Sunteți pe pagina 1din 11

WSN Based Data Collection Framework and Protocols for Disaster

Mitigation and Rescue Operation

Suman Saha , Mitsuji Matsumoto


GITS, Waseda University, Tokyo, Japan.
suman@asagi.waseda.jp, mmatsumoto@waseda.jp

ABSTRACT
Rapid development of Wireless Sensor Network (WSN) expedites the movement
of ubiquitous computing. Though people can deploy this WSN for disaster
mitigation and rescue operation after disaster, very few research works of WSN
consider about disaster management system. None of the existing works considers
one of the major effects of disaster such as some base stations might be collapsed
or unreachable due to catastrophe. Considering this deficiency, in this paper, a data
collection framework for disaster mitigation is demonstrated. This proposed
framework is designed based on large-scale hybrid networks of cellular networks
and WSNs. Considering the some base stations might be collapsed or unreachable
during or after disaster, in this framework, ARSs are deployed on the border
regions of every cell of cellular system. Sensor nodes of the corresponding cell
disseminate their data to the ARSs and ARSs route the received data to non-
collapsed base stations or nearby ARSs. In addition, in order to disseminate data
from WSN to ARSs, in this paper, a protocol for disaster mitigation is proposed.
This proposed protocol is a clustering and angular based routing protocol. Based
on sensor node residual energy and node centrality, cluster head selection is
performed. Simulation results prove that the performance of proposed protocol in
the proposed framework outperforms those of LEACH and PEGASIS protocol.
Though at first ARS is mentioned in [5], in this paper, the ARS placement strategy
is updated and this update is justified by the simulation results. Furthermore, a
small-scale data collection framework is proposed for rescue operation system
after disaster. The proposed framework for rescue operation considers that rescue
operators or first responders use some portable devices named cnode to collect
data from some particular disaster areas for rescue operation. During rescue
operation, cnode disseminates task over the sensor nodes of the disaster area by
directional antenna to inform the sensor nodes to send their sense data to the cnode.
For energy efficient and lower delay based data routing from these sensor nodes to
cnode, in this paper, a protocol for rescue operation is proposed. Performance
evaluation results prove that the performance of proposed rescue operation
protocol in proposed framework outperforms that of SENDROM [1].

Keywords: Wireless Sensor Network(WSN), Cellular network, ARS(Adhoc Relay


Station), Clustering, Routing.

1 INTRODUCTION Communications Commission) has required to


cellular systems the role for locating the position of a
Our disaster management system is not so terminal [10]. To satisfy the requirement, several
efficient comparing with our rapid development in technologies have been studied such as the Global
communication system. Current communication Positioning System (GPS) and so on [10], [11], [12].
systems could meet disaster management needs However, no system has been demonstrated to
sufficiently, as long as communication services are overcome the requirement.
uninterrupted. However, when these infrastructures Lots of research works ([2], [3]) use cellular
break down, the pipeline for essential information is network as a data gateway to convey data from
cut. A famous example of such a situation occurred disaster location. But almost in every case, after a
immediately after the Kobe earthquake in 1995 and catastrophe, rescuers cannot communicate with
the attack on the World Trade Center (WTC) in New survivors due to destruction in an established cellular
York on September 11, 2001. The FCC (Federal network. Main reason is collapsed or unreachable

