Voice over IP is a method of send oning voice as packages utilizing the Internet Protocol ( IP ) over an IP web [ 9 ] . Sending a voice over a packet-switched web has some advantages such as cost nest eggs and improved services [ 11 ] . But the quality of the voice sent has non ever been competitory. Packet shift has a ‘best attempt ‘ public presentation [ 12 ] . The web sends every package every bit fast as possible but there is no penchant in handling the packages. Besides, there is no warrant that all packages have been delivered successfully [ 12 ] . Harmonizing to V.Hardman et al. , 1995 [ 5 ] , ‘packet loss is a relentless job and the end-to-end hold is besides a critical factor. ‘ Jitter is besides a job happening in Voice over IP and in the survey done by Shveni Mehta, 2005 [ 6 ] , the codec used to transform the parallel signals into digital signifier is a factor to be considered every bit good.
2.1 Packet loss and discards
Loss of packages greatly affects the voice quality and unluckily they are really common. Some beginnings of package loss are:
Congestion of routers and gateways [ 5 ] , [ 6 ] . When there are excessively many packages which are being sent, the router can non manage all of them. The packages start line uping and there is a buffer flood. The geting packages are discarded by routers [ 13 ] . The size of the buffer is limited.
The cyberspace routes packages one at clip. Some of the packages may be delayed during their transmittal [ 5 ] , [ 6 ] , [ 14 ] . If they are late for excessively long, they become useless in some instances and are discarded by the receiving system [ 6 ] .
Bit mistake can be another job. In certain instances, there are some spots which get changed when the packages travel from one topographic point to another [ 5 ] . The amount of spots present in the package is attached to it. When the package is received, the router checks the content of the package and the amount of spots. If they are different, the package gets discarded.
Besides packages can acquire lost if it ‘s time-to-live ( TTL ) in its heading is expired. Due to this ground, even with an infinite buffer size, package loss is non eliminated as stated by Nagle 1987 [ 15 ] .
2.2 Types of holds
Packages are sent to their finish via a series of intermediate nodes. The amount of all holds experienced by a package on its manner to the finish is called the end-to-end hold. In a research conducted by T.Brady, 1971 [ 16 ] , if the conversation forms are non to be broken down, the end-to-end hold should be kept below 600ms in the absence of reverberations. The package size straight affects the end-to-end hold. There are some holds which are comparatively fixed such as coding algorithms and decryption and there are some which rely on the web conditions [ 8 ] .
Processing hold: the clip a routing device takes to get a package in and send on it [ 17 ] . The information takes a certain sum of clip to go across the web and to make the other terminal. Router need to analyze the package ‘s heading and direct it, pull off the informations flow and choose the best way.
Queuing/Buffering hold: clip taken for the package to be buffered before transmittal of packages onto the nexus [ 17 ] . When a user sends many packages at a clip, the router can non cover with all of them together. So it assigns precedences and a waiting line is created. The packages wait until the router processes so.
Propagation hold: clip taken for a digital signal to be transmitted via the wire [ 17 ] .
Transmission hold: clip taken for a router to direct the package onto the wire [ 17 ] .
Packetization hold: clip taken for the encoded information to be placed in a package. This hold is besides called accretion hold since the voice accumulates in a buffer before being released [ 18 ] . Packetization hold depends on the block size and the figure of blocks.
Jitter is the fluctuation of hold of the reachings of the package and it is present merely in packet-based webs [ 9 ] . There is an expected clip interval for the packages to make its finish but they may see different holds. Edward J.Daniel et al. , 2003 [ 19 ] examined the features and causes of the web hold jitter and developed a theoretical account for simulation of jitter. In the survey, the chief cause of jitter is the queuing holds experienced by the packages at the nodes. Besides when there is congestion in the IP web, the packages take different waies to make their finish and this may take to packet hold jitter. To cut down this consequence, jitter buffers or play-out buffers are used [ 8 ] . The buffer will present a little sum of hold so that the timing fluctuations are smooth. However, utilizing buffers to overcome this job can take to end-to-end hold [ 19 ] .
