to packet loss since it is not constrained to the error handling mechanism of the codec. They compared new front end method with the conventional approach called Automatic Speech Recognition (ASR). There are two types of ASR. They are Speaker independent continuous speech and speaker independent isolated speech. The proposed method is compared with both ASR techniques. This scheme outperforms the conventional procedure. They concluded that the improvement is higher even though the network condition was worst.

(20) The author proposed Global Local Search--Time Scale Modification (GLS--TSM) receiver based scheme. This scheme is classified as sender-receiver based, network based, receiver based. This work focused only receiver based. This method provides flexible arrival delay cutoffs, reducing packet loss at the receiver, Low computational complexity, lost packets concealed effectively and no additional delay. But the performance is limited in silence, noise substitution and packet repetition. Rigorous objective and subjective tests for large number of input speech samples with varying network condition were conducted. These tests confirmed better performance. They concluded this fully receiver based scheme is suitable for any practical voip system.

(6) Describes an adaptive Joint Play out buffer and Forward Error Correction scheme. FEC techniques can be classified as media independent and media dependant. Media independent FEC uses block codes to provide redundant information. FEC send redundant information along with the original information. The benefits are, avoids delay, performs better than existing algorithms and recovered packet loss. The Drawback of this scheme was it introduces additional delay, uses block codes, provides redundant information. They compared the performance of play first and play best strategies. Play first is a delay aware FEC scheme. Play best is a non delay aware FEC scheme. Both lead similar results. Among these the author recommends delay aware play first strategy because of its simplicity. There is a real benefit using joint method.

(24) Discusses the maximal rate algorithm, proportionally fair algorithm and simple admission control scheme. The proportionally fair algorithm is suitable for elastic traffic when the channel condition is considered. The maximal rate algorithm shows twice of the loss rate for the same delay bound and load and it is a good choice to improve the whole performance. Simple admission control scheme controls the average portion of slots occupied by voip packets. They compared hard and soft algorithm. The frame structure divides in to two parts. The first part of the frame gets more priority. The second part is distributed to normal data which do not need urgent delivery. There are two algorithms to schedule voip. One is maximal rate algorithm the other one is proportionally fair algorithm. Each scheduler is divided into 2 categories. They are maxhard, maxsoft, pfhard, pfsoft. They concluded that pfsoft showed the best result. But in this method if the traffic is high, the drop probability is also high.
Table1: Year wise findings.

YEAR       AUTHOR                      CONCEPT

1998 (18)  C.Perkins et al.            FEC

1999 (4)   Bolot et al.                Adaptive Delay aware error
                                       control

2001 (17)  Palaezmoreno et al          New front end approach
                                       speech recognition over
                                       IP

2002 (20)  Samar Agnihotri et al       GLS--TSM receiver based
                                       scheme

2003 (6)   Catherine Boutremans et     Adaptive joint playout
           al                          buffer & FEC

2004 (24)  Young-June Choi et al       Maximal rate algorithm
                                       Proportionally fair
                                       algorithm Simple admission
                                       control scheme

2004 (15)  Lingfen Sun et al           New method for predictive
                                       voice quality for
                                       buffer design /
                                       optimization

2004 (12)  K.Kondo et al               Linear prediction in both
                                       forward and backward
                                       direction

2005 (21)  Shveni P metha              Comparative study of
                                       techniques to minimize
                                       packet loss

2005 (8)   Fernando Silveira Filho et  Adaptive forward error
           al                          correction for interactive
                                       streaming over the
                                       internet

2005 (14)  Kiki Karadimou et al        Source-filter model for
                                       multichannel audio

2006 (9)   Hanoch et al                Interleaving--packet
                                       dispersion

2007 (1)   An chan et al               Clique nalytical call
                                       admission in multiple
                                       WLAN

2007 (2)   Ashwin Kashyap et al        Zero stuffing and packet
                                       repetition scheme

2007 (11)  Jes Thyssen et al           a candidate for the ITU-T
                                       G.722 packet loss
                                       concealment standard

YEAR       FINDINGS                    DRAWBACK

1998 (18)  Reduces packet loss         Increases end to end
                                       delay

1999 (4)   Reduces Delay               Not considered losses Not
                                       managed additional delay
                                       due to FEC

2001 (17)  Very effective to packet    Performance was better
           loss                        without considering the
                                       network condition

2002 (20)  Reduces packet loss at the  Performance is limited
           receiver, Low complexity    with silence, noise
           and Lost packets concealed  substitution and packet
           effectively                 repetition

2003 (6)   Performs better than        Additional delay,
           existing algorithm          blockcodes, redundancy