Ubiquitous Computing and Communication Journal 1


base stations. Best example of this is 2004 Nigata future study.
earthquake in Japan. To overcome these deficiencies,
wireless sensor network can be deployed to make 2 PREVIOUS WORKS
communications among rescuers and survivors. In
addition, before disaster emergency workers need Few research works are for disaster management
varieties of sensing data, as an instance, which can ([1], [2], [3], [4], [14]). In [1], based on state-of the-
be applicable in case of tsunami. Because most cases art technologies, sensor network architecture is
earthquakes in the ocean areas precede tsunamis. If developed for rescuer operation after disaster. But
emergency workers can get the data of these the data dissemination technique is not efficient
earthquakes earlier, according to significant of the because of overall delayed route establishment time
earthquake they can evacuate or warn the people of and energy inefficiency. Although a protocol of ad
coastal region. For data collection from ocean, Under hoc sensor network for disaster management is
Water Sensor Network (hereafter, it is denoted as presented in [4] along with its energy efficiency,
UWSN) should be considered. Links in underwater which achieved by the combination of a low power
networks are usually based on acoustics wireless mode algorithm and a power aware routing strategy,
communications [6]. The unique characteristics of this network is not viable for real world
the underwater acoustic communication channel, implementation and its sensor prototype is
such as limited bandwidth capacity and high expensive. In [2], a location aware distributed sensor
propagation delays [6], require very efficient and networks is presented, but there is no mention about
reliable data communication protocols. These hybrid network interface such like sensor and
limitations make a gap between UWSN and cellular network interface. Though [3] is defined on
terrestrial wireless sensor network. a hybrid of sensor and cellular network, it does not
In this paper, the disaster management system is consider collapsed or unreachable base stations. In
divided into two subsystems: disaster mitigation and [2], clustering method does not regard energy
rescue operation. Here disaster mitigation subsystem efficiency, while in [3], routing and access protocol
stands for a large-scale heterogeneous system of for sensor network is not energy efficient though it
WSN, Cellular networks and Ad hoc Relay Station can be easily deployable in a cellular network
(ARS)s networks that can be used for 24 hours without considering collapsed base stations.
disaster surveillances, while rescue operation can In this paper, the proposed protocol for disaster
work in a small area for rescue operation after mitigation is a cluster-based protocol. Among the
disaster. Considering both of these subsystems of distributed cluster-based sensor networks protocols,
disaster management system, a data collection LEACH [9] is very popular. But clustering technique
used in LEACH wastes energy in terms of long run.
framework is proposed in this paper. In addition,
In LEACH, when clusters are created, each node of n
based on energy efficiency and lower delay,
autonomously decides if it will be a cluster head for
protocols for data collection from disaster areas for
the next round. The selection is stochastically: each
disaster mitigation and rescue operation are proposed. node determines a random number between 0 and 1.
Here proposed data collection framework for disaster If this number is lower than a threshold T(n), the
mitigation considers collapsed or unreachable base node becomes a cluster head. T(n) is determined
stations, which was not regarded as a significant area according to the equation
in previous works [2], [3], [4]. So in this framework,
static ARSs [5] are placed in bordering areas of cells P (1)
T (n ) =
for conveying data from collapsed base station areas ⎛ 1 ⎞
1 − P × ⎜ r m od ⎟
to their nearest base stations. Details motivations of ⎝ P ⎠
proposed works are discussed in section 2. Though in
proposed framework, UWSN framework is for nodes that have not been cluster head in the last
considered for ocean data collection, proposed WSN 1/P rounds, otherwise T(n) is zero. Here P is an a
protocols are only for terrestrial WSN. For UWSN, priori determined number that determines the
delay sensitive routing protocol [6] might be a good average number of cluster heads during a round, r is
choice. the number of the current round. Using this
The remainder of the paper is structured as algorithm, each node will be a cluster head exactly
follows. Section 2 will discuss related works and once within 1/P rounds. But in long run, there may
their drawbacks briefly. We present a data collection be lots of cluster heads in a densely small zone or
framework in section 3. Section 4 presents our there are no cluster heads in a large area of network.
proposed protocol for disaster mitigation along with So cluster formation is not uniformly distributed over
its performance evaluation. The proposed protocol the network. Another point LEACH considers the
for rescue operation and its performance evaluation number of times the node has been a cluster head up
is proposed in section 5. In section 6, we conclude to current round but not energy status of node. For
the paper and highlight the possible avenues for long run, energy is one of the main factors of every

Ubiquitous Computing and Communication Journal 2


sensor node with limited energy battery. Fig.1 depicts our considered network model.
Among the distributed techniques, Power Right side of the figure presents UWSN framework.
Efficient Gathering in Sensor Information Systems Here we count the UWSN, because we can collect
(PEGASIS), a chain-based routing protocol, further the data of earthquakes, which are originated in
enhances network lifetime among same class ocean-areas; and in ocean-areas, sometimes tsunamis
techniques by increasing local collaboration among follow the earthquakes. Due to lots of challenges of
sensor nodes [15]. In PEGASIS, nodes are organized UWSN such as overall infrastructure and
into a chain using a greedy algorithm so that each communication, it is different from terrestrial WSN.
node transmits to and receives from only one of its In this paper, the proposed WSN protocol is only for
neighbors. In each round, a randomly chosen node terrestrial WSN. Here, we consider that the delay
from the chain will transmit the aggregated data to
sensitive routing protocol [6] is the routing protocol
the base station, thus reducing the per round energy
of our UWSN.
expenditure compared to LEACH. But average delay
Left side of Fig. 1 presents a combined view of
per round makes PEGASIS an ineligible to disaster
data collection framework of terrestrial WSN, which
mitigation technique.
is used for disaster mitigation and rescue operation.
Form the discussion of the related works it is
For disaster mitigation, data dissemination network
evident that until now there is no large scale WSN
architecture is defined based on a hybrid model,
and cellular network based hybrid disaster mitigation
whose heterogeneity is confirmed by sensor, ARSs
network system, which can consider collapsed or
and cellular network. Deployed sensor nodes for
unreachable base stations. Normally used disaster
forming sensor networks and to collect information
management system runs for a short time and for a
from surrounding areas, and hybrid network, referred
small scale, so this does not have to consider energy
as an access network, hereafter, which combines a
scarcity of the sensor network. This type of system is
cellular network and ad hoc relay station (ARS) [5]
designed based on only lower data transmission
networks. We also initiate some concentric circular
delay. Due to rapid development of ubiquitous
zones in each cell for facilitating routing, which is
computing, we can deploy a large-scale disaster
called zoning. For better illustration of disaster
mitigation network that will track disaster
mitigation architecture, we present the Fig. 2. This
information for a long time. But in this case, energy
figure is presented with a cellular system with 7 cells,
efficiency and lower data delay should be considered
where some base stations are collapsed or
in together. To the best of my knowledge, in this
unreachable. In [5], author discusses the number of
paper, two major factors: energy efficiency and
ARSs needed to cover the entire system, and
lower data delay are considered in together for the
proposes a seed growing approach for the case only a
first time for the two proposed protocols of disaster
limited number of ARSs are available. Seed ARS is
mitigation and rescue operation. Furthermore, the
placed on each pair of shared edges along the border
proposed framework can be easily deployed in real
between two cells. According to [5], Fig. 3 depicts a
world cellular network. So, in the proposed
cellular system with 19 cells, where major
protocols’ part only WSN protocols are considered.
classification of cell is two: boundary cell and non-
Some researchers have studied a hybrid network for
boundary cell. But in Fig. 2, our network framework
cellular systems [5], [13], [16], [17]. The objectives
puts ARS in every edge without considering whether
of those systems are to achieve high speed, high
it is a shared edge or non-shared edge. In order to
capacity and wide area coverage by way of multi-
increase lifetime of sensor networks, we maintain
hopping. In [5], ad hoc relay station (ARS) is
every cell as a non-boundary cell. For performance
presented with its way of placement and various air
study, in section 4, we divide boundary cell class into
interfaces, though in this paper ARS placement
two subclasses: boundary cell with two non-shared
strategy is updated.
edges and boundary cell with three non-shared edges.
As an example, in Fig. 3, cell 1 is a boundary cell
3 DATA COLLECTION FRAMEWORK
with three non-shared edges, where cell 2 and cell 5
are boundary cell with two non-shared edges and
We have already mentioned, our framework of
non-boundary cell respectively. In section 4, by
disaster management system consists of two
evaluating performances of the protocol with respect
subsystems: disaster mitigation and rescue operation. to above-mentioned three kinds of cell, we justify
Former one is a general monitoring system for that we have to consider every cell as a non-
disasters’ tracking. For instance, the first sign of boundary cell. A WSN protocol for this proposed
tsunamis is earthquakes in the ocean-areas. If we disaster mitigation architecture is developed in
know about these earthquakes in real time, we can section 4. In this paper, we assume that every ARS
evacuate the residents of tsunami prone areas before supports two types of interface: ad hoc relay
the tsunamis taking effect on those areas. Latter interface and cellular interface. By ad hoc interface,
subsystem is for rescue purpose after disaster. ARS can communicate with other ARSs and sensor
networks. ARS uses cellular interface to