2.4 Packet Recovery Techniques
‘Packet loss is ineluctable ‘ as stated by T J.Kostas et al. , 1998 [ 8 ] but it can be controlled so that a better quality of address is produced. Many methods were proposed by many writers and they can be separated into two techniques [ 7 ] :
2.4.1 Receiver-based techniques
Receiver-based techniques execute its action at the receiving system merely. This technique performs error-concealment whereby merely an estimation of the losing package is obtained [ 20 ] . There are three chief classs in the received-based fix:
Figure 2.1: Receiver based strategies [ 7 ]
188.8.131.52 Insertion-based strategies
In this type of strategy, a ‘fill-in ‘ frame is inserted for a lost package. This technique is easy to implement. A particular characteristic of this technique is that the characteristics of the signal are non used to assist Reconstruction. But the public presentation is normally of hapless quality [ 7 ] .
In splice, the doomed package is replaced by zero-length stand-in. There is no spread left but the timing of the information is changed [ 7 ] . A survey on this technique has been performed by J. G. Gruber and L. Strawczynski, 1985 [ 21 ] . Splicing was shown to execute decrepit and can be used for really low loss rates ( & lt ; 3 % ) .
In this technique, the lost packages are replaced by the value ‘0 ‘ . The infinite left by a losing package is filled up by silence so as the timing relationship between the neighboring packages is maintained [ 7 ] . This method is widely used because it is really simple to implement. However as the package sizes and loss rate additions, the consequence produced by silence permutation gets worst. The public presentation of silence permutation is good for short package lengths ( & lt ; 4 MS ) and low loss rates ( & lt ; 2 % ) [ 22 ] .
The usage of noise can be a replacing for a lost package. Background noise, normally linear white Gaussian noise, is inserted in the infinite left by the losing package. Warren, 1982 [ 23 ] investigated the human perceptual experience of interrupted address. It was shown that the human encephalon has the capacity of mending the losing address section with a noise instead than a minute of silence. This is done of course by the human encephalon. This consequence is known as ‘phonemic Restoration ‘ . The quality of the address seems to be better [ 24 ] and it has an improved intelligibility [ 23 ] with noise permutation. The timing relationship can still be preserved. Furthermore during the soundless periods, the transmitter can direct a ‘comfort noise ‘ for the lost package. Therefore noise permutation is normally more recommended than silence permutation [ 5 ] , [ 7 ] .
Packet repeat is the replacing of the losing package by a transcript of the old package that reached merely before the loss. The public presentation of package repeat is good and it is less complex. Attenuation of the repeated units can be made to better the quality of repeat. The signal amplitude is decreased to zero. Packet repeat with attenuation is a measure towards the insertion techniques [ 7 ] .
184.108.40.206 Interpolation-based strategies
There are several Interpolation-based methods and they try to extrapolate from packages which are found near a loss so as to move as a replacement for the losing package [ 7 ] . Interpolation has a cardinal advantage compared to insertion-based techniques. The changing characteristics of a signal are taken into consideration for the replacing. But they are more hard to implement [ 7 ] .
In this method, a sound is used before and non-compulsorily after so as to happen a signal which is suited to replace the losing package. From the correct address which is received, a section is taken to make full in the doomed package. A survey of wave form permutation has been performed by Goodman et al. , 1986 [ 25 ] . The sound quality has been found to be better than utilizing silence permutation or package repeat.
Pitch Waveform Replication
Wasem et Al, 1988 [ 26 ] used a pitch sensing algorithm. In this strategy, positive and negative extremums in the wave form which give an estimate of the pitch are continuously searched. It has been found that Pitch Waveform Replication gives a better consequence than wave form permutation.
Time Scale Modification
Time Scale Modification enables the audio signal to be stretched across the infinite created left by the losing package. Sanneck et Al, 1996 [ 27 ] presented a strategy where vectors of pitch rhythms which intertwine each other on each side of the loss, are offset to counterbalance for the loss and at the topographic point of overlapping, the vectors are averaged. Time scale alteration is computationally heavier but the consequence is better than waveform permutation and pitch wave form reproduction [ 7 ] .
220.127.116.11 Regeneration-Based strategies
Regeneration-Based strategy uses algorithms for audio compaction so as to obtain codec factors to bring forth a replacing wave form. The consequence is expected to be good since a batch of information is used in the fix but this theoretical account is used seldom because it is hard to implement [ 7 ] . There are two types for this strategy:
Interpolation of familial province
The decrypting portion can construe what the province codec should be in. The reproduced signal bit by bit fades when there are more losingss. This method is non simple to implement [ 7 ] .
