Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
CHAPTER 5
CHAPTER 5
Cross Layer Cluster Approach for Energy Efficient Data Gathering Scheme for Wireless Sensor Networks
5.1 INTRODUCTION
Research in Ad hoc wireless networks and wireless technology has
played a major role in the development of smart sensing. WSNs have a
wide variety of applications in industrial process, target detection and
tracking, healthcare monitoring, military applications, public places, traffic
surveillance and so on. Traffic congestion is a major problem in
metropolitan cities. This problem of congestion needs to be alleviated.
Some solutions to this problem include location of the specific vehicle and
then monitoring its movement, real time automatic traffic data collection
during rush hour for efficient management and soon. The work here
describes cross layer approach for optimizing data gathering from sensor
nodes and in turn increasing the lifetime of the network [89]. It is
important to design a sensor node that is power efficient as the power
source in WSNs is battery operated and has limited power capacity. The
design of the sensor node should be power efficient as the battery is
limited. Cross layer cluster approach makes use of both routing and MAC
layers information to reduce congestion, increase packet delivery ratio and
minimize energy usage. The time spent on data movement can be reduced
83
by using cross layer approach for energy efficient [41, 46] MAC layer in
WSN. MAC protocol, clustering, routing and various other attributes and
parameters need to be considered for cross layer design approach. While
designing sensor networks, a trade off between those parameters and
attributes are required. Some of the important attributes include scalability,
energy efficiency, channel utilization, adaptability to network topology,
node deployment and collision avoidance. Of all the attributes mentioned,
the most important attribute is energy efficiency [71] [72]. Three important
domains of energy consumption that takes place in sensor networks are
sensing, data processing and communication. The reason MAC layer is
considered here in cross layer is because it controls directly the radio of the
nodes in the networks. Thus, the design of MAC protocols plays an
important role in node's energy consumption which influences the lifetime
of sensor networks. Packet overhead, overhearing, over emitting, idle
listening and collision are the major sources of energy waste of MAC
layer.
Single layer communication protocol for designing optimal
strategies for sensor networks has been proposed by most researchers. The
traditional approach does not include overall performance improvement.
Protocols designed for MAC layer are power efficient and are categorised
84
as: on demand wakeup and schedule wakeup. On demand wakeup
protocols [48] make use of some form of out-banding signaling technique
through a separate radio to wake up nodes with adding extra cost to the
network deployment. Scheduled wake up MAC schemes can be further
classified as asynchronous schemes and synchronous schemes. Clock
synchronization is not needed in asynchronous approaches. It is easier to
implement and can ensure network connectivity even in highly dynamic
networks. Asynchronous schemes are simple to implement. Asynchronous
schemes are less efficient than synchronous schemes and cannot provide
guarantees on the worst-case delay. For synchronous scheduled wake up
protocols, TDMA and duty cycling are the most commonly used
techniques. Although TDMA protocols are usually designed to create
contention free medium access for communication, they can schedule the
wakeups of the sensor nodes, as well. If the network topology is dense and
dynamic than TDMA protocols require slot allocation and management,
which can be difficult. MAC protocols for WSN can also be broadly
classified as contention based and contention less protocols. The nodes can
transmit without having any predetermined time assigned to them in case
of contention based MAC. The nodes will compete for a shared channel.
Collisions exist if multiple nodes access the medium simultaneously. A
mechanism to avoid collisions has to be provided by the protocol. One
85
such protocol is standardized IEEE 802.11 Distributed Coordination
Function (DCF). In order to reduce the energy consumption of idle
listening, various kinds of schedules and wake up techniques as on
demand, asynchronous schemes, duty cycling is being adopted. Examples
of contention based MAC protocols are S-MAC, D-MAC, T-MAC. The
MAC protocols mentioned are about wake-up schemes and listen/sleep
schedule allocation and reduce processing time in a node [22]. Contention
less MAC protocols are normally based on TDMA approaches. These
protocols have a natural advantage of energy conservation in comparison
with contention based protocols since there are no collisions. Schemes
based on the several sources of power consumption also exist. One such
example is the purpose of saving energy by minimizing the idle time and
also by optimizing the route algorithms [50] is achieved. Adjust the power
by optimizing the physical layer in order to transmit proper power. These
schemes are for each layer, and they have their own limitation. If the
network layer is being optimized then, the parameter of MAC layer may
change. Hence it is not comprehensive to optimize a signal layer. No
research exists about how to reduce energy consumption with the cross
layer design approach and also incorporating cluster formation mechanism
helpful in data gathering.
