You are on page 1of 16

Wireless Netw (2016) 22:335–350

DOI 10.1007/s11276-015-1025-x

A review on energy efficient protocols in wireless sensor networks


Sarika Yadav1 • Rama Shankar Yadav1

Published online: 6 August 2015


 Springer Science+Business Media New York 2015

Abstract In past decade, wireless sensor networks have 1 Introduction


gained attention by researchers, manufacturers as well as
the users for remotely monitoring tasks and effective data On 2nd August 2007, unexpectedly a highway bridge col-
gathering in diverse environment. The wireless sensor lapsed into fast flowing Mississippi River in Minnesota, nine
nodes are tiny battery powered devices having limited people were killed in the event. The potential causes behind
lifetime, hence for longevity and reliability, the foremost the accident were wear and tear, weather conditions and load
concern is minimizing energy consumption and maximiz- on the bridge as determined by National Transportation
ing network lifetime while designing protocols and appli- Safety Board, Minnesota [1]. In another incident, on 2nd July
cations. In this paper, we review the main design issues 2013, fire broke down in the forests of California where 19
based on the model of wireless sensor networks: structure- firefighters succumbed to death due to choking and high rise
free and structured for data collection and aggregation in temperature [2]. These incidents can be taken care off if
where role of clustering and routing is discussed for energy ample information such as weather forecasting, temperature,
conservation and enhancing network lifetime. These design high density smoke zone, load conditions are provided and
strategies are the foundation of any networking protocol necessary actions have been taken in time. With recent
from the energy saving point of view. A comprehensive advances in Micro-Electro-Mechanical Systems (MEMS)
tabular overview of different approaches under structure- technology, wireless communications and digital electronics
free and structured wireless sensor networks for data col- have facilitated the development of low-cost, low-power
lection and aggregation, clustering and routing is presented multi-functional, small-size wireless sensor nodes proficient
with key issues. in observing the changes in physical environment, process-
ing the sensed data, taking decisions and performing
Keywords WSN  Data aggregation  Network lifetime  appropriate actions. Wireless sensor networks are used as an
Cluster head  Base station  Network connectivity and integral part of systems such as health monitoring, battle-
coverage field surveillance, missile target tracking, natural disasters
like floods, earthquake, fire-break down, thunder storm etc.
[3]. In a wireless sensor network, a sensor node can be made-
up by a small size battery, storage, a mote (processor/radio
board) and depending on its function, sensor/data acquisition
board that supports the detection and measurement of envi-
& Sarika Yadav ronmental changes [4]. Wireless sensor boards offered in the
sarikayash18@gmail.com market have a variety of applications such as light sensors,
Rama Shankar Yadav barometric pressure sensors, accelerometers, humidity sen-
rsy@mnnit.ac.in sors, temperature sensors, GPS modules, acoustic sensors,
1 magnetic RPM sensors, magnetometers, wind speed sensors,
Department of Computer Science and Engineering, Motilal
Nehru National Institute of Technology, Allahabad 211004, moisture sensors, solar radiation sensors, temperature sen-
India sors, seismic sensors and rainfall meters [4]. These sensor

