Predicting End-to-end Network Load

Vashist, Akshay; Mau, Siun-Chuon; Poylisher, Alexander; Chadha, Ritu; Ghosh, Abhijeet

doi:10.1109/icmla.2010.145

Cited by 2 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also use graph models to describe the DFM, and the calculation is also based on LDA. Vashist et al [8] proposed LDA, and because of a reasonable hypothesis and better performance, it was soon applied to many areas of research.…”

Section: Related Workmentioning

confidence: 99%

“…They also used the MEP to measure the parameter values, and with these parameters, they used a linear algorithm to predict the future's RTT. This type of prediction model can be found in numerous other study areas, such as economics, and many classic algorithms (Grey Theory, Time Series Analysis) can be found in these applications [6]- [8]. The system models currently used to predict the regularity of our data are relatively simple and have low accuracy.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Network Delay Factor Model Based on the Hidden Markov Model and Latent Dirichlet Allocation

Yuchi

Yang

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Network systems are widely used in today's digital communication systems and applications. Network delay is a very important factor affecting the efficiency and reliability of the system, and researchers often use a model to predict and understand the causes and patterns of delay. For our model, we propose beginning with the Hidden Markov Model (HMM) of the transmission network, which is used to analyze the network's operational situation and to predict the approximate state of the system into the future using the HMM prediction algorithm. Then, employing Latent Dirichlet Allocation (LDA), we put forward the Delay Factor Model (DFM) value for the delay. In the DFM, we need to map the delay's interval to an integer, designated as DII (Delay Interval Integer), and the factors returned define the hidden states of the HMM. From the view of the DFM, DII is generated from a factor and the previous DII randomly. We use the Gibbs Sampling approach to obtain an estimation of the DFM's parameters. By defining the HMM and DFM, we can forecast future delay with high accuracy, and the result can be shown to follow the peaks and troughs of the real operation of the network system's delay patterns.INDEX TERMS Delay factor model, derivation, hidden Markov model, prediction.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%