Ubiquitous Computing and Communication Journal 3


communicate with base stations of cellular network. integrator. Section 5 presents a protocol for this
proposed rescue operation architecture.

Satellite

Onshore sink

Surface station
Base Station Ad hoc Relay Station (ARS)
Zoning Sensor Network Node
Cnode
Underwater sensor nodes
Sensor Network Node

Figure 1: Network model

Non-shared Edge
1 A pair of Shared Edge
2 3
4 5 6 ARS
7 8
9 10 11
Boundary Cell
12 13
14 15 16
17 18
19 Non-Boundary Cell

Base Station Ad hoc Relay Station (ARS)


Figure 3: Cellular network with 19 cells
Zoning
Collapsed Base Station Sensor Network Node

4 DISASTER MITIGATION
Figure 2: A 7 cell-model for disaster mitigation
One of the major advances in the field of
For rescue operation, rescue workers or first
surveillance technology is the deployment of
responders work for rescue purpose using portable
distributed WSN. This has provided a means to
data collectors named cnode and mobile access
monitor areas, which are either unreachable or
points. When cnode queries the sensor nodes, they
hostile to human existence. So for disaster mitigation,
switch from idle mode to active mode and start
WSN might be a strong medium of disaster data
reporting the sensed data to the cnode. Sensor nodes
dissemination. In pervious section, we have
of rescue operation are different from those of
presented a framework for this and now in this
disaster mitigation. RFID sensor can be used as
section, we present a protocol for this WSN along
location tracking interrogator for rescue operations
with its performance evaluation.
[7]. Non-battery RFID tags or passive RFID tags are
ubiquitously placed along roadsides and function as
4.1 WSN Protocol For Mitigation
information-storage units. Information such as
Mainly based on energy efficiency and
location of refuges and safety assessments of
considering some base stations are unreachable, here
damaged buildings is remotely downloaded for the
we propose a protocol for terrestrial WSN that is
tags by the RFID integrators or sensors. The
applicable for disaster mitigation purpose.
maximum communication distance between the
4.1.1 Addressing scheme
passive or non-battery RFID tag and the interrogator
The addressing scheme in traditional networks is
is roughly 1.5 meters. Battery-driven RFIDS can
fixed x-y coordinate address. But in our proposed
extend the communication distance with the
protocol, the addressing format is <Location ID,