123
336 Wireless Netw (2016) 22:335–350

nodes perform three basic tasks: (1) sampling of a physical deployed with a primary objective of replacing wired net-
quantity for the specified surrounding environment, (2) works in the area where either deployment of wired net-
processing and storing (if required) of the sensed data, and work was complicated due to geographic locations or
(3) transferring the sensed data to a data collection point costly in terms of its utilization and random requirements.
called base station or sink node all the way through wireless The major concern in a wireless sensor network is to
communication [3]. These sensor nodes can communicate extend the network lifetime in addition to its robustness as
with a gateway unit (base station) that is proficient in com- compared to traditional wired networks where the center of
municating with other computers by means of additional attention is to maximize channel throughput with minimum
networks, such as a LAN, WLAN, WPAN and the Internet. node exploitation.
The base stations are fixed nodes and are connected to other While the base station in wireless sensor networks can
nodes forming the infrastructure for wireless sensor network. have continuous power supply, a sensor node has limited
At the base station, a snapshot of the adjoining region is resources, for e.g. limited processing capability, limited
obtained by assembling the received sampling values for the memory, limited battery backup and unpredictable har-
corresponding locations. The sensor nodes communicate vesting capabilities which determines its own longevity
with each other and the base station through radios and the and whole network lifetime also. A sensor node’s lifetime
base station to exchange data with applications to accom- is restricted by backup available at a time, amount of
plish their assigned operation. Due to their unconventional computation as well as communication it carries out
behavior, wireless sensor networks stand for a number of depending on the distance with communicating node.
unique technical problems during data processing, commu- Once power backup exhaust below threshold, it is treated
nication and sensor management bare to various challenges as dead node and is disconnected. Therefore, energy con-
in network detection, network control, mutual information servation techniques present a major research challenge for
processing/exchange, bandwidth bottleneck and energy wireless sensor networks and a number of research papers
availability as well as consumption constraints [5] due to the proposed solutions to one or more of the above problems.
following factors [6] as a part of wireless sensor network: As available in research literature, Fig. 1 shows the
percentage of energy consumption by various activities
• Ad hoc deployment The regions which have no
performed by a sensor node during its active time. From
infrastructure at all, for example in a forest, the sensor
the Fig. 1, it is noticeable that more than 50 % of energy is
nodes have to identify their connectivity being
spent during communication activity by a sensor node. For
deployed by tossing from an aeroplane where these
communication we have to focus on maximum coverage
nodes are either densely accumulated or not connected
area where network connectivity should be maintained
at all.
within the deployed wireless sensor network. In this paper,
• Self organization In most cases, after deployment,
our review focuses on the suggested energy efficient
sensor networks have no human interference and the
solutions as wireless sensor nodes have only a small and
sensor nodes themselves are responsible for reconfig-
finite source of energy with the most energy consuming
uration in case of any change in their location either
parts—sensors, microcontroller and radio module. Sec-
due to strong wind, thunder storm or floods. This
tion 2 deals with issues related to design requirements of
requires that sensor nodes be familiar with their
energy efficient schemes for data collection and aggrega-
geographic position (either absolute or with respect to
tion, clustering and routing. In Sect. 3, we present review
some coordinate system) and time-synchronized with
of related work for data collection and aggregation for
each other.
structure-free wireless sensor networks. In Sect. 4, we
• Limited resources The wireless sensor nodes are
discuss related work for data collection and aggregation for
equipped with limited energy capacity source which
structured wireless sensor networks along with need and
must be efficiently utilized for processing and
issues associated with clustering and routing through
communication.
review of literature in Sects. 5 and 6. In Sect. 7, future
• Resource dynamics The resources - energy, bandwidth,
directions for research in wireless sensor networks having
and the processing power of wireless sensor network
perspective of these issues is discussed.
change dynamically; hence the network system should
operate autonomously as addition of more sensor nodes
1.1 Design requirements for wireless sensor
or failing of sensor nodes take place.
networks
Several key factors should be taken into consideration
while designing a wireless sensor network and its protocols Data collection and aggregation is regular compilation of
as described in Table 1. Wireless sensor networks were sensed data from multiple sensor nodes to be transmitted to

123
Wireless Netw (2016) 22:335–350 337

Table 1 Key factors [7] in wireless sensor network


Key factors Description

Network Sensor nodes are connected with each other and base station via wireless links to share information. It is preferred as node
connectivity failures may cause the network topology and network size to change and loss of information
Area coverage Region where the sensor nodes are deployed for monitoring and data collection. It needs maximum coverage by sensor
node/base station transmission range
Node deployment Strategy to deterministically place the sensor nodes in order to meet the desired performance goals such as coverage of
monitored region
Fault tolerance Ability of sensor nodes to provide functionality even when few sensor nodes may fail and stop the data transmission due to
power shortage, physical damage or environment interferences and restrain the purpose of the network
Network lifetime Minimum time at which maximum numbers of sensor nodes are dead
Data aggregation Guiding principle to combine the sensed data from numerous sensor nodes with elimination of redundancy and provide
collective information to the base station
Clustering Course of action to group sensor nodes having similar role with an assigned cluster head in a densely deployed large-scale
wireless sensor network
Routing The process of determining a network path for a packet from source node to its destination
Network Changes in the topology of wireless sensor network due to sensor node failure or its mobility. It makes periodic monitoring
dynamics difficult as path stability, bandwidth, energy estimation are required