86
5.2 RELATED WORK
Initially lot of research work on energy consumption was based on
single layered approach. Gradually as the efficiency of sensor network had
to be improved and to achieve the goal of saving energy, cross layered
design scheme was incorporated by researchers.
Cross layered [45] approach that combined the MAC layer and the
routing layer was started by Van H L et al. [50]. The PHY +MAC
+Routing scheme was incorporated by Kong I Y, Hwang W J [52] by
considering the power control in the physical layer, the scheduling in MAC
layer, and the overhead in routing the layer. W. Ye [61] explained about an
energy-efficient MAC protocol for WSNs. There exists some research on
the cross-layer interaction and design in developing new communication
protocols [39]. However, as discussed in [39] in detail, these works either
provide analytical results without any communication protocol design, or
perform pairwise cross-layer design within limited scope, e.g., only routing
and MAC layers [75], which do not consider all of the networking layers
involving in the communication in WSN such as transport, routing,
medium access and physical layers. S-MAC (Sensor-MAC) [49][76] is an
energy efficient protocol, but it introduces significant delivery latency and
provides poor traffic throughput. S-MAC uses three novel techniques to
87
reduce energy consumption and support self-configuration. To reduce
energy consumption in listening to an idle channel, nodes periodically
sleep. Neighbouring nodes form virtual clusters to auto-synchronize on
sleep schedules. Inspired by Power Aware Multi-Access Protocol
(PAMAS), S-MAC also sets the radio to sleep during transmissions of
other nodes. Unlike PAMAS, it only uses in-channel signaling. Finally, S-
MAC applies the message passing to reduce contention latency for sensor-
network applications that require store-and-forward processing as data
move through the network. Shu Du [53] presented a new duty-cycle MAC
protocol, called as Routing enhanced MAC protocol (RMAC) [53]. RMAC
exploits cross layer routing information in order to avoid these problems
without sacrificing energy efficiency. RMAC can deliver a data packet
multiple hops in a single operational cycle. Cross layer clustering approach
would save energy [77]. In this work, it is investigated that, there is a
decrease in energy consumption among nodes in the network using cross
layer cluster approach of WSN.
5.3 PROPOSED CROSS LAYER APPROACH DESIGN
MAC protocol and routing are combined to reduce data moving
between cross layer in the proposed work. The MAC protocol S-MAC is
used in wireless sensor networks in which mobile node execute both
88
receiving and sending of MAC frames at the same interface. IEEE 802.11
DCF protocol is used as MAC protocol with cross layer approach as better
results are obtained as IEEE 802.11 DCF produces RTS and CTS frames
which consume less energy. The same cross layer approach is used with S-
MAC protocol and avoiding data movement between MAC and network
layer. Every node must perform data movement between NIC (Network
Interface Card) memory and host memory twice and looking up the routing
table frame decapsulation and encapsulation. The address of the next hop
is decided by the network layer. These operations consume a lot of
resources of the node such as processing time, memory and energy. But
still packets have to be delivered to the desired destination using minimum
energy. Clustering approach is used to find the nodes nearer to the Base
Station (BS) which work in a longer time and consume much energy.
Nodes near BS will exhaust their energy in a short time in this condition.
The nodes far away from BS in one hop are considered, and the nodes
wake up randomly in a period, time-synchronized sensors form on-off
schedules that enable the sensors to be awake only when necessary.