In this technique, address is regenerated to suit in the losing section to cover the loss. [ 7 ]
2.4.2 Sender-Based Repair techniques
Figure 2.2: taxonomy of sender-based fix techniques [ 7 ]
The sound encoding format is modified by the Sender-Based Repair technique by adding a certain sum of redundancy information to it unlike receiver-based techniques [ 28 ] . The sender-based fix can be divided into two parts [ 7 ] :
Passive channel cryptography
Retransmission is simple to work with. To retransmit, it is non necessary to utilize the original informations. The address can be changed to a lower bandwidth depending on how much operating expense can be accepted. But it adds on to the communicating latency. Retransmission requires information like the package ‘s sequence figure and an acknowledgement therefore increasing the sum of operating expenses [ 7 ] .
2.5 Forward Error Correction ( FEC )
A common illustration of inactive channel cryptography is frontward error rectification ( FEC ) . FEC techniques are by and large based on the usage of mistake sensing and rectification. In FEC, a controlled sum of excess packages is transmitted together with the original packages. Mistakes can be spotted and corrected without retransmitting the message once more [ 29 ] . There are many FEC algorithms viz. Hamming codification, Bose-Chandhuri-Hocquenghem codification and Reed-Solomon codification. The information transmittal channel can greatly impact a codification ‘s public presentation [ 29 ] . FEC can be media-independent, that is it does non depend on the contents of the information. However FEC schemes introduce extra holds and an increased sum of bandwidth is used. Media-specific FEC technique transmits each unit of the address in many different packages. In instance of a package loss, another package with same unit can be used alternatively [ 7 ] , [ 30 ] . Error-correcting codifications, though being more complex than mistake sensing codifications are normally given more attending in communicating applications [ 31 ] . It can be divided into:
Message blocks of fixed length are formed from the binary information sequence. Each message block contains thousand information spots and they are encoded into a block of N codeword digits. Parity spots are attached to the information bits organizing the group of n spots. Linear block codifications are characterised by the notation ( n, K ) with a block codification of length N and 2k codification words [ 32 ] . Reed-Solomon codification is the most known block codification. Block codifications are normally chosen in instances where applications need high velocity to execute.
2.5.1 Convolutional codifications
Convolutional codification adds excess spots. Its notation is ( n, K, K ) . The ratio k/n is known as the codification rate. The whole number K is the restraint length and it represents the figure of K ( k spot ) stages that the displacement registries consists of. In a convolutional encoder, the input sequence of k-bit information is passed through displacement registries. The end product spots of the registries are sampled to organize the binary codification symbols and are so transmitted. The original information sequence can be found if the decipherer knows the encoder ‘s province sequence. An of import characteristic of the convolutional encoder is that it has a memory. The end products of the encoder do non depend merely on the input K, but besides on the old K-1 input [ 32 ] .
Figure 2.3: Convolutional encoder [ 32 ]
Viterbi Convolutional decryption
The viterbi algorithm reconstructs the maximal likelihood way in the treillage. The distance between the standard signal at a clip T and all the trellis waies go through each province at that clip is calculated. This algorithm causes less heavy burden [ 32 ] .
2.5.2 Reed-Solomon codifications
18.104.22.168 Overview and Properties
Reed-Solomon codifications are non-binary cyclic mistake rectifying codifications and they form portion of optimum erasure codifications. Since its find in 1959 by Irving Reed and Gus Solomon, the Reed-Solomon codification has become an indispensable portion of many wireless communiction applications, satellite communicating, storage devices, digital telecasting and broadband modems ( ADSL ) [ 33 ] .
Reed-Solomon codifications add excess information to the original informations. After encoding, the encoded information may incorporate mistakes. The decipherer will so observe where mistakes are found in the end product informations and will rectify them with the aid of the excess information added. The sum of redundancy is of import since the figure of mistakes that can be corrected will depend on it.
The entire figure of codification symbols in the encoded block, N, nowadays in a block codification consist of K information spots and R para spots. A Reed-Solomon codification is represented by the notation ( n, K ) . The codification has n figure of symbols which consist of thousand figure of spots. The figure of thousand information symbols is besides known as the dimension of the codification.
n = 2m – 1 [ 2.1 ]
The difference ( n-k ) which represents the figure of para symbols is besides called 2t. The figure of symbols that can be corrected by the Reed-Solomon decipherer is up to ( n-k ) /2 [ 34 ] .
