TY - GEN
T1 - A study on QoS of VoIP networks: a random neural network (RNN) approach
AU - Radhakrishnan, Kapilan
AU - Larijani, Hadi
N1 - <p>Paper presented at the 2010 Spring Simulation Multi-Conference, Orlando, Florida, 11-15 April 2010. Conference website: <a href="http://www.scs.org/springsim/2010">http://www.scs.org/springsim/2010</a></p>
PY - 2010/1/1
Y1 - 2010/1/1
N2 - Voice over Internet Protocol (VoIP) is predicted to be the replacement of the traditional PSTN telephone system. Quality of Service (QoS) of VoIP systems are more difficult to measure and implement compared to PSTN systems. The nature of QoS in VoIP networks is very variable and hence it is important to be able to measure the QoS offered by the system in real time with a low computational cost. So it is very important to measure the quality of service offered by VoIP networks. In this paper we discuss a new novel model to calculate the perceived voice quality using Random Neural Network (RNN). The RNN is an open Markovian queuing model, motivated by spiking behaviour of biological neurons that has been used for several applications. We used the feedforward architecture with different numbers of hidden neurons to test the stability of our model. We study the RNN model with 4, 5, and 6 hidden layers of neurons. Our model shows a high degree of accuracy.
AB - Voice over Internet Protocol (VoIP) is predicted to be the replacement of the traditional PSTN telephone system. Quality of Service (QoS) of VoIP systems are more difficult to measure and implement compared to PSTN systems. The nature of QoS in VoIP networks is very variable and hence it is important to be able to measure the QoS offered by the system in real time with a low computational cost. So it is very important to measure the quality of service offered by VoIP networks. In this paper we discuss a new novel model to calculate the perceived voice quality using Random Neural Network (RNN). The RNN is an open Markovian queuing model, motivated by spiking behaviour of biological neurons that has been used for several applications. We used the feedforward architecture with different numbers of hidden neurons to test the stability of our model. We study the RNN model with 4, 5, and 6 hidden layers of neurons. Our model shows a high degree of accuracy.
KW - communications technology
KW - random neural networks
KW - engineering
U2 - 10.1145/1878537.1878656
DO - 10.1145/1878537.1878656
M3 - Conference contribution
SN - 9781450300698
BT - Proceedings of the 2010 Spring Simulation Multi-Conference
PB - Society for Computer Simulation International
ER -