densely deployed large-scale wireless sensor network. The


energy limitation on sensor nodes result in a limited net-
work lifetime, hence this fact must be considered as
appropriate clustering can reduce the overall energy uti-
lization in a network and improve the network lifetime.
Moreover, when a sensor node drains out energy, it is
not able to receive or transmit packets anymore, thus gets
disconnected from the network and lost its coverage area
also. When data is aggregated, relaying of data from the
sensor nodes to the base station should be done by means
of energy-efficient route detection so that the network
lifetime is maximized. It implies that for increased network
lifetime, wireless sensor network must have following
design consideration:
• Maximum Area Coverage
Fig. 1 Energy consumption by various activities of a wireless sensor • Maximum Connectivity
node [8]
• Minimum Energy Consumption
Table 2 represents the various issues concerning the
the base station for processing ultimately. Since the wire- three major activities performed by wireless sensor
less sensor nodes are distance and energy-constrained; it is networks.
inefficient for all the wireless sensor nodes to transmit the
data directly to the base station. Moreover, if multiple
wireless sensor nodes are sensing same event, the data will 2 Data collection and aggregation
be redundant and massive. Hence, we need such techniques
to combine redundant data into valuable information at the Wireless sensor networks are mainly utilized for data col-
wireless sensor nodes which can reduce the total number of lection and aggregation. Data collection is defined as the
packets transmitted to the base station as estimated organized integration of sensed data from multiple sensors
reduction in energy consumption and bandwidth. Cluster- transmitted ultimately to the base station for processing.
ing algorithms play a crucial role in achieving the targeted However, data generated from neighboring sensor nodes is
design goals for data collection and aggregation with persistently redundant and highly interrelated. In such
maximum area coverage. Clustering is the course of action scenarios, sensor nodes can transmit data to a local col-
to group sensor nodes with an assigned cluster head in a lector or assigned head which combines data from all the

123
338 Wireless Netw (2016) 22:335–350

Table 2 Design requirements for energy efficient schemes in WSN more practical environment where sensor nodes may move
Requirement type Design constraints
or fail unexpectedly, there is too much necessary con-
struction and maintenance overhead associated with
Data collection and 1. Restriction on redundant data structured mechanism that cannot be reimbursed with the
aggregation 2. Low computational overhead advantage from structured data aggregation. On the con-
3. Minimum memory storage trary, structure free approaches do not disburse energy for
Clustering 1. Maximum supported coverage constructing any structure. In this section, we present a
area review of data aggregation protocols in wireless sensor
2. Distribution of sensor nodes networks for both structured as well as structure free sce-
3. Connection density on cluster narios with evaluation of key features and short comings.
head
4. Fresh node addition
Routing 1. Shortest path
3 Data aggregation protocols for structure-free
2. Network connectivity
wireless sensor networks
3. Maximum supported size
4. Low communication overhead In structure-free networks, each sensor node deployed in a
specific region has similar responsibility and holds out
almost same kind of battery. In such networks, data
sensor nodes that reside in its connectivity and transmits aggregation is done by data centric routing where the base
the concise packet to the base station consequentially station regularly broadcast a query message to the sensor
reducing the total number of packet transmission thus, nodes. Such structure-free networks are generally deployed
conserving bandwidth and energy. This can be accom- for crisis management as they are meant for short duration
plished by data aggregation. Data aggregation [9] can be varying from few hours to few days. Structure-free
defined as the guiding principle to combine the sensed data approaches do not pay out energy on establishing any
from numerous sensor nodes with elimination of redun- structure and provide solutions with benefits of reduced
dancy and provide collective information to the base sta- average delay, reduced maintenance overhead and
tion. In the literature, data aggregation protocols could be improved robustness when incurring sensor node failures.
categorized into two groups: structure-free and structured Such structure-free protocols have major role in providing
[10] as shown in Fig. 2. For efficient data aggregation, fault-tolerance to energy-constrained time-critical applica-
structured approaches employ either tree-based or a cluster tions and simultaneously, tend to manage aggregation such
based structure at the network initialization phase. Such as the incident information during a crisis, for e.g. fire
structured protocols are well suited to a stable environment evolution, safe zones discovery, gas diffusion, natural
where sensor nodes function continually. Nevertheless, in a disaster, etc., along with tracking and orientation of rescue

