Communication Availability in Communications-Based TCSs ◾ 103
6.5 Modeling Data CommunicationSystem
Behavior with DSPNs
e main problem in our proposed CTMC model in Section 6.3 is that we
assume that the time between two successive handos is exponentially distributed.
However, the distance between successive APs and the train speed both may not
follow exponential distribution. In order to show the soundness of the approxima-
tion in our CTMC model, we formulate the data communication system behavior
with DSPNs.
e motivations of selecting the DSPN approach are as follows: (1) As a kind
of SPN, DSPN provides an intuitive and ecient way of describing the link
failure behaviors in WLAN-based data communication systems, especially facili-
tating hando behavior modeling. (2) DSPN allows timed transitions to have an
exponentially distributed time delay or a deterministic timed delay, which can
accurately model the situation when the time between two successive handos is
relatively constant. (3) SPN has been successfully used to analyze system avail-
ability in safety-critical on-demand systems [20] and industry plants [21], among
others.
6.5.1 Introduction to DSPNs
A Petri net is a directed bipartite graph with two types of nodes called places and
transitions, which are represented by circles and rectangles (or bars), respectively
[22]. Arcs connecting places to transitions are referred to as input arcs, whereas
the connections from transitions to places are called output arcs. A nonnegative
integer (the default value is 1) may be associated with an arc, which is referred to as
multiplicity or weight. Places correspond to state variables of the system, whereas
transitions correspond to actions that induce changes of states. A place may con-
tain tokens that are represented by dots in the Petri net. e state of the Petri net
is dened by its marking, which is represented by a vector
Ml
k
(, ,...,
12
, where
p
is the number of tokens in place
k
. Here,
is a mapping func-
tion from a place to the number of tokens assigned to it. A transition is enabled
if the number of tokens in each of its input places is larger than the weight of its
corresponding input arc. An enabled transition can re, and as many tokens as
the weight of the corresponding input arcs are moved from the input place to the
output place.
SPNs are one kind of Petri nets in which an exponentially distributed time delay
is associated with each transition [22]. Generalized SPNs (GSPNs) extend the mod-
eling power of SPN and divide the transitions into two classes: the exponentially
distributed timed transitions (represented by blank rectangles), which are used to
model the random delays associated with the execution of activities, and immediate
transitions (represented by bars), which are devoted to the representation of logical