Ubiquitous Computing and Communication Journal 4


Node Type ID>. The Location ID identifies the over, a new cluster head is selected in the respective
location of a node that conducts sensing activities in cluster based on residual energy of sensor nodes and
a specified region of the network. Locations of these node centrality of that cluster (i.e. sub-area). Every
nodes are recorded in terms of polar co-ordinates (r, node transmits a message containing the information
θ) with respect to the base station as origin of the about weighted summation of its residual energy and
coordinate system (0,0). Based on r value, each node node centrality to its neighbors along with the node
falls in one of the logical zones, which are concentric. address and sub-area identification number. In Eq.
Also, every sensor node keeps records of its polar (4), Ci represents the weighted summation of its
co-ordinates (ri, θi) with respect to i-th ARS, where i residual energy and node centrality of ith
= 1,2,…,6. Protocol is applicable to stationary sensor
nodes and a stationary base station. The 1− γ
Ci = γ Ei + (4)
implementation can be extended to mobile and ad di
hoc networks with GPS enabling of the nodes, which
is an added expenditure. In that case, it is assumed node. In this equation, a constant γ is introduced to
that the base station keeps up-to-date information on weight the residual energy and node centrality,
the location of all the nodes in the network with the whose value is 0.7. A node i, receiving the Cj
help of GPS. Each node within the cluster is further information of all other nodes j, compares its Ci with
provided with a Node Type ID that describes the Cj of all other nodes j in the same sub-area (i.e.
functionality of the sensor such as seismic sensing, cluster). If its Ci is less than Cj of other nodes j of
thermal sensing, RFID sensing and so on. same sub-area, then it can detect the node having the
4.1.2 Clustering and scheduling maximum Cj from the received information, and
The proposed disaster mitigation protocol is a elect the corresponding j as the cluster head. Thus
cluster-based protocol. For facilitating clustering and the node having maximum Cj becomes the cluster
routing of the protocol, each cell is divided into some head for the particular sub-area. After being a new
concentric zones, what is already mentioned in cluster head, the cluster head updates the routing
chapter 3. Each zone area is divided into some sub tables of its nearby clusters and the cluster itself. Fig.
areas. Each sub-area has unique identification 4 represents the way of update the route. By
number. Every sensor node within this sub-area uses broadcasting the route discovery message with its
this identification number along with its own address sub area identification, new cluster head informs the
for communication purpose. nearby clusters that the cluster, which is representing
Each sub-area stands for a cluster, while a sensor the corresponding sub area, has changed its head. As
node acts as a cluster head for the respective cluster. a consequence, every nearby cluster updates its
Cluster head selection is based on two properties that routing table information for the corresponding
are residual energy and node centrality of the nodes cluster. When the new cluster head receives polar
of a cluster. Node centrality- a value that classifies coordinates from nearby clusters, it calculates the
the nodes based on how central the node is to the minimum angular deviation in shorter distance with
cluster. To find node centrality, every node keeps
the records of distances to the other nodes of
Algorithm Update_Route()
respective cluster and calculates the reciprocal of
{
sum of the squared distances of others node of the // All the distance and angle (in radians) measured
cluster. Since the transmission energy is proportional // with respect to nearest ARS of initiating cluster head
to d2, the higher value of the node centrality, the r0 ← new cluster head radial distance;
lower the amount of energy required by other nodes θ0 ← new cluster head polar angle;
// Transmit signal with E such that (Q≤E<2Q)
to transmit data to that node as a cluster head. For Broadcast_Route_Discovering_Message ( );
elaboration, see Eq. (2) and Eq. (3), where SN is the // receiving heads response with their polar
set of sensor nodes of the Nth sub area (i.e. cluster). // coordinates with respect to respective
// ARS as acknowledgement.
r1, r2,….., rn ← radial distances of responding heads.
di = ∑ d i2, j (2) θ1, θ2,….., θn ← polar angles of responding heads.
j ∈{ S N − i } Receive_Co-ordinates((r1,θ1),(r2,θ2),…,( rn,θn));
// calculation of minimum angular deviation with
1
Node Centrality of node i = (3) //respect to all i node, where ri < r0
di Calculate_Minima((|θ0-θ1|),( |θ0-θ2|),…,(|θ0-θn|));
// return polar co-ordinate of next head (with
Every cluster head maintains a routing table, // respect to respective ARS)
retrun (r, θ)
which keeps the location information of nearby }
cluster-heads’ locations in polar coordinates. When
energy level Ei of a sensor node i (who is acting as a
cluster head) reaches a threshold level or interval Figure 4: Algorithm for next head selection of
time between two consecutive clustering process is routing

