State Transition Factor of Three-state Markov
Channels
Liang Huang, Jiefan Qiu, and Li Ping Qian
College of Computer Science and Technology
Zhejiang University of Technology, Hangzhou, China
{lianghuang, qiujiefan, lpqian}@zjut.edu.cn
Abstract—In this paper, we model the wireless fading channels
with Poisson arrivals and three-state Markov modulated service
processes (MMSP) as M/MMSP/1 queues. We derive the gen-
eralized Pollaczek-Khinchin formula that precisely characterizes
the delay performance of three-state Markov channels. Our key
contribution mainly lies in the state transition factor defined in
this paper that clearly describes the impact of channel varying on
the delay performance of these Markov channels. Moreover, we
resort to an innovative linear approximation of the start-service
probability in terms of server utilization to obtain closed-form
expressions, which are all verified by simulations.
Keywords—three-state; Markov channels; state transition fac-
tor; delay analysis;
I. INTRODUCTION
Time-varying communication channel is well studied as
two-state Markov channels since Gilbert and Elliott in [1] and
[2], in which they assumed that the channel is either noiseless
or totally noisy. In practice, sometimes it is more appropriate to
model communication systems with another intermediate state,
as three-state Markov channels [3]. These channel models are
generalized to finite-state Markov channel (FSMC) models by
Wang and Moayeri in [4], which explicitly established the
link between the physical parameters of wireless channels
and these Markov channel states. All these models have been
widely applied to study the design of wireless communication
systems.
In previous considerations of delay-sensitive wireless sys-
tems requiring quality of service (QoS) guarantees, the two-
state Markov channel is traceable and widely adopted for
general analysis. Its generating functions of queueing per-
formance are widely used in the literature. Recently, we
derived a generalized Pollaczek-Khinchin (P-K) formula for
two-state Markov channels in [5], in which we demonstrated
that the queueing delay of Markov channels can be fully
characterized by a newly defined state transition factor. The
delay performance of the FSMC with more than two states
is usually investigated by the matrix-geometric method [6].
These matrix-form solutions usually only provide numerical
results with very little physical insights for practical system
operation and design. The FSMC is commonly used to study
the performance of some specific transmission schemes. The
three-state Markov channel is an alternative model for analysis
This research was supported in part by the National Natural Science Founda-
tion of China under Grants No. 61502428, No. 61502427, and No. 61379122,
and by the Zhejiang Provincial Natural Science Foundation of China under
Grants No. LQ15F010003, LY16F020034, and No. LR16F010003.
when the two-state Markov channel is inappropriate [3], [7]–
[9]. However, the delay performance from either generating
functions as two-state Markov channel or matrix-geometric
method as FSMC is untraceable.
In this paper, we focus on deriving analytic expressions
with clear physical interpretations for three-state Markov chan-
nels. The wireless fading channels with Poisson arrivals, and
three-state Markov modulated service processes (MMSP) are
modeled as M/MMSP/1 queues. We study the state transition
factor of three-state Markov channels, and derive a generalized
P-K formula for the three-state M/MMSP/1 from residual
service time and start-service probability.
A. Related Work
Fritchman investigated a three-state Markov model with
two Good states and one Bad state in [3], which captures
error-free intervals more precisely. In [7], a system deployed
with two antennas which fade independently was analyzed by
modeling the channel with three Markov states: both channels
are Good, only one of the channels is Good, and both chan-
nels are Bad. Yet another three-state Markov channel model
proposed in [8] was employed to analyze the performance of
Hybrid Automatic Repeat Request (Hybrid ARQ) systems, but
only numerical results were reported. Although three Markov
states provide a better representation than the two-state Markov
channel, but beside the numerical solutions obtained from
matrix-geometric method, the complexity of the model makes
the delay analysis mathematically intractable.
B. Main results
We provide closed-form expressions to evaluate the delay
performance of three-state Markov channels, and show that the
queueing delay is highly related to the derived state transition
factor. Our work is not only a straightforward extension of
the generalized P-K formula for two-state Markov channels
reported in [5], but we also make the following innovative
contributions:
1) We derive the state transition factor of three-state
Markov channels.
2) We introduce a linear approximation of the start-
service probability in terms of the server utilization
without invoking to the matrix-geometric method.
The remainder of this paper is organized as follows:
we describe the three-state Markov channel model and the
2016 International Conference on Computing, Networking and Communications (ICNC), Workshop on Computing, Networking and
Communications (CNC)
978-1-4673-8579-4/16/$31.00 ©2016 IEEE