2004 (24)  Good choice to improve the  Drop probability increases
           whole performance           if the traffic increases

2004 (15)  Achieves the optimum        Delay distribution model
           perceived voice
           quality

2004 (12)  Reduces complexity and      Improved performance
           processing delay

2005 (21)  Reduced packet loss         Redundancy

2005 (8)   This method not only        It increases bandwidth
           recovers more packets but   requirements
           also it performs
           efficiently

2005 (14)  Reconstructs the lost       Small overhead and delays
           information exclusively at  for the total encoding /
           the receiver side without   decoding process
           any overhead to the
           transmitter

2006 (9)   Improves quality, balance   Concentrated on burst
           the load                    losses

2007 (1)   Prevent packet collision    Header overhead, packet
           Solves multi cell mutual    aggregation
           interference Increases
           voip capacity

2007 (2)   works well with G.711 and   Artifacts are introduced
           G.722 codecs
           simultaneously

2007 (11)  alternative codec for       Additional computational
           packet loss concealment     complexity and memory
                                       usage


(15) Present their analysis with an efficient new method for predicting voice quality for buffer design/optimization. In this method first, nonlinear regression models are derived for a variety of codecs (e.g.G.723.1/G.729/AMR/iLBC) with the aid of ITU PESQ and the E-model. Second, they propose the use of minimum overall impairment as a criterion for buffer optimization. This criterion is more efficient than using traditional maximum Mean Opinion Score (MOS). Third, they show that the delay characteristics of Voice over IP traffic are better characterized by a Weibull distribution than a Pareto or an Exponential distribution. Based on the new voice quality prediction model, the Weibull delay distribution model and the minimum impairment criterion, they propose a perceptual optimization buffer algorithm. Preliminary results show that the proposed algorithm can achieve the optimum perceived voice quality compared with other algorithms under all network conditions considered. Preliminary results show that the proposed algorithm can achieve the optimum perceived voice quality compared with other algorithms under all network conditions considered.

(12) Move ahead on the idea of the linear prediction both in the forward and backward direction was proposed. Subjective quality is compared between the proposed method and the packet loss concealment algorithm. This method showed higher scores. There is a complexity and processing delay. They recommended adaptive LPC prediction to improve the quality. The adaptive LPC prediction order depends on the consecutive number of repetitive prediction. They planned to reduce the complexity using gradient LPC coefficient updates. The adaptive forward bidirectional prediction modes depending on the measured packet loss ratio is planned to reduce the processing delay.

(9) In their work, Delivery of real time streaming applications such as voice and video over IP in packet switched networks is based on dividing the stream into packets and shipping all the packets over a single path along the network. In contrast to traditional approach, the packets are dispersed over multiple paths. The reason is to improve quality, balance the load. The noticeable loss rate was used as a measure. They analyzed Bernoulli and Gilbert model for burst losses. The results suggested that the use of packet dispersion can be useful for voip applications.

In (1), a clique analytical call admission in the multiple wireless LAN scheme was proposed. Nowadays infrastructure WLAN is the most widely deployed network architecture. A 2-layer coloring problem was formulated to assign coarse time slots and frequency channels to voip sessions. The benefits are, prevents packet collision, solves multi cell mutual interference. The header overhead and packet aggregation is the big problem found by authors. The proposed scheme increases voip capacity in the multi cell environment. In the single cell scenario, all client stations are within the same cell and associated with the same AP. In a multi cell WLAN instead of one clique, multiple cliques can be formed.

Bolot et al. (4) proposed an adaptive rate/error control that optimizes a subjective measure of quality and incorporates a rate control. Their algorithm describes that the destination plays the best received copy of a given packet. They neither consider losses due to play out buffer overflow nor try to optimize the overall end-to-end delay. They do not manage the additional delay due to FEC. It is recognized that the end-to-end delay has a great impact on the perceived quality of interactive communications, with a threshold effect around 150ms. As a result the FEC scheme increases the delay. An adaptive delay aware error control was proposed in (16) to overcome this problem. This algorithm is based on the assumptions that if the source went to the trouble of adding some redundancy then the destination should wait for the redundant information to arrive.

