16-05-2012, 03:38 PM
Mobile Agent Based Wireless Sensor Networks
Mobile Agent Based Wireless Sensor Networks.pdf (Size: 398.22 KB / Downloads: 26)
Abstract
Recently, mobile agents have been proposed for
efficient data dissemination in sensor networks. In the
traditional client/server-based computing architecture, data
at multiple sources are transferred to a destination; whereas
in the mobile-agent based computing paradigm, a taskspecific
executable code traverses the relevant sources to
gather data. Mobile agents can be used to greatly reduce the
communication cost, especially over low bandwidth links,
by moving the processing function to the data rather than
bringing the data to a central processor. This paper
proposes to use the mobile agent paradigm for reducing and
aggregating data in a planar sensor network architecture.
The proposed architecture is called mobile agent based
wireless sensor network (MAWSN). Extensive simulation
shows that MAWSN exhibits better performance than
client/server communications in terms of energy
consumption and the packet delivery ratio. However,
MAWSN has a longer end-to-end latency than client/server
communications in certain conditions.
Index Terms—mobile agent, energy efficient, aggregate,
data dissemination, wireless sensor networks
I. INTRODUCTION
Recent years have witnessed a growing interest in
deploying large numbers of micro-sensors that
collaborate in a distributed manner on data gathering and
processing. Sensors are expected to be inexpensive and
can be deployed in a large scale in harsh environments,
which implies that sensors are typically operating
unattended. Energy-efficient data delivery is crucial
because sensor nodes operate with limited battery power.
Currently, most energy-efficient proposals [1] in wireless
sensor network (WSN) are based on the client/server
computing model, where each sensor node sends its
sensory data to a back-end processing center or a sink
node. Because the link bandwidth of a WSN is typically
much lower than that of a wired network, a sensor
network's data traffic may exceed the network capacity.
To solve the problem of the overwhelming data traffic,
Qi. et al [3] proposed the mobile agent based distributed
sensor network (MADSN) for scalable and energyefficient
data aggregation. By transmitting the software
code, called a mobile agent (MA), to sensor nodes, a
large amount of sensory data can be reduced or
transformed into a small amount of data by eliminating
the redundancy. However, the operation of an MADSN is
based on the following assumptions: (1) the sensor
network architecture is clustering based; (2) each source
node is within one hop from a clusterhead; (3) much
redundancy exists among the sensory data which can be
fused into a single data packet with a fixed size. These
assumptions pose much limitation on the range of
applications that can be supported by an MADSN. Thus,
we will consider MA in multi-hop environments with the
absence of a clusterhead. In this paper, a MA is exploited
in two levels to reduce the information redundancy in a
planar WSN. Specifically, the MA is proposed to perform
the following functions: (1) eliminating data redundancy
among sensors by application context-aware local
processing at the node level; (2) eliminating spatial
redundancy among closely-located sensors by data
aggregation at the task level; (3) reducing communication
overhead by concatenating data at the combined task
level. The proposed architecture is called mobile agent
based wireless sensor network (MAWSN).
Extensive simulation-based comparisons between
MAWSN and client/server based WSN (CSWSN) show
that, depending on the parameters, MAWSN can
significantly reduce the energy consumption while
conditionally improving the end-to-end delay.
The rest of this paper is organized as follows. Section
II presents related work. We describe the MAWSN
architecture and design issues in Sections III. Simulation
model and experiment results are presented in Sections
IV and V, respectively. Section VI concludes the paper.
II. RELATED WORK
Recently, MAs have been proposed for efficient data
dissemination in WSNs [3-8]. In a typical client/server
based WSN, the occurrence of certain events will alert
sensors to collect data and send them to a sink node.
However, the use of MAs leads to a new computing
paradigm, which is in marked contrast to the traditional
client/server-based computing. The MA is a special kind
of software that propagates over the network either
periodically or on demand (when required by the
applications). It performs data processing autonomously
while migrating from node to node. Q. Wu et. al. [5]
14 JOURNAL OF COMPUTERS, VOL. 1, NO. 1, APRIL 2006
© 2006 ACADEMY PUBLISHER
presents a genetic algorithm based solution to compute an
approximation to the optimal source-visiting sequence.
The use of MAs in computer networks has certain
advantages and disadvantages [10], such as code caching,
safety and security, depending on the particular scenario.
Regardless, they have been successful deployed in many
applications ranging from e-commerce to military
situation awareness [7]. As described in [3], many
inherent advantages (e.g., scalability, extensibility,
energy awareness, reliability) of the MA architecture
make it more suitable for WNSs than the client/server
architecture. In [4], MAs are found to be particularly
useful for data fusion tasks in distributed WSNs.
In our previous work [2], we only presented a
description of data dissemination using MA in a planar
WSN, where MAs are exploited at three levels (i.e., node
level, task level, and combined task level). We extend
that work in this paper by proposing a scheme for MA
migrating in MAWSN. Then, we verify the efficacy of
MAWSN by extensive simulations.
III. SYSTEM ARCHITECTURE AND DESIGN OF MAWSN
In this section, we present the architecture and design
of the MAWSN. We first give an overview of the
network organization, and then describe MA assisted
information redundancy reduction at three levels. Lastly,
we present the operation of MAWSN in detail.
A. Overview
In the architecture illustrated in Fig. 1, a sink queries
multiple targets simultaneously by means of the MA. The
data in the target region is collected from the targets one
by one. The operation of the basic MAWSN will be
described in detail in Section III.C.
In traditional scenarios, multiple requests for different
physical information arrive at different times. We believe
that applications that require multiple different requested
tasks to be executed concurrently will become
widespread in the future. The reaction to a single task can
range from the simple return of a result by collaborative
processing among some sensor nodes (e.g., in the
application to obtain the population of the objects, the
system aggregates reports of individual objects right at
the point of data source and sends the already-aggregated
object counts), to a complex return of a large volume of
sensed data (e.g,. a picture captured by an image sensor).
Due to protocol overheads, the communication cost of
sending a longer message is usually less than sending the
same amount of data using many short messages. For the
concurrent tasks associated with small amounts of data,
we can perform them by a single packet carrying multiple
requests (e.g., one request for each task) and also
concatenate their results into a single packet to save
communication overhead. In Fig. 1, the combined
multiple tasks will be executed one by one, so that the
overall processing will take a longer time. If the
application’s minimum quality of service requirement
(e.g., latency bound) is not violated, especially in the case
that the target region is far from the sink node, the energy
savings of this combined execution can be significant.