Merely half of the para symbols are corrected. One para symbol is used to follow the mistake and another one to rectify the mistake. One utile feature of Reed-Solomon codification is that information symbols added to an RS codification of length Ns wiill non diminish its minimal distance [ 32 ] . The minimal distance is given as:
dmin = n – K +1 [ 2.2 ]
One belongings of Reed-Solomon codifications is that they can rectify burst mistakes [ 35 ] . These mistakes are caused due to melting in the communicating channel. RS codification can rectify a symbol with merely one spot mistake and besides a symbol which has mistakes in all its spots since it will take it as a individual mistake. In instance of erasures, the mistake is already situated. So merely one para symbol is used so as to rectify the mistake. Reed-Solomon codification is the best option for encoding and decrypting for its ability to rectify burst mistakes and erasures harmonizing to Mohit Agrawal, 2010-2011 [ 35 ] .
Reed-Solomon codification is a good pick when long block codifications need to be transmitted because when the codification block size additions, the mistake public presentation is besides better. As the sum of redundancy added additions, the codification rate lessenings and the error-correcting capableness besides increases [ 32 ] .
22.214.171.124 Galois field
Encoding and decrypting with Reed-Solomon codifications is based on a field called Galois field. A field is a ensuing aggregation of operations like add-on, minus, division or generation and they are capable to the Torahs of commutativity, distributivity and associativity [ 36 ] . Galois Fieldss are finite and can be represented by a fixed length binary word. A galois field GF ( P ) contains p elements.
The Galois field can be widened to GF ( autopsy ) where m is non zero positive whole number [ 32 ] . A generator multinomial generates each component of the field. There can be different multinomials which will bring forth different Fieldss. Crude multinomial is used in Reed-Solomon codifications and it defines the finite Fieldss ( 2m ) [ 32 ] . A multinomial is usually written get downing with low order to the high order.
2.6 Theoretical Model
In this subdivision, some theories of digital communicating will be reviewed.
Digital voice communicating
Voiced and voiceless sounds
In this portion we will discourse how speech signals are produced. Speech production can be grouped into three different constituents [ 37 ] :
The first 1 is the quasi-periodical pulsation.
The 2nd instance is where the input excitement is noise-like in nature.
And the last 1 is where there is no excitement.
Voice address occurs when the input excitement is about periodic. Oscillatory quivers of the vocal cords signifier voiced sounds. The vocal Fords halt the air blown out of the lungs through the windpipe and the glottal moving ridge is produced [ 37 ] . There are some cardinal frequence and its harmonics in the spectrum of the sonant address. The being of the harmonic construction is defined by the frequence constituents which are repeated at regular intervals.
Degree centigrades: UsersMY PCDesktopmashoudavoiced nd unvoiced_filesexperiment3-theory-fig4.JPG
Figure: Block diagram representation of sonant address production [ 37 ]
The continuance of each rhythm is known as the cardinal period ( T0 ) .The cardinal frequence ( F0 ) of input excitement is called theA pitch frequencyA and it is one of most indispensable factor of the voice beginning [ 38 ] , [ 39 ] . The pitch depends on the strength and composing. The pitch is much higher in female and kids voice, approximately 200 Hz for an mean female voice and 200-300 Hz for kids whereas for work forces it is normally about 100 Hz [ 40 ] .
In voiceless address, the air is forced through a vocal piece of land obstructor ensuing in a turbulency and the sound caused is normally represented by a noise beginning. There is neither cardinal frequence nor any harmonic construction in the excitement signal [ 41 ] . It has a comparatively level spectrum. This is how a sonant and an voiceless address can be distinguished [ 37 ] .
Degree centigrades: UsersMY PCDesktopmashoudavoiced nd unvoiced_filesexperiment3-theory-fig61.jpg
Figure: Block diagram representation of voiceless address production [ 37 ]
The sonant and voiceless address is produced in sequence and they are separated by a silence part. In this part, there is no address end product. However silence is of import since the address becomes clearer and the information nowadays in the address can be identified [ 37 ] .
In the parallel, the voice transmittal frequence spectrum is technically 4 KHz. For digital telecommunication, the signal is 8 KHz, that is it is sampled twice the rate [ 42 ] .
Figure: PCM Communication [ 43 ]
Figure shows the stairss required for PCM communicating. Pulse Code Modulation is used to change over parallel signals into digital signifier [ 44 ] . The input signal which is an linear signal is foremost passed through a low base on balls filter of a certain cutoff frequence. All frequence constituents above this cutoff frequence will be blocked. The signal is so sampled to bring forth a Pulse Amplitude Modulated signal. Sampling is a procedure whereby the values of the filtered input signal can be obtained at distinct clip intervals, that is at a changeless sampling frequence [ 45 ] . The trying rate ( fs = 2fm ) is besides known as the Nyquist rate. The sampling frequence should be selected above the Nyquist rate so that there is sufficient figure of samples to stand for the parallel wave form ( aliasing ) [ 43 ] .
degree Fahrenheit a‰? 2fm
The PAM signal is uninterrupted in amplitude and discrete in clip. The signal is converted to a digital signifier. Each sample obtained is allocated a distinct value from a scope of possible values which is reliant on the figure of spots used to qualify each sample and this procedure is called quantisation. Each sample is assigned to the quantisation degree nearest to the value of the sample. Quantization noise or mistake is obtained by doing the difference between the original address and the distinct value assigned to it. It can be reduced by increasing the figure of quantisation degrees. When Quantization noise additions, the signal-to-noise ratio of a signal decreases since there are more mistakes. [ 45 ] .
Quantization can be unvarying or non-uniform. In the unvarying quantisation, the quantisation degrees are uniformly spaced. The quantisation noise is the same for all the magnitudes since noise is dependent on the measure size.
A non-uniform quantisation procedure is besides known as companding. In non-uniform quantisation, the measure size varies.A The quantisation noise is relative to the signal size [ 43 ] . Noise is reduced for the weak prima signals but for the seldom occurring signals, noise additions [ 43 ] .
Compressing the signal to be transmitted at the transmittal side and spread outing it at the having side forms the companding procedure. There are two companding strategies [ 45 ] viz. :
Aµ-law companding ( used in North America )
A-law companding ( used in Europe )
Address cryptography is the procedure of compacting the voice signals for efficient transmittal. Coding algorithm is used to minimise the spot rate in the digital representation of a signal without a important loss of the signal. A digital address is changed into a coded representation by a address programmer and a address decipherer reconstructs the address [ 46 ] . Address programmers are different in footings of spot rate, hold, degree of complexness and perceptual quality of the address [ 46 ] . A good address cryptography is one which uses less bit rate to stand for a address while continuing a good quality of address. Address can be processed in blocks utilizing the address programmers but this causes a communicating hold. There are chiefly two address coding techniques:
It tries to reproduce the address wave form every bit indistinguishable as possible [ 46 ] . It is at high spot rates that this type of coding gives a good quality of address [ 47 ] .
They keep merely the spectral belongingss of the address. Even a lower spot rates, a clear address can be produced [ 47 ] .
Address programmers are used in cellular communicating, videoconferencing and voice over IP.
2.6.4 Gilbert Model
Figure: Gilbert Model
The Gilbert loss theoretical account, besides known as the 2-state Markov concatenation theoretical account is used to implement burst package loss. It is simple and is good accepted to be used in voice over IP. The web is modeled with two provinces. State ‘1 ‘ represents a package loss and province ‘0 ‘ represents bringing of the package to its finish [ 48 ] . Figure shows the different provinces and whether a package is lost or delivered.
Gilbert theoretical account is normally a better estimate for the procedures of package loss. The parametric quantity P denotes the passage chance from province ‘0 ‘ to province ‘1 ‘ . It is the chance that a package will be dropped following given that the old package is non lost. The parametric quantity Q denotes the chance to stay in province ‘1 ‘ . It is the chance of a package being dropped given that the old package is dropped.
The matrix of passage chance of Gilbert theoretical account is:
2.6.5 Erasure codifications
Erasure codification is a forward mistake rectification codification for the binary erasure channel. To protect information from acquiring lost, erasure codifications provide space-optimal informations redundancy [ 49 ] . These codifications are used in communicating systems and in storage systems [ 49 ] . K blocks of beginning informations generates n blocks of encoded informations such that the original informations can be recovered back from a subset of the K blocks. The receiving system protects the information up to n-k nodes. The codification is represented as an ( N, K ) codification [ 50 ] . The codification rate is as follows:
R = ( )
There are different types of erasure codifications:
Optimum erasure codification
Near optimum erasure codification ( illustrations: LT codifications, Raptor codifications )
Rateless erasure code/ Near optimum fountain
In this undertaking, both receiving system and sender-based fix techniques are used for the privacy of package loss. The address is encoded utilizing two different FEC strategies, Reed-Solomon codification and Convolutional codification. After go throughing through the Gilbert package loss theoretical account, the lost packages are replaced utilizing silence permutation and package repeat techniques. The two FEC strategies and the receiver-based techniques are compared to cognize which combinations of techniques perform better.
[ 1 ] You Do n’t Know Jack About VoIP, THE COMMUNICATIONS THEY ARE A-CHANGIN ‘ .
PHIL SHERBURNE AND CARY FITZGERALD, CISCO, September 1, 2004 at:
hypertext transfer protocol: //queue.acm.org/detail.cfm? id=1028895
[ 2 ] A Brief History of VoIP, Document One, Joe Hallock – The Past, November 26, 2004
Development and Trends in Digital Media Technologies – COM 538 Masters of Communication in Digital Media University of Washington at:
hypertext transfer protocol: //www.joehallock.com/edu/pdfs/Hallock_J_VoIP_Past.pdf
[ 3 ] JISC Voice over IP: what it is, why people want it, and where it is traveling, Jane Dudman, Lending writer: Gaynor Backhouse, September 2006 at:
hypertext transfer protocol: //www.ciscovb.com/up/uploads/files/domain-357d053b63.pdf
[ 4 ] QuickStudy: Packet-Switched vs. Circuit-Switched Networks, By Lee Copeland, March 20, 2000 12:00 PM at: hypertext transfer protocol: //www.computerworld.com/s/article/41904/Packet_Switched_vs._Circuit_Switched_Networks
[ 5 ] Vicky Hardman, Martina Angela Sasse, Mark Handley, Anna Watson, , ”Reliable sound for usage over the Internet, ” Proc.INET ’95, 1995.
[ 6 ] Comparative Study of Techniques to minimise package loss in VoIP, Shveni P Mehta, 21st Computer Science Seminar SB3-T2-1, 2005 at:
hypertext transfer protocol: //www.ewp.rpi.edu/hartford/~rhb/cs_seminar_2005/SessionB3/mehta.pdf
[ 7 ] C. Perkins, O. Hodson, and V. Hardman, “ A study of package loss recovery techniques for streaming sound, ” IEEE Network, 1998, pp. 40-48.
[ 8 ] Real-time Voice Over Packet-Switched Networks
Thomas J.Kostas, Michael S.Borella, Ikhlaq Sidhu, Guido M.Schuster, Jacek Grabiec and Jerry Mahler 3COM, IEEE Network, January/February 1998
[ 9 ] AN OVERVIEW OF VOICE OVER INTERNET PROTOCOL ( VOIP )
Graduate pupil, M.S. in Computer Science Program, Rivier College
RIVIER COLLEGE ONLINE ACADEMIC JOURNAL, VOLUME 2, NUMBER 1, Spring 2006
[ 10 ] An Analytic and Experimental Study on the Impact of Jitter Playout Buffer on the
E-model in VoIP Quality Measurement
Olusegun Obafemi, Tibor Gyires, Yongning Tang
ICN 2011: The Tenth International Conference on Networks
[ 11 ] The benefits of VoIP, By Yuval Shavit, Features Writer, This was foremost published in November 2007 at: hypertext transfer protocol: //searchnetworkingchannel.techtarget.com/feature/The-benefits-of-VoIP
[ 12 ] Best Efforts Networking, Geoff Huston July 2001, at:
hypertext transfer protocol: //www.potaroo.net/ispcol/2001-09/2001-09-best.pdf
[ 13 ] Tier 1, MPLS Specialists Telecom Et Wan Solutions Worldwide, What Causes Packet Loss on the Internet? , October 30, 2007 by sgarson at: hypertext transfer protocol: //www.mpls-experts.com/what-causes-packet-loss-on-the-internet/ ? doing_wp_cron=1361361661.8698310852050781250000
[ 14 ] SHORT Paper
International Journal of Recent Trends in Engineering, Vol 2, No. 3, November 2009 120
AN IMPROVED PACKET LOSS RECOVERY IN VOIP USING COMBINED SOURCE AND RECEIVER BASED TECHNIQUE
K.Maheswari1, Dr. M.Punithavalli 2
1SNR Sons College
Dept. of Computer Applications, coimbatore, India
2 Sri Ramakrishna College of Arts and Science for Women
Dept. of Computer Science and Applications, coimbatore, India
hypertext transfer protocol: //ijrte.academypublisher.com/vol02/no03/ijrte0203120122.pdf
[ 15 ] On Packet Switches with Infinite Storage, This paper appears in:
Communicationss, IEEE Transactions on
Date of Publication: A Apr 1987
Merchandise Type: A Journals & A ; Magazines
On Packet Switches with Infinite Storage, JOHN B. NAGLE, IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. COM-35, NO. 4, APRIL 1987
[ 16 ] Brady P.T. ‘Effects of Transmission hold on Conversational Behaviour on Echo-Free Telephone Circuits ‘ Bell System Technical Journal, pp 115-134, January 1971.