(21) The author proposed many techniques to minimize the packet loss in voip. The first technique to replace lost packets, Interleaving, Repetition, and Interleaving with Repetition were used. In Interleaving, the information of a speech part is distributed in multiple packets. The data units are regrouped in a crossed form before transmission such that they are distributed, and at the receiver they are arranged in their original form. Thus instead of losing the whole packet small parts from distributed packets are lost. In Repetition, lost packets are replaced by copies of last received packets. In Interleaving with Repetition, the data are interleaved before sending and then any missing part is substituted using the repetition technique at the receiver. The second method was Forward Error correction and Concealment (FEC) adds redundancy to the transmission so that lost packets can be recovered, as long as the following packets are received successfully. Finally, Optimized unequal error protection method was used. In this method, certain packets are allocated more FEC protection than others depending on their perceived importance.

(18) FEC is used to mitigate the impact of packet losses. It increases the end to end delay since the destination has to wait for the redundant packets to be received in order to repair packet losses. It increases the bit rate requirement of an audio source. The sender driven mechanisms for error correction was proposed. They are Retransmission, Insertion based error concealment, and Interpolation based repair. Retransmission works well for small loss rates, In Retransmission Interleaving, FEC was discussed. Interleaving disperses the effect of packet loss whereas FEC is media dependant and media independent. In Insertion based error concealment, two schemes were used by author. One is Silence substitution, which fills the gap left by a lost packet with silence in order to maintain the timing relationship between the surrounding packets. It is only effective for short packet lengths and low loss rates. The other one is Noise substitution, instead of filling in the gap left by a lost packet with silence, background voice is inserted. In Interpolation based repair, pitch waveform replication, time scale modification, and Regeneration based repair were discussed. In pitch waveform replication, unvoiced speech segments are repaired using packet repetition and voiced losses repeat a waveform of appropriate pitch length. This performs better than wave form substitution. The Time scale modification performs better than both pitch waveform replication and waveform substitution. The interpolation of transmitted state and model based recovery are Regeneration Based Repair.

(8) Developed an adaptive mechanism for FEC selection using a predictive model. This method not only recovers more packets but also it performs efficiently. The computations required for the entire control mechanism must be fast. It increases bandwidth requirements, and controls redundancy. This methodology is applicable to video-conferencing.

(2) The author proposed zero stuffing and packet repetition schemes to reduce the packet loss. This scheme works well for G.711 and G.722 codecs simultaneously. This is useful in multirate system where both narrowband and wideband speech were supported. It introduces artifacts in both the methods.

(14) The source/filter model for multichannel audio was proposed by authors. This is useful for both stored recordings and streaming applications. This method reconstructs the lost packet only at the receiver side without redundancy. There is a small overhead for encoding decoding process.

(11) Suggested an alternative approach for G.722 packet loss concealment standard. The algorithm is based on waveform extrapolation in the speech domain. They compared many PLC codec algorithms. An additional complexity and memory usage are drawbacks.

Future Work

Repair methods for packet loss are known as voice reconstruction mechanisms (23). Better performance was provided by adaptive FEC schemes (7), (16), (19), and (22). The performance of these schemes is limited by potentially high buffering delays introduced and poor quality of speech delivered when schemes such as splicing, silence or noise substitution and packet repetition. The lost packets should be concealed as much as possible. For further research, the performance will be increased without introducing additional delay.

Conclusion

This survey provides an overview of existing approaches for packet loss techniques. Voice over IP is the new fancy development in the telecom industry. It promises to deliver cost savings to users and service providers and is driving the convergence of network and telecom. It offers improvements in quality, interoperability and applications in the near future. This paper surveyed packet loss techniques in voip. A variety of proposals for error recovery are reviewed. The proposed packet loss concealment algorithm gives a significant improvement in the quality of speech in voip. This mechanism is suitable for both unicast and multicast connections under all types of network conditions. The processing of damaged packets has been established as a suitable topic for further research. Packet loss tends to be a major cause of lost voice signals. It arises primarily from network congestion. Voice traffic can tolerate some packet loss. However, if the packet loss rate is greater than 5% it is considered harmful to the voice quality and a good concealment technique is required for reconstruction of the lost packets. In future an effort was directed to the development of a concealment algorithm that would maintain the quality of voice for lost packets.

References

(1) An Chan, Soung Chang Liew, "voip over multiple IEEE 802.11 wireless LANs", IEEE Trans on mobile computing, December 2007, pages: 1-14.

(2) Ashwin Kashyap, Mikael K.Rudberg, "A Low Complexity Loss Concealment Algorithm for G.711 and G.722", 2007.

(3) BUR GOODE, Senior Member, IEEE, "Voice over Internet Protocol (VoIP)", Proceedings of the IEEE, VOL. 90, NO. 9, September 2002.

(4) J.C.Bolot, S.F.Parisis and D.Towsley, "Adaptive FEC-based error control for Internet Telephony", in Infocomm'99, March 1999.