Fig. 2 Taxonomy of data aggregation protocols for energy conservation in wireless sensor network

123
Wireless Netw (2016) 22:335–350 339

persons and intrusion robots. In the rest of this subsection, 3.2 Data-aware anycast and randomized waiting
we describe these protocols and highlight their key fea- (DAA 1 RW) [10]
tures, merits and demerits in Table 3.
Fan et al. proposed a structure-free event-driven reporting
3.1 Sensor protocol for information via negotiation scheme (DAA ? RW) for data aggregation in wireless
(SPIN) [11] sensor networks. The sensor node that detects an event sends
an anycast RTS first to determine the next hop node to report
In this scheme, the sensor nodes that detect an event are event data to the base station. The sensor node which
active participants and initiate the diffusion of data while receives this RTS is a next hop candidate. Sensor nodes that
the base station responds to these participants. In SPIN, have same event data to transmit or are closer to base station
each sensor node imparts two phases: negotiation and are assigned with a priority to respond CTS in order to
resource adaptation. The sensor nodes need a descriptor to achieve better aggregation efficiency. Furthermore, to
concisely describe their sensed data for ideal data negoti- reduce the total number of transmissions, a randomized
ation. Whenever a sensor node (initiator) detects an event, waiting scheme is set up so that each sensor node that has to
it advertises the data to the neighboring nodes in the report event data can start its transmission after a random
wireless sensor network using the metadata. If the neigh- waiting time. Feasible aggregation is prompted when a node
boring node is interested in such kind of data, it sends a nearer to the base station prefers a longer waiting time.
request to the initiator sensor node for data. To keep
account of energy usage by a sensor node, a resource 3.3 Structure-free and energy- balanced data
manager is associated with every sensor node. Before data aggregation (SFEBDA)
transmission, each node polls its resources such as battery
power. This allows sensor nodes to hold back on certain Chih-Min et al. [12] presented a technique SFEBDA for
tasks when its energy is low. Simulation shows SPIN gains wireless sensor networks consisting of two stages. In the first
3.5 times less energy consumption and is able to distribute stage, as an event occurs, some sensor nodes are selected as
60 % more data per unit energy as compared to flooding. primary aggregators (PA) and secondary aggregators (SA)
SPIN is compatible for environments where the forwarding depending on their positions and regarded as a pair (PA/SA
decisions are based on local neighborhood information as pair) for the next step. For each PA/SA pair, the sensor node
in case of mobile sensor nodes. closer to base station is designated as the aggregator and

Table 3 Analysis of structure-free data aggregation protocols


Protocol Selection of Aggregation Merits Demerits
Head

SPIN [11] Based on descriptors Each sensor node requires the knowledge of its Unable to guarantee data delivery
single hop neighbors only
Topological changes are localized
DAA ? RW Aggregation based on Each node with same event data to report or is Aggregation efficiency is poor if the
[10] event detection closer to base station has higher priority and can randomized waiting time selected by the
start its transmission sensor nodes nearer to base station is
Avoids structure maintenance overhead shorter
SFEBDA Primary and secondary Efficient data gathering and balanced energy Computation overhead while selecting
[12] aggregator (PA/SA) pair consumption dynamic aggregator
based on their location Two-phase aggregation process Network holes created
Dynamic aggregator selection
Mechanism
MADA [13] Multi agent- structure free Co-operating agents eliminate the inter-sensor- Aggregator node chosen on basis of event
nodes redundancy and combine processed detection
information of a gathering session into one
message
EEDAA [14] Multi agent- structure free Co-operating agents reduce redundancy Aggregator node selected on basis highest
Aggregation head is selected depending on its residual energy may not have sufficient
residual energy energy for transmission to base station

