Hidden Markov modeling for network communication channels

Abstract

In this paper we perform the statistical analysis of an Internet communication channel. Our study is based on a Hidden Markov Model (HMM). The channel switches between different states; to each state corresponds the probability that a packet sent by the transmitter will be lost. The transition between the different states of the channel is governed by a Markov chain; this Markov chain is not observed directly, but the received packet flow provides some probabilistic information about the current state of the channel, as well as some information about the parameters of the model. In this paper we detail some useful algorithms for the estimation of the channel parameters, and for making inference about the state of the channel. We discuss the relevance of the Markov model of the channel; we also discuss how many states are required to pertinently model a real communication channel.

Performance analysis for -bottleneck cell in large-scale wireless networks

“…S. Tao and R. Guerin [11] developed a layered model for end-to-end communication performance across two different time scales. Each hidden state of HMM [12] is characterized by a parameter representing the individual packet loss probability. Another hierarchy HMM is proposed in [13], with a separate 2-state Markov chain operating inside each of the hidden chain's states.…”

“…By [20], we can obtain that ω is positive if and only if χ 1 (x) 0, where (12) χ 1 (x) 0 implies that, for x c 3 λ i /μ, ψ(x) 0 and so x η and, for x > c 3 λ i /μ, χ (x) 0 and so x > λ i /μ i . When μ/c 3 > μ i , from Lemma 5.1, we can know that c 3 λ i /μ < η < λ i /μ i .…”

A Markov Decision Process based flow assignment framework for heterogeneous network access

“…While the wired access networks may be characterized using simpler techniques, the fast variations in wireless access network characteristics require robust stochastic models. It is shown in the literature that Markov models can well characterize network characteristic variation behavior [21,22]. In addition, there exists well-established computational and theoretical methods to optimize Markovian processes [4].…”

An analytical model for end-to-end communication channel over PLCN based on QBDs

“…Su et al [5] developed an iteration algorithm for the Gilbert model that allows the evaluation of the probability of observing i errors out of j transmissions, P (i, j), conditioned on loss rate feedback from the channel. An analytical model of packet loss based on HMM (Hidden Markov Models) is discussed in [6] and [7]. Tao et al [6] developed a layered model for predicting end-to-end loss performance across two different time scales.…”

“…Their model tries to predict the long-term loss rates and the percentage of loss bursts shorter than 3 packets, displaying small prediction error in the former while failing to perform well in the latter. In the HMM [7], each hidden state is characterized by a parameter representing the individual packet loss probability. Since, in this model, hidden state transitions occur after every packet observation, the expected fraction of losses in a near future interval may converge too rapidly to the steady-state loss probability, ignoring short-term fluctuations in this measure.…”