(5) C.Boutremans and J.Y.Le Boudec, "Adaptive Delay Aware Error control for Internet Telephony", April 2001.

(6) Catherine Boutremans, Jean-Yves Le Boudec, "Adaptive Joint play out buffer and FEC adjustment for internet telephony", IEEE INFOCOM, 2003.

(7) P.DeLeon and C.J.Sreenan "An adaptive predictor for media playout buffering", in IEEE conf on Acoustics, speech and signal processing, 1999, vol.6.

(8) Fernando Silveira Filho, Edson H.Watanabe, Edmundo de Souza e Silva," Adaptive Forward Error Correction for Interactive Streaming over the Internet", 2005.

(9) Hanoch, Levy, Haim, Zlatokrilov," the effect of packet dispersion on voice applications in IP networks", IEEE/ACM transactions on networking, vol.14, issue: 2, Apr, 2006, vol.14, issue 2, pages: 277-288.

(10) Georg Carle and Ernst W. Biersack," Survey of Error Recovery Techniques for IP based Audio-Visual Multicast Applications", 1997.

(11) Jes Thyssen, Robert Zopf, Juin-Hwey Chen, and Niranjan Shetty, "A Candidate for the ITU-T G.722 Packet Loss Concealment Standard", IEEE, ICASSP 2007.

(12) K. Kondo and K. Nakagawa, "A packet loss concealment method using recursive linear prediction", proc. international conference on spoken language processing Interspeech-ICSLP), October 2004.

(13) Kevork R.Piloyan, Vahe Nerguizian, "Novel Architecture for Routing Packetized Voice over Existing Internet Infrastructure without Using the Internet Protocol", IJCSNS International Journal of Computer Science and Network Security, VOL.6 No.7B, July 2006.

(14) Kiki Karadimou, Athanasios Mouchtaris,and Panagiotis Tsakalides, "Packet Loss Concealment for Multichannel Audio Using the Multiband Source/Filter Model", 2005.

(15) Lingfen Sun and Emmanuel Ifeachor, "New Models for Perceived Voice Quality Prediction and their Applications in Playout Buffer Optimization for VoIP Networks", IEEE communications society, 2004.

(16) S.B.Moon, J.Kurose, and D.Towsley, "Packet audio playout delay adjustment: Performance bounds and algorithms", August 1995.

(17) Pelaez-Moreno, C.Gallardo-antolin, A.Diaz-de-Maria, F., "Recognizing Voice over IP: A Robust Front End for Speech Recognition on the World Wide Web", IEEE Trans. Multimedia, vol. 3, issue: 2, June.2001, pages: 209-218.

(18) C. Perkins, O. Hodson, and V. Hardman, "A Survey of Packet Loss Recovery Techniques for Streaming Audio," IEEE network, vol.12, no.5, pp.40-48, September 1998.

(19) R.Ramjee, J.Kurose, D.Towsley, and H.Schulzrinne, "AdaptivePlayout Mechanisms for Packetized Audio Applications in wide area networks", in proc IEEE INFOCOM, June 1994, vol-02.

(20) Samar Agnihotri, K.Aravindhan, H.S.Jamadagni, B.I.Pawate, "A New Technique for Improving Quality of Speech in voice over IP using Time Scale Modification", IEEE Trans. 2002.

(21) Shveni P metha, "comparative Study of Techniques to Minimize Packet loss in voip", 21st computer science seminar, 2005.

(22) C.J.Sreenan,J, C.Chen, P,Agarwal and B.Narendran, "delay reduction technique for playoutbuffering", IEEE Trans. multimedia, vol.2, June 2000.

(23) Vicky Hardman, Martina Angela Sasse, Mark Handley, Anna Watson," Reliable Audio for use over the internet, 1995.

(24) Young-June Choi, Saewoong Bahk, "Scheduling for voip service in cdma20001x EV-DO" IEEE communications society 2004.

K. Maheswari (1) and M. Punithavalli (2)

(1) SG.Lecturer, Dept. of Computer Applications, SNR SONS College, Coimbatore.

(2) Prof. and Head, Dept. of Computer Science and Applications, Sri Ramakrishna College of Arts and Science for Women, Coimbatore.
COPYRIGHT 2009 Research India Publications
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:voice over internet protocol
Author:Maheswari, K.; Punithavalli, M.
Publication:International Journal of Computational Intelligence Research
Article Type:Report
Date:Jan 1, 2009
Words:3680
Previous Article:Enhancing computational spped for search application through high performance grid using Globus Tool kit.
Next Article:Wavelet based image compression using ROI SPIHT coding.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |