SlideShare a Scribd company logo
1 of 3
Download to read offline
ANN BASED DISCRETIZED SOURCE MODEL FOR
MULTIMEDIA TRAFFIC
ABHISHEK SINGH S.C.GUPTA
DEPT. OF ELECTRICAL ENGG. DEPT. OF ELECTRICAL ENGG.
AND COMPUTER ENGG. IT - BHU
& COMPUTER SCIENCE VARANASI - 221005
UNIVERSITY OF CIN CINNATI INDIA.
OH - 45221 - 0030
U.S. A.
ABSTRACT:
Feed forward network for modelling of discrete multimedia traffic source has been proposed. ANN is trained
to classify each time slot into its transmission status & state. Training data are obtained using MMBP source model. It is
highly recommended for slowly changing video scenes.
Index Term: ATM, CAC, ANN, Source Utilisation.
I. INTRODUCTION
The Extensive use of video on demand
together with the multimedia interactive application
have led to the modelling of variable bit rate with the
aim of analysing its behaviour.
Multimedia traffic is the superposition of
traffic from many different sources. It is necessary to
model them correctly. Besides helping in modelling of
switch it also plays an important role in ATM networks
where CAC parameters cannot be known until traffic
parameters are known. In the earlier approaches
renewal traffic models like Poisson, Bernoulli or phase
- type renewal models were proposed. In these, the
cell inter-arrival time for each individual traffic is
assumed to have independent identical distribution.
However, since it does not capture the autocorrelation,
so it cannot explain traffic burstiness which is the
major characteristics of broadband network and
LAN[4]. To remove this limitation markov modulated
traffic models were proposed and studied. In these and
auxiliary markov process M is defined and probability
law of arrivals is controlled or modulated by the state
of this process. The most commonly used Markov
modulated model is Markov modulated Poisson’s
process and Markov modulated Bernoulli process.
In Markov modulated Poisson’s process
arrival occurs according to Poisson’s process at a rate,
λ k in the state ‘k’ of M. Markov modulated Bernoulli
process has been used extensively for modelling of
discrete sources in [ 6 - 8 ] owing to its ability to
capture the strong correlation in traffic intensity and yet
be analytically tractable. In the MMBP source models,
the theoretical and practical value of source utilisation
matches after considering a large no. of time slots [ 10
]. For lower no. of time slots (around 1000), this
mismatch is highly significant.
This paper presents novel method for modelling of
discrete time sources. ANN have been used in
diversified application like for example congestion
control Their application for traffic modelling has not
been made. An attempt has been made in this work to
use ANN for modelling of discrete sources. ANN
source models diminishes error due to conventional
method modelling approximation by employing
learning capacity of neural networks. It is being
trained to learn and classify each time slots. A lesser
number of packets will have to generated to get exact
characteristics of source if an ANN based source model
is trained with actual source parameters as inputs. In
this work model of ANN has been trained using
MMBP source. During testing phase output of ANN
has been compared with the output of MMBP. Studies
showed that it could be used as an alternative way.
II. MMBP SOURCE MODELS.
Time in MMBP is discretized into slots. The
probability that a Slot contains a cell is a Bernoulli
process with a parameter that varies according to an r-
state Markov process which is independent of arrival
process. At the end of each slot, the Markov process
moves from state ‘i’ to state ‘j’ with probability for i =
1......r. At state i, a slot contains a cell with probability
a and no cell with probability (1-α). For 2 state MMBP
when the multimedia source is in state 1(2) it generates
a cell with probability α(β) and may remain in this
state in the next period with probability p(q). For a
value of α, β, p and q probability that a slot contains a
cell is given by [1].
ρ =[(1 - q)α + (1 - p)β] / [2 - p - q]
2 state MMBP model
Consider the time an arrival occurs when the Markov
process is in state 1. In next slot
* MMBP may remain in state 1 and arrival may
occur which happens w.p. ρα
* MMBP may remain state 2 an arrival may
occur which happen
wp (1-ρ)β.
* MMBP may remain state 1 and no arrival will
occur which happens
w.p.ρ(1-α)
* MMBP may move to state 2 and no arrival
occur which happen
w.p.(1-ρ)(1-β).
From ANN point of view this problem can be
viewed as classification problem. Time is discretized
into slots with each slot containing a cell. Here for 2-
state model a time slot is classified into
*Transmission state or non transmission state.