123
340 Wireless Netw (2016) 22:335–350

other sensor node as the forwarding node. Such a mecha- route request message (RREQ) with a unique RREQ
nism achieves both the forwarding node selection and early identification number. The message will reach the neigh-
aggregation. In the second stage, these aggregators send the boring sensor nodes where each sensor node calculate its
collected packets back to the base station where the number participation relevance and reply with their residual energy
of transmissions for control packets is twice that of data to the requesting sensor node using route reply (RREP)
packets. In SFEB phase-2, the sensor nodes that have to send message along with updating the routing table and setting
packets for reporting include the PA of each aggregation set up a reverse route to the sender sensor node. If the residual
and those sensor nodes that fail to find an aggregator to energy of the requesting sensor node is highest then it is
collect their data. The sensor nodes that fail to send data to elected as the aggregation head otherwise among its
aggregators will also transmit their packets to the base sta- neighbor sensor nodes, the sensor node which has the
tion. The authors assumed the sensor nodes are self-con- highest residual energy is assigned as the aggregation head.
figured, localized and synchronized. Once a sensor node is selected as aggregation head, it
collects the data from all the sensor nodes that want to
3.4 Multi-agent data aggregation (MADA) [13] participate, aggregates it and finally send the data to the
base station.
Sardouk et al. proposed MADA mechanism for structure-
free event-driven wireless sensor networks. The proposed
data aggregation protocol works spontaneously where a 4 Data aggregation protocols for structured
sensor node detects significant information—it asks its wireless sensor networks
adjoining neighbors to participate in order to aggregate the
most representative information to be sent to the base In an unstructured network, aggregation node has excessive
station. Each sensor node in the wireless sensor network communication and computation load hence its battery
decides to participate or not in the data aggregation process depletes faster. The death of such a crucial sensor node
depending on relevance participation factor which is breaks down the functionality of the network [16]. Hence,
determined by various parameters i.e. importance of the in vision of scalability and energy efficiency, several
information, its density, the criticality of its position and its structured data aggregation approaches have been
residual energy. These parameters are weighted and used in proposed.
the decision function. Hence, a group of nodes will be In this section, we survey quite a few data aggregation
formed to aggregate the information and transmit their data protocols which have primarily been designed for struc-
to the initiator node that has detected the event and been tured wireless sensor networks and highlight their merits
designated as aggregation head (AH). The AH transmits and demerits in Table 4.
the aggregated data to the base station in a single hop or
multiple hops depending on the network configuration. At 4.1 Cluster based data aggregation
the end of this aggregation mechanism, these groups dis-
appear and then a new aggregation session is initiated at In energy constrained wireless sensor networks of large
time of new event detection. This data aggregation mech- size, the major traffic during data collection is the reported
anism enhances the energy conservation of the wireless data from each sensor node to the base station. It is
sensor network in terms of reduction of total sensor nodes’ incompetent for sensor nodes to transmit the data directly
participation. to the base station as energy hole problem was reported and
discussed in [17], where sensor nodes closer to the base
3.5 Energy efficient data aggregation (EEDAA) [14] station are depleted quickly in case of multi hop routing
due to traffic relays and create a hole shape area that leaves
Boughami et al. have proposed EEDAA in which the the remaining network disconnected from the base station,
energy conservation is enhanced by the selection of resulting network disconnectivity and reduction in cover-
aggregator node (called aggregation head) and participant age area of the base station. In such circumstances, few
sensor nodes all the way throughout. In this mechanism, sensor nodes can be designated as aggregator or cluster
the sensor node which is aggregating and transmitting the head that collect and aggregate data from all the sensor
concise message to the base station is selected on basis of nodes belonging to its locale which in turn transmit the
its residual energy and proximity to the base station. When concise message to the base station.
a sensor node detects an event, it sends a request message In Fig. 3, a cluster based wireless sensor network
along with its residual energy to all its neighbor sensor organization is shown. The cluster heads can communicate
nodes using ad–hoc on demand distance vector routing with the base station directly via long range transmissions
(AODV) protocol [15] which establishes routes, using or multi hopping through other cluster heads. Many data