In order to minimize processing delay between layers, packet is not
required to send data twice through same interface. MAC layer does the
packet decapsulation of the network layer header. Only a portion of
89
network layer header is sent to the network layer to process and establish a
route while allowing data frames to reside in NIC memory itself. All the
nodes in the network need not have to move the data packet between the
MAC and the network layer. Whenever MAC layer gets a data frame, it
checks the MAC destination address field in the frame and finds out
whether it is intended to it or not. It stores the MAC frame and removes the
MAC header to further examine it if the data frame belongs to it. MAC
layer itself removes the network layer which is supposed to be done by the
network protocols at network layer. Instead of sending complete data
frame, only network layer header with the destination address is sent to
help in finding the next hop address and then transferring next hop MAC
address to the MAC layer. Now, MAC layer encapsulates a new frame
with next hop address and the data frame reserved in NIC memory. Thus,
the amount of data moving between layers, processing time and end-to-end
delay is reduced. Once after finding the next hop, the packets have to be
routed through nodes to the desired destination. The route must be chosen
before fransmitting the packets, and choosing the correct route is the first
phase using a cluster approach in WSNs. Sensor nodes dynamically create
on-off schedules in a way that the nodes will be awake only when needed
and asleep the rest of the time. This can be done in two distinct phases in
routing:
90
1, The setup and reconfiguration phase: This phase takes place during the
initializing of the network. And it is shorter in comparison to the steady
phase. The routing failure may happen because of the impact of the
external environment. It is necessary to reconfigure the route when it fails.
The goal of this phase is to set up the schedules that will be used during the
steady state phase and the setup phase corresponds to writmg the routing
tables.
2. The steady state phase: It takes place between consecutive setup and
reconfiguration phase. The nodes will transmit the data steadily in this
phase.
WSNs are application-oriented. The amount of data needs to be
transferred is very huge in applications like audio and video monitoring.
The data field in MAC frames will be large too. These applications are
usually real time applications. Thus, this approach is helpful for these
applications. The efficient S-MAC protocol and cluster approach for
minimizing processing time and energy consumption has been considered.
5.4 SIMULATION SETUP
NS2 [40] discrete simulation is used to evaluate the performance of
cross layer cluster approach. The setup consists of a testbed of 25 nodes
randomly deployed in the area of 1500 m x 1500 m. 10 nodes are 91
considered to be one hop away from the base station out of 25 nodes. The
nodes are finally increased to 100. The MAC protocol S-MAC is
incorporated into CLD-SMAC (Cross Layer Design S-MAC) and consume
energy for reading, writing and comparing a bit. With the increase in
processing time, energy consumption also increases. Processing delay
includes header processing, data movement between layers and routing
table lookup. In addition, it is helpfiil to use cluster approach for routing
data between the nodes whose residual energy is greater. The simulation
parameters are listed in table 5.1.
Simulation Parameters Number of Nodes Simulation Time Initial Energy Transmit Power Receive Power Idle listening Power Number nodes one hop near to BS/total nodes Transmission Range
Value 25, 50, 75,100 60 seconds 1000 joules 3.0 Watts 2.0 Watts 0.04 Watts 10/25,20/50, 35/75 and 45/100 25 meters
Table 5.1 Simulation parameters used for cross layer design
5.5 RESULTS and DISCUSSIONS
Evaluation of node processing delay and end-to-end processing
delay of CLD-SMAC is the primary purpose of simulation. A UDP
92
connection between the source node and the base station is estabHshed. At
1 .Os, UDP packets are sent to the base station from the source node. Packet
size considered is 50 bytes. It has been assumed that the period is 5.0s, so
the sensor network needs to work for 10 periods and the number of the
nodes closer to the base station are 10.
The simulation result is shown in Figure 5.1. Figure 5.1 shows that
the performance of cross layer approach is better than a single layer
approach. Every node in the network is capable of forwarding traffic to any
destination and always it tries to find the best route to every packet.