* State A or State B.
Since feed forward network with error back
propagation has been recommended for classification
and pattern recognition problem and this problem
being of similar type. So an attempt has been made in
this paper to solve this problem using feed forward
ANN network. The ANN model can be trained on
actual source to get more accurate results. The
digitised value of discrete time source and time will act
as input to ANN model which would classify it into
corresponding class.
III. SIMULATION ASPECT
Since MMBP model are being used
discretized source modelling so trained data set have
been generated by suing MMBP source model. The
value of p,q, α,β have been fixed to 0.2, 0.3, 0.4,0.3
respectively. For different value of t training data set is
generated. Since p, q, α, β, t are the inputs so there
are five inputs nodes. For a two phase MMBP outputs
can have four combinations (or divided into four
classes).
* State one with transmission
* State one with no transmission
* State two with transmission
* State two with no transmission
So total of four output nodes will be there for
two phase MMBP. Where two nodes will denote
transmission status and two node will denote state.
Node1 Node2 Node3 Node4 Meaning
1 0 1 0 No transmission
State A
1 0 0 1 No transmission
State B.
0 1 1 0 Transmission
State A
0 1 0 1 Transmission
State B
For fast learning of ANN architecture a
transformation scheme has been adopted. The
combination of output has been mapped into these
classes respectively.
1010 1000
1001 0100
0110 0010
0101 0001
By hit and trial 21 nodes in single hidden
layer has been found to give good result. Initially
momentum has been fixed to 0.78 and is decremented
by 0.00001 in each iteration.
IV. RESULT
After extensive training it showed an accuracy
of 88%. It is being able to classify 88% of time slots
into respective state and into respective transmission
status. If a class stays for only one time slot then ANN
model fails recognise it. If a class stays for two time
slots then it fails to classify first time slot however it
correctly predicts the second time slot and there is
transition to next state in subsequent time slots then the
model is able to predict this also correctly. In some
cases if a class is for fourteen to fifteen time slots then
model shows error in two or three slots mostly at the
last time slots. However this can be removed by
further training.
V. CONCLUSION & DISCUSSION
The main characteristics of multimedia traffic
is its great diversity of correlation and burstiness.
Since ANN based model is being trained to capture
each time slot so it is recommended for discretized
source modelling; If the transition probability between
sources is low then it will occupy many time slots. So
ANN based source model is recommended for those
time slots which has low transition probability like
slowly changing vided scenes etc.
REFERENCES
[1] Raif O. Onvural, “Asynchronous Transfer
Mode Networks : Performance issues”, Artech House
Boston London, 1994.
[2] H. KRONER, B. Hebuterne, P. Bayer and A.
Gravey, “Priority management in ATM switching
nodes”, IEEE selected Areas in communication, April
1991, 9(3), PP 418-427.
[3] H. Kroner, “Comparative performernce study
of space priority mechanism for ATM Networks”,
proceedings of INFOCOM’90, Jun 1990, PP 1136-
1143.
[4] S. Suri, D. Tipper and G. Mmempat, “A
comparative of space priority strategies in ATM
Networks”, proceedings of INFOCOM’94 June,1994
PP 515-523.
[5] R. Chipalkatti, J.F. Kurose and D. Towsle,
“Scheduling Policies for real time traffic in a statistical
multiplexer”, proceedings of INFOCOM’89, April
1989, PP 774-782.
[6] P.Yegani. M. Krunz and H. Hughes,
“Congestion central scheme in prioritized ATM
Networks”, Proceedings of ICC’94, May 1994, Vol 2
PP 1169-1173.
[7] M. Krunz, M. Hughes and P. Yegani, “Design
and analysis of buffer management scheme for
multimedia traffic with loss and delay priroity”,
proceedings of GLOBECOM’94, NOV, 1994 PP
1560-1594.
[8] K.S. Meler-Hellstern, “The analysis of a
queue arising in a overflow models”, IEEE Trans.
April 1989, COM 37(4) PP 367-372.
[9] V.S. Frost and B. Melamed “Traffic
modelling for telecommunication networks” Proc.
IEEE Feb. 1991, 79(2).
[10] Abhishek Singh, “ATM neworks performance
issues for source modelling and policing mechanism”,
B.Tech Thesis, Department of Electrical Engg.
Institute of Technology, BHU, 1999.
[11] M.M. Lee, Y.S. Mun, B.G.Kim, “Performance
analysis of delaying loss priority mechanism using
Markov Modulated arrival stream”, IEE Proc.
communication Vo.144 No.5 October 1997.
[12] Jacek.M. Zurada, “Introduction to Artificial
Neural Systems, Jacio Publishing House, 1997.

More Related Content

Similar to ANN.pdf

MM-PNEMO: a mathematical model to assess handoff delay and packet loss
MM-PNEMO: a mathematical model to assess handoff delay and packet lossMM-PNEMO: a mathematical model to assess handoff delay and packet loss
MM-PNEMO: a mathematical model to assess handoff delay and packet loss
journalBEEI
 
Paper id 2120147
Paper id 2120147Paper id 2120147
Paper id 2120147
IJRAT
 
Dynamic Manycasting in Optical Split-Incapable WDM Networksf
Dynamic Manycasting in Optical Split-Incapable WDM NetworksfDynamic Manycasting in Optical Split-Incapable WDM Networksf
Dynamic Manycasting in Optical Split-Incapable WDM Networksf
AlyciaGold776
 
A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...
A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...
A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...
International Journal of Modern Research in Engineering and Technology
 

Similar to ANN.pdf (20)

MM-PNEMO: a mathematical model to assess handoff delay and packet loss
MM-PNEMO: a mathematical model to assess handoff delay and packet lossMM-PNEMO: a mathematical model to assess handoff delay and packet loss
MM-PNEMO: a mathematical model to assess handoff delay and packet loss
 
THROUGHPUT ANALYSIS OF MOBILE WIMAX NETWORK UNDER MULTIPATH RICIAN FADING CHA...
THROUGHPUT ANALYSIS OF MOBILE WIMAX NETWORK UNDER MULTIPATH RICIAN FADING CHA...THROUGHPUT ANALYSIS OF MOBILE WIMAX NETWORK UNDER MULTIPATH RICIAN FADING CHA...
THROUGHPUT ANALYSIS OF MOBILE WIMAX NETWORK UNDER MULTIPATH RICIAN FADING CHA...
 
M.E Computer Science Mobile Computing Projects
M.E Computer Science Mobile Computing ProjectsM.E Computer Science Mobile Computing Projects
M.E Computer Science Mobile Computing Projects
 
M phil-computer-science-mobile-computing-projects
M phil-computer-science-mobile-computing-projectsM phil-computer-science-mobile-computing-projects
M phil-computer-science-mobile-computing-projects
 
M.Phil Computer Science Mobile Computing Projects
M.Phil Computer Science Mobile Computing ProjectsM.Phil Computer Science Mobile Computing Projects
M.Phil Computer Science Mobile Computing Projects
 
D04622933
D04622933D04622933
D04622933
 
Modeling and Performance Evaluation TAODV Routing Protocol Using Stochastic P...
Modeling and Performance Evaluation TAODV Routing Protocol Using Stochastic P...Modeling and Performance Evaluation TAODV Routing Protocol Using Stochastic P...
Modeling and Performance Evaluation TAODV Routing Protocol Using Stochastic P...
 
Performance evaluation on the basis of bit error rate for different order of ...
Performance evaluation on the basis of bit error rate for different order of ...Performance evaluation on the basis of bit error rate for different order of ...
Performance evaluation on the basis of bit error rate for different order of ...
 