Ubiquitous Computing and Communication Journal 5


the nearest ARS. The method of communication here, during disaster some base stations may be
between the nodes under one particular cluster head collapsed. So data sending from initiating cluster
and the cluster head is same as that of the LEACH. head to the nearest ARS is done through multi-
Radio interference caused by neighboring hoping by cluster heads. After receiving data, ARS
clusters that could impede the operation of any given sends data to one of its neighboring base stations if
cluster. This protocol makes use of code–division one of these two is not collapsed. If both base
multiple access (CDMA) codes to counteract this stations are collapsed, ARS sends data to the nearest
problem. Each cluster is assigned a spreading code ARS. Routing protocols of ARSs and base stations
that the nodes in the cluster use to distinguish their discussed in [5] might be the best choice for
data transmission from those of nodes in the proposed framework and disaster mitigation data
neighboring clusters. dissemination from ARSs to base stations. For
4.1.3 Routing integrated networks with heterogeneous technologies
After clustering and scheduling, cluster nodes such as cellular and ad hoc relaying, [5] describes the
send their data to the cluster head based on scheduled proposed signaling and routing protocols for iCAR to
slots. Once data from all sensor nodes have been establish and release bandwidth guaranteed
received, the cluster head performs data fusion on the connections possibly involving ARS relaying. Such
collected data, and reduces the amount of raw data protocols aim to addressing the QoS need of IP based
that needs to send to the nearest ARS. In general, real time applications.
data is routed to the base station by single hop or by
multi hop. But here, to reduce average energy Base
dissipation of sensor nodes and consider unreachable Cluster
or collapsed base stations, data is routed to the
nearest ARSs rather than base stations. And energy is
Head
distributed over the sensor network by keeping
Actual
multi-hop routing from cluster head to the nearest
ARS.
During routing, cluster head node forward its
data to a neighbor cluster head that is selected
ARS
applying the algorithm presented in Fig. 4. After
forwarding data to selected cluster head, the sensor Radial
node listens to that head to ensure that it repeats the
data packet i.e. implicit acknowledgement. If the
packet is not repeated, this indicates a transmission
failure. A transmission failure invokes route update
process (i.e. Fig. 4). Sometimes, a node fails to Figure 5: Data routing outline
forward data to none of its neighbors or it does not
have any negihbor within its transmission range. 4.2 Performance Evaluation
Usually this occurs when the node observes a void To evaluate the performance of the proposed
region between itself and the destination ARS. At protocol, using C programming language a
this situation, it broadcast the sensed data with the simulation program is developed based on proposed
maximum transmission power. It is already framework. Though here the performances of the
mentioned due to severe disaster some base stations protocol are studied in two ways: based on 3 types of
can be collapsed. So unreachable or collapsed base cell and comparing with others protocols such as
stations have to be considered during developing LEACH, PEGASIS; in both cases main evaluation
data routing technique for disaster management. In metric is system life.
this proposed protocol, clusters send data to ARSs 4.2.1 Based on 3 Types of Cell
and then ARSs send data to the base stations To evaluate the performance of the proposed
considering their status: collapsed or not. If the base protocol, the proposed protocol is simulated and
station of respective cell is collapsed, in that case compared its performance regarding three different
ARSs send their data to a live base station or ARSs. types of cell: non- boundary cell, boundary cell with
During disaster, any ARS may be collapsed. But 2 non-shared edges, and boundary cell with 3 non-
there is little chance to collapse all ARSs of a cell. shared edges, which are briefly discussed with Fig. 3.
Only one ARS is enough to convey data from sensor Here performance metrics are average energy
network of a cell to a base station. dissipation, successful data delivery and number of
Fig. 5 shows data routing outline, how data is live nodes. In a cell, 500 sensor nodes are uniformly
routed from data source cluster head to ARS. This distributed with center-located base station and
outline is a modified version of [2], where authors energy level of each node is assigned initially 5 Joule.
consider that data from cluster head is routed to base Here the simulation area is kept small regarding
station through multi hoping by cluster heads. But general sensor network simulation technique and

Ubiquitous Computing and Communication Journal 6


sensor network readers can get an idea easily from data messages than other types of cell. Non-
the following simulation results. Eventually, the boundary cell offers improvements in data delivery
radius of cell is 50 meter. Data packet size of sensor by factor of 12 percent and 49 percent over boundary
node is 1 kilobit. Number of operation is divided into cell with 2 non-shared edges and boundary cell with
rounds like LEACH protocol, and the number of 3 non-shared edges respectively.
rounds is 2500. In analysis, radio model of [9] is
used that has already been discussed briefly in
section 4.1.1.
In Fig. 6, the plot shows the total number of live 5

Average energy dissipation (Joule)


nodes that remain alive over the number of rounds of 4.5
activity for proposed protocol regarding three types 4
of cell. With non-boundary cell, 486 nodes out of
3.5
500 remain alive after the end of the simulation,
3
while corresponding numbers for boundary cell with
2 non-shared edges and boundary cell with 3 non- 2.5

shared edges are 341 and 125 respectively. 2

Eventually performance of non-boundary cell 1.5


justifies the update of ARS placement in chapter 3. 1
The improvement of gained through non- 0.5
boundary cell is further exemplified by the average
0
energy dissipation graph in Fig. 7, which shows the 0 500 1000 1500 2000 2500
average energy dissipation of the protocols under Number of rounds
study over the number of rounds of operation. The
plot clearly portraits that non-boundary cell has a
much more desirable energy expenditure curve than Figure 7: Average energy dissipation over the
those of others. After 2500 rounds, 4% energy number of rounds of operation
remained in boundary cell with 3 non-shared edges
and 21% energy remained in boundary cell with 2
non-shared edges , while in non-boundary cell,
remaining energy is 50% of the initial network 1200000
Number of data messages received

energy.
1000000

800000

600000
500
450 400000
Number of live nodes

400
350
200000
300
250
0
200
0 500 1000 1500 2000 2500
150
Number of rounds
100
50
0
0 500 1000 1500 2000 2500
Figure 8: Total number of data messages received at
Number of rounds the ARSs over the number of rounds