[ 17 ] Characterizing Network Processing Delay
Ramaswamy Ramaswamy, Ning Weng and Tilman Wolf
Department of Electrical and Computer Engineering
University of Massachusetts
Amherst, MA 01003
[ 18 ] hypertext transfer protocol: //www.cisco.com/en/US/tech/tk652/tk698/technologies_white_paper09186a00800a8993.shtml
[ 19 ] An Inter-arrival Delay Jitter Model utilizing Multi-Structure
Network Delay Characteristics for Packet Networks
EdwardJ Daniel, Christopher M White, and Keith A. Teague
School of Electrical and Computer Engineering
Oklahoma State University
[ 20 ] Packet Loss Concealment for Audio Streaming
Submitted in Partial Fulfillment of The Requirements for the Degree of Master of Science in Electrical Engineering
Submitted to the Senate of the Technion – Israel Institute of TechnologySivan, 5766 Haifa June, 2006
[ 21 ] J. G. Gruber and L. Strawczynski, “ Subjective effects of variable hold and cutting in dynamically managed voice systems, ” IEEE Trans. Commun. , vol.COM-33, no. 8, Aug. 1985, pp. 801-8.
[ 22 ] N. S. Jayant and S. W. Christenssen, “ Effectss of package losingss in wave form
coded address and betterments due to an odd-even sample-interpolation process, ” IEEE Trans. Commun. , vol. COM-29, no. 2, Feb. 1981, pp. 101-9.
[ 23 ] R. M. Warren, Auditory Perception, Pergamon Press, 1982
[ 24 ] G. A. Miller and J. C. R. Licklider, “ The intelligibility of interrupted address, ”
J. Acoust. Soc. Amer. , vol. 22, no. 2, 1950, pp. 167-73.
[ 25 ] [ D.J Goodman, G.B Lockhrt, O.J. Wasem and W.-C Wong, ”Waveform Substitution Techniques for Recovering Missing Speech Segments in Packet Voice Communications ” , IEEE Transactions on Acoustics, Speech and Signal Processing, vol 34, num6, pp 1440-1448, December 1986
[ 26 ] C. O.J.Wasem, D.J.Goodman and H.G.Page, The effects of wave form permutation on the quality of pcm package communicating, ” IEEE Transactions on Acoustics, Speech, and Signal Processing, vol. 36, no. 3, pp. 342-348, 1988.
[ 27 ] K. H.Sanneck, A.Stenger and B.Girod, A new technique for audio package loss privacy, ” in IEEE Global Telecommunications Conference, pp. 48-52, November 1996.
[ 28 ] Frequency-Domain Stochastic Error Concealmentfor Wireless Audio Applications, Andreas Floros & A ; Markos Avlonitis & A ; Panayiotis Vlamos
# Springer Science + Business Media, LLC 2008
[ 29 ] Use Forward Error Correction To Improve Data Communications, Electronic Design, Aug 20,2000 at:
hypertext transfer protocol: //electronicdesign.com/communications/use-forward-error-correction-improve-data-communications
[ 30 ] J.C Bolot and A.Vega.Garcia, ”The instance for FEC based mistake control for package sound in the Internet, ” to look, ACM Multimedia Sys.
[ 31 ] Forward Error Correction ( FEC )
David R.Goff, Fiber Optic Video Transmission, 1st ed Focal Press: Woburn, Massachusetts, 2003 and other private Hagiographas.