Performance evaluation with a
Performance evaluation with aPerformance evaluation with a
Performance evaluation with a
 
Review of crosstalk free Network
Review of crosstalk free NetworkReview of crosstalk free Network
Review of crosstalk free Network
 
Peak detection using wavelet transform
Peak detection using wavelet transformPeak detection using wavelet transform
Peak detection using wavelet transform
 
Comparative analysis of congestion
Comparative analysis of congestionComparative analysis of congestion
Comparative analysis of congestion
 
Hybrid Adaptive Channel Estimation Technique in Time and Frequency Domain for...
Hybrid Adaptive Channel Estimation Technique in Time and Frequency Domain for...Hybrid Adaptive Channel Estimation Technique in Time and Frequency Domain for...
Hybrid Adaptive Channel Estimation Technique in Time and Frequency Domain for...
 
Implementation of OFDM System Using Various Channel Modulation Schemes
Implementation of OFDM System Using Various Channel Modulation SchemesImplementation of OFDM System Using Various Channel Modulation Schemes
Implementation of OFDM System Using Various Channel Modulation Schemes
 
Research paper (channel_estimation)
Research paper (channel_estimation)Research paper (channel_estimation)
Research paper (channel_estimation)
 
Paper id 2120147
Paper id 2120147Paper id 2120147
Paper id 2120147
 
Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...
Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...
Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...
 
Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...
Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...
Investigation of Ant Colony Optimization Algorithm for Efficient Energy Utili...
 
Dynamic Manycasting in Optical Split-Incapable WDM Networksf
Dynamic Manycasting in Optical Split-Incapable WDM NetworksfDynamic Manycasting in Optical Split-Incapable WDM Networksf
Dynamic Manycasting in Optical Split-Incapable WDM Networksf
 
A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...
A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...
A Distributed Rendezvous Point Source (RPS) For Congestion Control in A Relia...
 

Recently uploaded

Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
notes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptnotes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.ppt
MsecMca
 
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak HamilCara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Kandungan 087776558899
 

Recently uploaded (20)

Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
 
notes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptnotes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.ppt
 
Booking open Available Pune Call Girls Koregaon Park 6297143586 Call Hot Ind...
Booking open Available Pune Call Girls Koregaon Park  6297143586 Call Hot Ind...Booking open Available Pune Call Girls Koregaon Park  6297143586 Call Hot Ind...
Booking open Available Pune Call Girls Koregaon Park 6297143586 Call Hot Ind...
 
(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7
(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7
(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7
 
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
 
chapter 5.pptx: drainage and irrigation engineering
chapter 5.pptx: drainage and irrigation engineeringchapter 5.pptx: drainage and irrigation engineering
chapter 5.pptx: drainage and irrigation engineering
 
NFPA 5000 2024 standard .
NFPA 5000 2024 standard                                  .NFPA 5000 2024 standard                                  .
NFPA 5000 2024 standard .
 
Bhosari ( Call Girls ) Pune 6297143586 Hot Model With Sexy Bhabi Ready For ...
Bhosari ( Call Girls ) Pune  6297143586  Hot Model With Sexy Bhabi Ready For ...Bhosari ( Call Girls ) Pune  6297143586  Hot Model With Sexy Bhabi Ready For ...
Bhosari ( Call Girls ) Pune 6297143586 Hot Model With Sexy Bhabi Ready For ...
 
UNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its PerformanceUNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its Performance
 
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
 
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
 
Online banking management system project.pdf
Online banking management system project.pdfOnline banking management system project.pdf
Online banking management system project.pdf
 
Thermal Engineering-R & A / C - unit - V
Thermal Engineering-R & A / C - unit - VThermal Engineering-R & A / C - unit - V
Thermal Engineering-R & A / C - unit - V
 
Unleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leapUnleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leap
 
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
 
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak HamilCara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
 
Unit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfUnit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdf
 
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
 
Double Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueDouble Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torque
 
Generative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPTGenerative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPT
 

ANN.pdf

  • 1. ANN BASED DISCRETIZED SOURCE MODEL FOR MULTIMEDIA TRAFFIC ABHISHEK SINGH S.C.GUPTA DEPT. OF ELECTRICAL ENGG. DEPT. OF ELECTRICAL ENGG. AND COMPUTER ENGG. IT - BHU & COMPUTER SCIENCE VARANASI - 221005 UNIVERSITY OF CIN CINNATI INDIA. OH - 45221 - 0030 U.S. A. ABSTRACT: Feed forward network for modelling of discrete multimedia traffic source has been proposed. ANN is trained to classify each time slot into its transmission status & state. Training data are obtained using MMBP source model. It is highly recommended for slowly changing video scenes. Index Term: ATM, CAC, ANN, Source Utilisation. I. INTRODUCTION The Extensive use of video on demand together with the multimedia interactive application have led to the modelling of variable bit rate with the aim of analysing its behaviour. Multimedia traffic is the superposition of traffic from many different sources. It is necessary to model them correctly. Besides helping in modelling of switch it also plays an important role in ATM networks where CAC parameters cannot be known until traffic parameters are known. In the earlier approaches renewal traffic models like Poisson, Bernoulli or phase - type renewal models were proposed. In these, the cell inter-arrival time for each individual traffic is assumed to have independent identical distribution. However, since it does not capture the autocorrelation, so it cannot explain traffic burstiness which is the major characteristics of broadband network and LAN[4]. To remove this limitation markov modulated traffic models were proposed and studied. In these and auxiliary markov process M is defined and probability law of arrivals is controlled or modulated by the state of this process. The most commonly used Markov modulated model is Markov modulated Poisson’s process and Markov modulated Bernoulli process. In Markov modulated Poisson’s process arrival occurs according to Poisson’s process at a rate, λ k in the state ‘k’ of M. Markov modulated Bernoulli process has been used extensively for modelling of discrete sources in [ 6 - 8 ] owing to its ability to capture the strong correlation in traffic intensity and yet be analytically tractable. In the MMBP source models, the theoretical and practical value of source utilisation matches after considering a large no. of time slots [ 10 ]. For lower no. of time slots (around 1000), this mismatch is highly significant. This paper presents novel method for modelling of discrete time sources. ANN have been used in diversified application like for example congestion control Their application for traffic modelling has not been made. An attempt has been made in this work to use ANN for modelling of discrete sources. ANN source models diminishes error due to conventional method modelling approximation by employing learning capacity of neural networks. It is being trained to learn and classify each time slots. A lesser number of packets will have to generated to get exact characteristics of source if an ANN based source model is trained with actual source parameters as inputs. In this work model of ANN has been trained using MMBP source. During testing phase output of ANN has been compared with the output of MMBP. Studies showed that it could be used as an alternative way. II. MMBP SOURCE MODELS. Time in MMBP is discretized into slots. The probability that a Slot contains a cell is a Bernoulli process with a parameter that varies according to an r- state Markov process which is independent of arrival process. At the end of each slot, the Markov process moves from state ‘i’ to state ‘j’ with probability for i = 1......r. At state i, a slot contains a cell with probability a and no cell with probability (1-α). For 2 state MMBP when the multimedia source is in state 1(2) it generates a cell with probability α(β) and may remain in this state in the next period with probability p(q). For a
  • 2. value of α, β, p and q probability that a slot contains a cell is given by [1]. ρ =[(1 - q)α + (1 - p)β] / [2 - p - q] 2 state MMBP model Consider the time an arrival occurs when the Markov process is in state 1. In next slot * MMBP may remain in state 1 and arrival may occur which happens w.p. ρα * MMBP may remain state 2 an arrival may occur which happen wp (1-ρ)β. * MMBP may remain state 1 and no arrival will occur which happens w.p.ρ(1-α) * MMBP may move to state 2 and no arrival occur which happen w.p.(1-ρ)(1-β). From ANN point of view this problem can be viewed as classification problem. Time is discretized into slots with each slot containing a cell. Here for 2- state model a time slot is classified into *Transmission state or non transmission state. * State A or State B. Since feed forward network with error back propagation has been recommended for classification and pattern recognition problem and this problem being of similar type. So an attempt has been made in this paper to solve this problem using feed forward ANN network. The ANN model can be trained on actual source to get more accurate results. The digitised value of discrete time source and time will act as input to ANN model which would classify it into corresponding class. III. SIMULATION ASPECT Since MMBP model are being used discretized source modelling so trained data set have been generated by suing MMBP source model. The value of p,q, α,β have been fixed to 0.2, 0.3, 0.4,0.3 respectively. For different value of t training data set is generated. Since p, q, α, β, t are the inputs so there are five inputs nodes. For a two phase MMBP outputs can have four combinations (or divided into four classes). * State one with transmission * State one with no transmission * State two with transmission * State two with no transmission So total of four output nodes will be there for two phase MMBP. Where two nodes will denote transmission status and two node will denote state. Node1 Node2 Node3 Node4 Meaning 1 0 1 0 No transmission State A 1 0 0 1 No transmission State B. 0 1 1 0 Transmission State A 0 1 0 1 Transmission State B For fast learning of ANN architecture a transformation scheme has been adopted. The combination of output has been mapped into these classes respectively. 1010 1000 1001 0100 0110 0010 0101 0001 By hit and trial 21 nodes in single hidden layer has been found to give good result. Initially momentum has been fixed to 0.78 and is decremented by 0.00001 in each iteration. IV. RESULT After extensive training it showed an accuracy of 88%. It is being able to classify 88% of time slots into respective state and into respective transmission status. If a class stays for only one time slot then ANN model fails recognise it. If a class stays for two time slots then it fails to classify first time slot however it correctly predicts the second time slot and there is transition to next state in subsequent time slots then the model is able to predict this also correctly. In some cases if a class is for fourteen to fifteen time slots then model shows error in two or three slots mostly at the last time slots. However this can be removed by further training.
  • 3. V. CONCLUSION & DISCUSSION The main characteristics of multimedia traffic is its great diversity of correlation and burstiness. Since ANN based model is being trained to capture each time slot so it is recommended for discretized source modelling; If the transition probability between sources is low then it will occupy many time slots. So ANN based source model is recommended for those time slots which has low transition probability like slowly changing vided scenes etc. REFERENCES [1] Raif O. Onvural, “Asynchronous Transfer Mode Networks : Performance issues”, Artech House Boston London, 1994. [2] H. KRONER, B. Hebuterne, P. Bayer and A. Gravey, “Priority management in ATM switching nodes”, IEEE selected Areas in communication, April 1991, 9(3), PP 418-427. [3] H. Kroner, “Comparative performernce study of space priority mechanism for ATM Networks”, proceedings of INFOCOM’90, Jun 1990, PP 1136- 1143. [4] S. Suri, D. Tipper and G. Mmempat, “A comparative of space priority strategies in ATM Networks”, proceedings of INFOCOM’94 June,1994 PP 515-523. [5] R. Chipalkatti, J.F. Kurose and D. Towsle, “Scheduling Policies for real time traffic in a statistical multiplexer”, proceedings of INFOCOM’89, April 1989, PP 774-782. [6] P.Yegani. M. Krunz and H. Hughes, “Congestion central scheme in prioritized ATM Networks”, Proceedings of ICC’94, May 1994, Vol 2 PP 1169-1173. [7] M. Krunz, M. Hughes and P. Yegani, “Design and analysis of buffer management scheme for multimedia traffic with loss and delay priroity”, proceedings of GLOBECOM’94, NOV, 1994 PP 1560-1594. [8] K.S. Meler-Hellstern, “The analysis of a queue arising in a overflow models”, IEEE Trans. April 1989, COM 37(4) PP 367-372. [9] V.S. Frost and B. Melamed “Traffic modelling for telecommunication networks” Proc. IEEE Feb. 1991, 79(2). [10] Abhishek Singh, “ATM neworks performance issues for source modelling and policing mechanism”, B.Tech Thesis, Department of Electrical Engg. Institute of Technology, BHU, 1999. [11] M.M. Lee, Y.S. Mun, B.G.Kim, “Performance analysis of delaying loss priority mechanism using Markov Modulated arrival stream”, IEE Proc. communication Vo.144 No.5 October 1997. [12] Jacek.M. Zurada, “Introduction to Artificial Neural Systems, Jacio Publishing House, 1997.