4.2.2 Comparison with PEGASIS and LEACH


Figure 6: Number of live nodes over the number of In this section, the performance of the proposed
rounds of operation. protocol is compared with those of PEGASIS and
LEACH protocol. Simulation environment is same as
Next the number of data messages received by section 4.2.1 except cell radius and number of
the sink (i.e. ARS) for three types of cells is analyzed. deployed sensor nodes. Here cell radius is 500 meter
Fig. 8 shows the total number of data messages (i.e. ten times bigger than previous environment).
received by the sink over the number of rounds of Like previous simulation environment,
activity. The plot clearly illustrates the effectiveness circumscribed hexagon is a regular hexagon.
of non-boundary cell in delivering significantly more Eventually the diagonals’ intersection of the hexagon

Ubiquitous Computing and Communication Journal 7


divides the hexagon into 6 equal areas. 1200 sensor PEGASIS LEACH Proposed Protocol
nodes are uniformly distributed over these 6 areas.

Number of rounds
Fig. 9 presents the node death percentage over 2500

the number of rounds of operation for the PEGASIS, 2000

LEACH and proposed protocol. With proposed 1500

protocol all nodes remain alive till 1498 rounds, 1000


500
while in PEGASIS and LEACH, all nodes remain
0
alive till 1005 and 498 rounds. After 1502 rounds, 0 20 40 60 80 100
there is no live node in PEGASIS, while proposed
Energy consumption percentage
protocol can maintain the system up to 2019 rounds.
Although 75% nodes with LEACH method dies
within 1000 rounds due to direct transmission, few
closer nodes of ARS can communicate up to round Figure 10: Energy consumption percentage over the
1902. number of rounds of operation.
Energy consumption percentage of the protocols
under study is demonstrated in Fig. 10, where
proposed protocol presents most desirable energy Table 1: Average time for a round.
curve for the proposed framework. In LEACH, 50 %
energy of initial energy of the whole network is Protocol PEGASIS LEACH Proposed
consumed by the 630 rounds of operation, while the Protocol
corresponding round’s number in PEGASIS and
proposed protocol are 855 and 1190. Though [42] Average 24 2.5 4.1
presents that PEGASIS is a protocol, whose energy time for a
consumption rate is near to optimal, these results round (sec.)
contradict with that because energy consumption rate
depends not only technique but also coverage areas. 5 RESCUE OPERATION
The further improvement of proposed protocol
over PEGASIS is exhibited by computing average For rescue operation, we assume, multiple
round time, which is present in Table 1. Due to sensors are deployed in each room of a building or in
making chain among the nodes, PEGASIS takes localized areas. These sensors may be several types
huge time to complete a round, while proposed such like RFID tag location sensor, high temperature
technique saves time by not doing cluster head sensor in the fire, and distortion sensor of the
selection for every round, which is executed after building. In the event a disaster happens, assuming
interval time between two consecutive clustering survivors are trapped in rubble, sensor nodes located
process or if current cluster head i energy level Ei near the survivors may detect them. After detecting
reaches to threshold level. Though here LEACH survivors’ information or other necessary
performs better than proposed protocol because of information, sensor nodes have to disseminate their
LEACH’s direct transmissions from cluster heads, in data to cnode in an energy efficient way.
real world LEACH implementation in a large
network will not viable. The performance of 5.1 WSN Protocol for Rescue
proposed protocol can be easily improved by adding In this section, mainly based on energy
more ARSs in every cell of proposed data collection efficiency and lower delay, we propose a WSN
framework. protocol only for rescue operation.
5.1.1 Addressing scheme
This node addressing scheme is similar to the
PEGASIS LEACH Proposed Protocol
scheme that described in section 4.1.1., where the
addressing format is <Location ID, Node Type ID>.
Number of rounds

2500
2000 But here node does not need to keep the record of its
1500 polar-coordinate with respect to ARSs.
1000 5.1.2 Routing
500 Two major issues are associated with this
0 proposed routing process, namely task dissemination
0 20 40 60 80 100
and data dissemination.
Node death percentage Task Dissemination: For rescue purpose, cnode
collects data from sensor nodes of disaster areas.
This data collection is similar to a query based data
Figure 9: Node death percentage over the number of collection, which is denoted as task based data
rounds of operation. collection. At the beginning, cnode broadcasts a task
message over the sensor nodes of a particular area.

Ubiquitous Computing and Communication Journal 8


Due to directional antennas in cnodes, sensor nodes individual task id. It only selects a next hop on the
in specific sections of the disaster area are invoked in following condition:
sensing system. In the task message, cnode includes -cnode changes its coordinate position.
its polar coordinate regarding base station as its -transmission failure.
origin. We assume, during task dissemination cnode After forwarding data to selected next hop, the
and sensor nodes are located in same cell of a sensor node listens to that hop to ensure that it
cellular system; and cnode has GPS facility though repeats the data packet i.e. implicit
here we consider that sensor nodes do not have this acknowledgement. If the packet is not repeated, this
facility. In addition, here considered sensor nodes are indicates a transmission failure. A transmission
stationary. For mobile sensor nodes, we can add GPS failure invokes next hope selection process.
facility with the sensor nodes, and in that case cost Sometimes, a node fails to forward data to none of its
will be expanded. After receiving the broadcasting neighbors or it does not have any negihbor within its
message from the cnode with particular task id and transmission range. Usually this occurs when the
polar coordinate, every sensor node updates its polar node observes a void region between itself and the
coordinate regarding current cnode as its origin. For destination cnode. At this situation, it broadcast the
instance, in Fig. 11, a sensor node S, whose polar sensed data with the maximum transmission power.
coordinate is (rs, θs) with respect to base station Fig. 12 depicts data routing outline.
O(0,0), calculates its polar coordinate with respect to
current cnode C(rc, θc) and that would be (CS, Radial Distance
∠SCD).