[ 32 ] Digital Communications: Fundamentalss and Applications, Second Edition, Bernard Sklar, Prentice Hall PTR, Upper Saddle River, New Jersey 07458, Published Jan 11, 2001 byA Prentice Hall
[ 33 ] QoS Evaluation of Sender-Based, Loss-Recovery Techniques for VoIP, Teck-Kuen Chua and David C. Pheanis, Arizona State University, IEEE Network November/December 2006 pg 16 at:
hypertext transfer protocol: //126.96.36.199/download/learn_web/Tong ( 97-1 ) — Multimedia_Comm/pdf/QoS % 20Evaluation % 20of % 20Sender-Based % 20Loss-Recovery % 20Techniques % 20for % 20VoIP.pdf
[ 34 ] Reed Solomon codifications, Joel Sylvester, January 2001, elektrobit at:
file transfer protocol: //ftp.radionetworkprocessor.com/pub/reed-solomon/reedsolomon-elektrobit.pdf
[ 35 ] IMPLEMENTATION OF REED SOLOMON ERROR CORRECTING CODES
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Bachelor of Technology in Electronics and Communication Engineering By MOHIT AGRAWAL, Department of Electronics & A ; Communication Engineering, National Institute of Technology Rourkela,2010-2011
[ 36 ] Reed Solomon Encoder/Decoder on the StarCorea„? SC140/SC1400 Cores, With Extended Examples By Jasmin Oz and Assaf Naor
Rev. 1, 12/2004 pg 1-2
[ 37 ] hypertext transfer protocol: //iitg.vlab.co.in/ ? sub=59 & A ; brch=164 & A ; sim=613 & A ; cnt=1
[ 38 ] Chapter 1
SPEECH Analysis: THE PRODUCTION-PERCEPTION PERSPECTIVE
Li Deng and Jianwu Dang
One Microsoft Way, Redmond, WA 98052
School of Information Science, Japan Advanced Institute of Science and Technology
[ 39 ] University of California Los Angeles The Voice Source in Speech Production: Data, Analysis and Models A thesis submitted in partial satisfaction of the demands for the degree Doctor of Philosophy in Electrical Engineering By Yen-Liang Shue 2010, pg3
[ 40 ] COMP449: Address Recognition
Department of Computing, Macquarie University, A Sydney, Australia
Chapter 7. The Source Filter Model of Speech Production,
Part I. Acoustics and Digital Signal Processing
Copyright A© 2002 Department of Calculating at:
hypertext transfer protocol: //web.science.mq.edu.au/~cassidy/comp449/html/ch07.html
[ 41 ] hypertext transfer protocol: //www.acoustics.hut.fi/publications/files/theses/lemmetty_mst/chap3.html
[ 42 ] Technology Review # 2001-2, Network Convergence and Voice over IP, Debashish Mitra, March 2001 at:
hypertext transfer protocol: //ebooks.allfree-stuff.com/eBooks_down/Networking/TCS % 20- % 20Computers % 20- % 20Networking % 20- % 20VoIP % 20 % 26 % 20Network % 20Converage.pdf
[ 44 ] Title: Pulse codification transition techniques: with applications in communications and informations entering /aˆ‹ Bill Waggener.
Writer: Waggener, William N.
Published: New York: Van Nostrand Reinhold, c1995.
[ 43 ] Communication Systems II, Dr. Wa’il A.H. Hadi
Digital Communication Systems at:
hypertext transfer protocol: //www.uotechnology.edu.iq/dep-eee/lectures/4th/Electrical/Communication % 20engineering % 202/part1.pdf
[ 45 ] hypertext transfer protocol: //www.technologyuk.net/telecommunications/telecom_principles/pulse_code_modulation.shtml
[ 46 ] SPEECH Cryptography: FUNDAMENTALS AND APPLICATIONS
University of Illinois at Urbana-Champaign
University of California at Los Angeles
Los Angeles, California at:
hypertext transfer protocol: //www.ee.ucla.edu/~spapl/paper/mark_eot156.pdf
[ 47 ] 10.2 Speech Coding, Bishnu S. Atal & A ; Nikil S. Jayant
AT & A ; T Bell Laboratories, Murray Hill, New Jersey, USA at: hypertext transfer protocol: //www.cslu.ogi.edu/HLTsurvey/ch10node4.html
[ 48 ] Packet Loss Recovery and Control for Voice Transmission over the Internet ( 2000 ) by Henning Sanneck pg 70-72 at:
hypertext transfer protocol: //sanneck.net/research/publications/thesis/Sann0010_Loss.pdf
[ 49 ] Using Erasure Codes Efficiently for Storage in a Distributed System, Marcos K. Aguilera, Ramaprabhu Janakiraman, Lihao Xu, 2005 at:
hypertext transfer protocol: //research.microsoft.com/en-us/people/aguilera/erasure-dsn2005.pdf
[ 50 ] Effective Erasure codification for Reliable computing machine Communication Protocols, Luigi Rizzo, Dip.di Ingegnaria dell’Informazione, Universita di Pisa via Diotisalvi 2-56126 Pisa ( Italy )
hypertext transfer protocol: //pages.cs.wisc.edu/~suman/courses/740/papers/rizzo97ccr.pdf
Published in: Newsletter ACM SIGCOMM Computer Communication Review, vol.27 issue 2, Apr. 1997, page 24-36