![](bg2.png)
problem consists in the application of these three re-
configuration types: what is the gain that we can get
by using any hardware reconfiguration, or software
reconfiguration or also the protocol reconfiguration?
If we reduce the communication by applying recon-
figuration scenarios, can we win in term of energy?
We try in this paper to answer these questions by
defining three forms of reconfiguration for low power
RWSNs. We define a new zone-based multi-agent ar-
chitecture for RWSN where a communication proto-
col is well-defined for useful distributed reconfigu-
rations. We decompose the RWSN to a set of zones
where each one gathers a number of nodes. The radius
of each zone is a parameter to be defined by users ac-
cording to several characteristics of the followed tech-
nology. We define a Controller Agent (CrA) that han-
dles the reconfiguration strategies of the whole net-
work, and assign a Zone Agent for each zone (ZA) to
control the local reconfiguration scenarios. Each node
of a particular zone is controlled by a Slave Agent
(SA) that monitors the local reconfiguration scenarios
inside the node. This original multi-agent architec-
ture combines all possible reconfiguration forms to
be adapted for the environment where we minimize
the energy consumption. This adaptive architecture
is modeled by nested state machines in order to con-
trol the specification complexity. With our solution,
we gain in term of energy to be consumed by each
node and the number of exchanged messages between
nodes in the network. This architecture supports the
delegation between agents and controls the complex-
ity by providing hierarchical structure of RWSN. We
apply and simulate the paper’s contribution to a case
study to be assumed as a running example, and com-
pare our results to some related works in order to
show the originality of this architecture.
The paper is organized as follows: after introduc-
tion and background, Section 3 presents our position
between related works. Section 4 proposes a new def-
inition of RWSN to be explained on a case study. The
multi-agent architecture of the RWSN is proposed in
Section 5. Section 6 presents the coordination pro-
tocol between different agents. The simulation and
evaluation of the paper’s contribution is provided in
Section 7 before concluding the paper in Section 8.
2 BACKGROUND
We briefly present the formalism of finite state ma-
chines to be used in the following for the modelling
of RWSN. Finite State Machine (FSM) is an abstract
machine that can be in one of finite number of states.
It changes the behavior from a state to another by fir-
ing a transition in response to particular event. A FSM
is an efficient way to specify constraints of the over-
all behavior of a system (Samek, 2003). A classic
form of a FSM is a direct graph with the following el-
ements: G=(Q, Σ, Z, δ, q
0
, F) where: (a) Vertices Q is
a finite set of states (Q
1
,Q
2
,...,Q
i
) such that each state
(Q
i
) models a system’s behavior at an instant t, (b) In-
put symbols Σ is a finite collection of input symbols
or designators. This part of graph represents the finite
set of initial states, (c) Output symbols Z is a finite
collection of output symbols or designators. This part
of graph represents the final state of the system, (d)
Edges δ represents transitions from one state to an-
other as caused by input symbols, (e) Start state q
0
is
the start state q
0
∈ Q, (f) Accepting state(s) F: F ∈ Q
is the set of accepting states. We define Nested State
Machines as a set of FSM such that a state of one cor-
responds to another machine. This solution is useful
for the modeling of a complex system where the in-
formation should be modeled on different hierarchical
levels in order to control the complexity.
3 STATE OF THE ART
Today, several researches deal with RWSN where a
reconfiguration can be applied in three levels: hard-
ware, software and data routing (Bellis et al., 2005),
(Jie CHEN and LUO, 2009). Hardware recon-
figurations are defined in (Bellis et al., 2005) by
adding FPGA-based intelligent modules to nodes. In
(Kindratenko1 and Pointer, 2005), the wireless au-
tonomous sensor and networks of actors (WASAN)
define hardware reconfigurations as dynamic oper-
ations that model platforms of evaluation and assis-
tance. To model well the protocol reconfiguration,
the existence of reconfigurable interfaces is essential;
Harish Ramamurthy in (Harish Ramamurthy, 2005)
presents the ReWINS project (Reconfigurable Wire-
less Interface for Networking), to manage the recon-
figurability thanks to a ’Central Control Unit’. The
Reconfigurable Wireless Sensor Network for Struc-
tural Health Monitoring (M. Bocca and Eriksson,
2009), is also another project of RWSN. This proposi-
tion has the possibility to reconfigure the parameters
of the monitoring application (software reconfigura-
tion), depending on the needs of the end-user oper-
ating at the sink node. To optimize the radio trans-
mission of data and avoid interferences (protocol re-
configuration), each sink node establishes a reserved
communication link with each of the sensor nodes. In
(Vlado Handziski, 2005), the TWIST project (a scal-
able and flexible tested architecture for indoor deploy-
ment of wireless sensor networks) defines two recon-
ReconfigurableWirelessSensorNetworks-NewAdaptiveDynamicSolutionsforFlexibleArchitectures
255