Actual Distance
S (rs,θs)

C (rc,θc) D Cnode Sensor node

O (0,0) Figure 12: Routing outline

Figure 11: Polar coordinate calculation of sensor 5.2 Performance Evaluation


node S with respect to current cnode C. To evaluate the performance of the proposed
protocol of rescue network architecture, we simulate
Data Dissemination: At the end of the task the protocol and compare its performance with that
dissemination, every sensor node in the specific area of SENDROM [1] architecture. Performance is
that is covered by directional antenna of cnode measured by system lifetime, initial route
senses data. The sensed data should be forwarded to establishment delay and recovery delay. Throughout
the cnode that broadcast the task. Here we use a the simulation, we consider 50 nodes are uniformly
multi-hoping technique for forwarding the sensed distributed in area of 160×160 meters and cnode is
data. For selecting the next hop, we use similar located on one of the corners. Operation is divided
algorithm that is presented in Fig. 4. But here all into number of events. Each individual task id
polar coordinates of sensor nodes are computed with represents individual event. Radio model used in [8]
respect to task initiating cnode. This is an angular is used for energy computation, where the power
deviation updated algorithm of [2]. Here a sensor dissipation is 35 mW in idle mode, 395 mW in
node selects its next hop based on minimum angular receive mode and 660mW in transmit mode. The
deviation on the way to the cnode. A sensor node data transmission rate is set to 1.6 Mbps.
broadcasts a route discovery message to its In Fig. 13, the plot shows remaining total energy
neighbors. After receiving the route discovery of the network over the number of events of activity
message, the neighbors who already received the for our proposed protocol and SENDROM. After 35
task message from the particular cnode send their events, with proposed protocol, remaining total
polar coordinates (where cnode is regarded as origin) energy is 73 percentage of initial total energy of the
to route discovery message initiating sensor node. network, while corresponding percentage for
And then the route discovery message initiating
SENDROM is 47. Proposed protocol offers
sensor node finds the minimum angular distance next
improvements in energy conservation by factor of
hop from the receiving polar coordinates of its
156 percent over SENDROM. In SENDROM, cnode
neighbors and updates its routing record. A sensor
node does not select a new next hop for every includes every node of the network during route
establishment phase. But some of these nodes that

Ubiquitous Computing and Communication Journal 9


waste their energy for route establishment are not SENDROM Proposed Protocol
14

Total number of failing nodes


used for data dissemination. In addition, angular
12
distance data dissemination technique of proposed
protocol invigorates the energy conservation process 10

of the network. 8
The improvement of life system through 6
proposed protocol is further exemplified by the
4
number of failing nodes in Fig. 14, which shows the
2
total number of failing nodes of the protocols under
study over the number of events of operation. The 0
0 5 10 15 20 25 30 35
plot clearly portraits that proposed protocol has a
Number of events
much more desirable failing nodes curve than that of
other. After 35 events, one-fourth nodes are failed to
communicate in SENDROM, while in proposed
protocol, this number is 5% of the total nodes. Figure 14: The number of nodes failed due to energy
Next we analyze the number of events depletion
transmitted to sink (e.g. cnode) for the SENDROM SENDROM Proposed Protocol
and our proposed protocol. Fig. 15 shows the total 35

Total number events transmitted


number of events transmitted to the sink over the 30
number of events of activity. The plot clearly
25
illustrates the effectiveness of proposed protocol in
delivering significantly more events than its to sink 20

counterpart. Proposed protocol offers improvement 15


in event delivery by factor of 135 percent over
SENDROM. 10

Fig. 16 represents the route establishment delay 5


and route recovery delay of the protocols under study.
0
Though route establishment delay and route recovery 0 5 10 15 20 25 30 35
delay are defined in the SENDROM system, in Number of events
proposed protocol, route establishment is not similar
to SENDROM. In SENDROM, cnode initiates route Figure 15: The number of events reported to the
establishment process and the process is completed network
for all nodes of the network by flooding the route
message. But in proposed protocol, this process is
14
not done in together. A node selects its next hop only
12
when it needs to that (see details in section 5.1.2).
Time (Sec)

10
For comparison purpose, in proposed protocol, we
8
define that total time for first time next hop selection
6
is denoted as route establishment time. When first
4
time selected next hop cannot forward data or
2
communicate, route update is needed; and the total
0
time to update route is denoted as route recovery SENDROM Proposed Protocol
time.
Route establishment delay Route Recovery delay
SENDROM Proposed Protocol
Remaining total energy of the

1
0.9
0.8
Figure 16: Route establishment and route recovery
network (*100%)