123
Wireless Netw (2016) 22:335–350 341

Table 4 Analysis of structured data aggregation protocols


Base Protocol Key features Merits Demerits
structure

Cluster LEACH Randomly selects a few sensor Each node has equal chance to There is a good chance that a node with
based [18] nodes as cluster heads (CHs) become cluster head very low energy gets selected as a CH
Rotates this role to evenly Use of TDMA keeps away CHs from Good CHs distribution cannot be
distribute the energy load among unnecessary collisions guaranteed
the sensors in the network Some nodes will not have any CH in their
range
Chain PEGASIS Nodes are organized into a linear Energy load is distributed uniformly No concern for redundant data
based [20] chain Reduce overhead due to dynamic Requires global knowledge of the location
Greedy approach cluster formation of all nodes at the sink.
Decrease number of data Long chain formation results in delay
transmission
Tree EADAT Node with the higher residual Sensor nodes with higher residual Network density at base station not
based [21] power and shorter path to the power have a higher chance to considered
sink is chosen as its parent become a non-leaf tree node
Average residual energy of all alive Lack of load balancing
sensor nodes decreases slowly If a child node does not receive help
Network lifetime increases linearly message or unable to switch to a new
with the network density parent, it enters into a danger state
Grid Grid [22] A set of data aggregators per grid Adaption to dynamic changes and Nodes in same grid do not communicate
based Data aggregation per Grid event mobility with each other
Data redundancy
In- Data aggregator per grid selected Each sensor node within the grid Overhead in selection of aggregation head
network on basis of signal strength communicates with its neighboring
[22] sensor nodes
Any sensor node within the grid can
attain the role of a data aggregator

aggregation protocols with cluster based organization have distributed protocol in which sensor nodes arrange them-
been proposed in literature and one of them is described in selves into clusters for data aggregation and is well-suited
this section. for applications requiring constant monitoring and periodic
data reporting. Data aggregation done periodically by
Low energy adaptive cluster hierarchy (LEACH)
cluster heads (CHs) reduces the redundancy in information
[18] Heinzelman et al. were the first to propose an energy
that is to be transmitted to the base station. The protocol
efficient cluster formation protocol called LEACH, a
comprises of two phases: setup phase (network is organized
into clusters and cluster heads are chosen) and steady state
phase (data aggregation and transmission to base station by
cluster heads). Nevertheless, LEACH is capable of esca-
lating the network lifetime.

4.2 Chain based data aggregation [19]

In cluster-based wireless sensor networks, if the cluster


head is far away from its member sensor nodes, excessive
energy is expend in communication. This drawback con-
structs the key idea for chain based data aggregation where
each sensor node transmits only to its closest neighbor.
Several protocols have been suggested in literature and one
of them is described as follows:
Power efficient data gathering protocol for sensor infor-
mation systems (PEGASIS) [20] Lindsey et al. presented
Fig. 3 Cluster based wireless sensor network a chain based data aggregation protocol PEGASIS with