The cross layer cluster approach is not only useful to route in the
best path, but it also saves energy and is as depicted in Figure 5.2.
in^n -.
1000 ^
> 980
c 960 -"n
-1 940
°̂ 920 -
900
880
-
K \ y V̂ N
*''' ' '*"'^ ^ ^
V > \ I ^ 7^ ^ ? ^ 7. / s -MAC
LD-SMAC ,
1 2 4 5 6 8 10
nodes
Figure 5.1 Residual Energy of each node which is closer to Base Station
93
In this scenario, a network with 25 nodes randomly distributed along
with 10 nodes placed one hop away from the Base Station. The length IP
header is 20 octets with packet length of 1156 octets that are the average
value between 0-2312. The header processing is done in the MAC layer,
and header information is sent to network layer to find the address of the
next hop. Figure 5.2 indicates the nodal processing delays of S-MAC and
CLD-SMAC. Simulation shows that CLD-SMAC outperforms S-MAC in
terms of the nodal processing delay. Thus, it is desirable that the energy
efficiency of CLD-SMAC will be better than that of other MAC protocols.
From Figure 5.3 and Figure 5.4, it is very clear that CLD-SMAC
outperforms in terms of the end-to-end processing delay for 50 and 100
nodes. The improvement when the network density changes have been
observed.
16000 1
14000
_g- 12300 • •8
gi 10000
S 8300 -
1 6000 -
Z 4300 -
2000 -
0 -
^^-^S/zz'-z.
-^^^H-- -
B _ . ^ ^ ^ H
^ ^ B Hcaacr KToccssing ^^^1 -^^1 -/ :. ^ ^ ^ ^ ^ ^ ^ ^ H
S-MAC CLD-SMAC
• •
Figure 5.2 Node Processing Delay for S-MAC/CLD-SMAC Protocol
94
innn -.
995
990 >• 5 985 -
1 980 -o Z 975 -c
UJ
970
965
qfin -
- . - . .. .__ . . . '
^^^^HH^l "̂ ^̂ H^̂ ^H"
^H ^M
: :;^H ^H
_^:.::::^^B ,,7 ^ 'T^~^^ ^M .r^z-:
^̂ H
m
——------
" / -"
T T ^ — - T
S-MAC CLD-SMAC
Figure 5.3 End-to-End Delay for S-MAC/CLD-SMAC Protocol with 50
nodes
1000
yyb
E c >. m V Q •a c UJ 1
0 4^ Tl UJ
990
985
ysu
97S
'J/U
965
960 S-MAC CLD-SMAC
Figure 5.4 End-to-End Delay for S-MAC/CLD-SMAC Protocol with 100
nodes
95
5.6 CONCLUSION
Node energy has become a key problem and interesting research
area in WSN. Various MAC layer protocols have been proposed for sensor
networks. Many schemes have improved the existing MAC protocols and
routing algorithms to extend the lifetime of WSNs. Depending on network
layer protocols, the routing path could be chosen. The data moving,
however, causes a portion of energy consumption that cannot be neglected
for each packet transferring. The basic idea is that a node should wake up
only when another node wants to communicate with it. An energy efficient
cross-layer cluster scheme for prolonging lifetime of WSNs by integrating
S-MAC MAC protocol is proposed. The goal to save energy is achieved
through averaging the energy of the nodes consume and reducing the idle
listening and minimizing the collision.
There is a significant improvement regarding the nodal processing
delay and the end-to-end processing delay and is as shown in simulation
results. It has also been proved that the proposed scheme is feasible for the
sensor network. Results show that there is a good progress in energy
saving through processing delay and finding out residual energy for the
entire sensor network. The performance of proposed cross layer cluster
approach shows better results than before compared with S-MAC.
96
There are open research problems on optimizing energy
consumption through cross layer approach with different protocols for
WSN based on this direction. The cross layer architectures enable vital
information exchange between the layers to fine tune protocol parameters
and to set route selection criteria. The network lifetime can thus be
improved by reducing energy consumption of nodes.
97