0.7
0.6 delay
0.5
0.4
0.3
0.2 6 CONCLUSION
0.1
0
0 5 10 15 20 25 30 35
In this paper, we present a framework for
Number of events disaster mitigation and rescue operation that can be
easily implemented in real world communication
system. Though current real world communication
system does not support various effects of disasters
Figure 13: The energy available in the network or catastrophes, while our proposed system removes

Ubiquitous Computing and Communication Journal 10


those drawbacks. In addition, here we present Transactions on Networking, pp. 2-16,
protocols for sensor networks, which are based on February,2003.
proposed framework, consider energy efficiency of [9] W. B. Heinzelman, A. P. Chandrakasan, and H.
Balakrishnan, “An Application-Specific
the sensor nodes strongly along with lower route Protocol Architecture for Wireless Microsensor
establishment delay. We update the ARS placement Networks,” IEEE Trans. Wireless Commun.,
of actual ARS placement in [5]. The simulation vol. 1, no. 4, , pp. 660–70, Oct. 2002.
results justify this update. In future, our future [10] Jeffrey H. Reed, Kevin J. Krizman, Brian D.
Woerner and Theodore S. Rappaport, "An
avenue is to develop a semantic query based protocol Overview of the Challenges and Progress in
for disaster management. Meeting the E-911 Requirement for Location
Service", IEEE Communications Magazine, no.4
7 REFERENCES pp.30-37, April 1998.
[11] J.J. Caffery and G.L. Stuber, "Overview of
[1] Erdal Cayirci, and Tolga Coplu, “SENDROM: radiolocation in CDMA cellular systems",
Sensor networks for disaster relief operations Communications Magazine, IEEE , vol.36, no.4
management”, Springer Journal on Wireless pp.38-45, Apr 1998.
Networks, vol. 13. no. 3, May 2007. [12] T. Kleine-Ostmann and A.E. Bell, "A data
[2] G. Ranjan, A. Kumar, G. Rammurthy, and M.B. fusion architecture for enhanced position
Srinivas, "A natural disasters management estimation in wireless networks", IEEE Comm.
system based on location aware distributed Letters, vol.5, no.8 pp.343-345, Aug 2001.
sensor networks," IEEE International [13] Y. Yamao, T. Otsu, A. Fujiwara, H. Murata and
Conference on Mobile Adhoc and Sensor S. Yoshida, "Multi-hop radio access cellular
Systems Conference, vol., no.pp. 180- 182, Nov. concept for fourth-generation mobile
7, 2005. communications system", The 13th IEEE
[3] T. Fujiwara, H. Makie, and T. Watanabe, "A International Symposium on Personal, Indoor
framework for data collection system with and Mobile Radio Communications, vol.1,
sensor networks in disaster circumstances," Int. no.pp. 59- 63 vol.1, 15-18 Sept. 2002.
Workshop on Ad-Hoc Networks, vol., no.pp. [14] Osamu Takizawa, “Ubiquitous communications
94- 98, June 2004. technology for disatser mitigation”, ” Proc. of
[4] N. Pogkas, G. Karastergios, C. Antonopoulos, S. NICT, pp-235-241, Sep2005.
Koubias, and G. Papadopoulos, "An ad-hoc [15] S. Lindsey, C. Raghavendra, and K. M.
sensor network for disaster relief operations," Sivalingam, “Data Gathering Algorithms in
10th IEEE Conference on Emerging Sensor Networks using Energy Metrics,” IEEE
Technologies and Factory Automation, vol.2, Trans. Parallel and Distrib. Sys., vol. 13, no. 9,
no.pp. 131- 139, 19-22 Sept. 2005. Sept. 2002, pp. 924–35.
[5] Hongyi Wu, Chunming Qiao, S. De., and O. [16] Haiyun Luo, Ramachandran Ramjee, Prasun
Tonguz, "Integrated cellular and ad hoc relaying Sinha, Li (Erran) Li, and Songwu Lu, “UCAN:
systems: iCAR," IEEE Journal on Selected a unified cellular and ad-hoc network
Areas in Comm., vol.19, no.10 pp.2105-2115,
Oct 2001. architecture”, Proc. of Mobicom, pp 353 - 367,
[6] D. Pompili, T. Melodia, and I. F. Akyildiz, Sep 2003.
"Routing Algorithms for Delay-insensitive and [17] Randeep Bhatia, Li (Erran) Li, Haiyun Luo, and
Delay-sensitive Applications in Underwater
Sensor Networks," Proc. of ACM Conference on Ram Ramjee, “ICAM: Integrated Cellular and
Mobile Computing and Networking Ad Hoc Multicast”, IEEE Transactions on
(MobiCom), Los Angeles, CA, September 2006. Mobile Computing, Vol. 5, No. 8, August 2006.
[7] O. Takizawa et al., “Research and Development
of an RFID Based Disaster-Relief System,”
Proc. of NICT, pp-259-277, Sep2005.
[8] C. Intanagonwiwat, R. Govindan, D. Estrin, and
J. Heidemann, and F. Silva. Directed Diffusion
for Wireless Sensor Networking. ACM/IEEE

Ubiquitous Computing and Communication Journal 11

S-ar putea să vă placă și