123
342 Wireless Netw (2016) 22:335–350

primarily objective to increase the lifetime of each sensor node or undefined state) and the number of hops from the
node by using cooperative techniques and allow local base station. After receiving the control message for the
coordination only between closer sensor nodes so that total first time, a sensor node sets up its timer that counts down
bandwidth utilization get reduced during wireless com- when the channel is idle. During this process, the sensor
munication. Chain formation as shown in Fig. 4 is initiated node chooses other sensor node as its parent having higher
by the farthest sensor node from the base station and at residual power and shorter path to the base station via a
each step, the contiguous neighboring sensor node is cho- control message. The process continues until each node
sen as the successor in the chain on the basis of signal broadcasts once and the result is an aggregation tree rooted
strength to measure the distance to all neighboring sensor at the sink. When the residual power of a sensor falls below
nodes so that only one sensor node can be heard. a threshold, it periodically broadcasts help messages for
The greedy chain formation approach used in [20] may certain time units and shuts down its radio.
result in some sensor nodes having relatively distant
neighbors along the chain. This problem is alleviated by
4.4 Grid based data aggregation
not allowing such sensor nodes to become leaders.
PEGASIS outperforms LEACH by 100 to 200 % [20] in
Vaidhyanathan et al. [22] have proposed two data
terms of the number of data gathering rounds for different
aggregation schemes: grid-based data aggregation and in-
network sizes. Such performance gain is achieved through
network data aggregation as shown in Fig. 6 by parti-
the elimination of the overhead, shorter distance and
tioning the region. In grid-based data aggregation, a set of
decreasing the number of transmissions and reception by
sensor nodes is assigned as data aggregators in fixed
using data aggregation.
regions of the wireless sensor network. The sensor nodes
in a particular grid transmit the data directly to the data
4.3 Tree based data aggregation
aggregator of that grid. Hence, the sensor nodes within a
grid do not communicate with each other. Grid-based data
In a tree based network, sensor nodes are prearranged into a
aggregation is compatible for nomadic environments such
tree where data aggregation is performed at intermediate
as military surveillance and weather forecasting where
nodes along the tree and a concise digest of data is trans-
adaption to dynamic changes and event mobility are
mitted to the root node as shown in Fig. 5.
required. In-network aggregation is analogous to grid
Tree based organization is well-suited for applications
based data aggregation but differs in two conducts, firstly,
that incorporate in-network data aggregation. In this sub-
each sensor node within the grid communicates with its
section, we will describe the organization of data aggre-
neighboring sensor nodes and secondly, any sensor node
gation tree by following protocol:
within the grid can attain the role of a data aggregator.
Energy aware distributed aggregation technique (EADAT) The in-network aggregation scheme is best suited for
[21] Ding et al. have proposed an energy aware dis- environments where events are highly localized. In terms
tributed heuristic (EADAT) to construct and maintain a of the data acquired (throughput), and data latency, the in-
data aggregation tree in wireless sensor networks. The network scheme perform better than grid-based aggrega-
technique is initiated by the base station which broadcasts a tion and no-aggregation scheme. The schemes have also
control message assuming the role of root node in the been compared with respect to the total energy con-
aggregation tree. The control message has five fields: ID, sumption as simulation results indicate that the energy
parent, power, status and hopcount indicating the sensor consumed by the grid based scheme is 2.2 less times than
ID, its parent, its residual power, the status (leaf, non-leaf the no-aggregation scheme.

Fig. 4 Chain based wireless


sensor network

123
Wireless Netw (2016) 22:335–350 343

Fig. 5 Tree based wireless


sensor network

Fig. 6 Grid based wireless


sensor network

Research issues in data collection and aggregation: aggregation process in wireless sensor networks [26].
Further, merits and demerits of the clustering protocols are
• To develop methods for combining data into high
given in Table 5.
quality information at the event detecting sensor node
or intermediate nodes which can reduce the number of
5.1 K-overlapping clustering algorithm (KOCA)
packets transmitted to the base station to conserve
[28]
energy and bandwidth.
Fernandess et al. [27] proposed K-clustering protocol
which can constitute maximum k-hop non-overlapping
5 Current state of art for structured wireless clusters with partial networks topology information instead
sensor network of the whole network topology and simultaneously saves
energy to prolong the network life time. On the other hand,
For structured wireless sensor networks, our review is it doesn’t consider cluster size and may form unbalanced
mainly focused on cluster based networks [23]. Clustering cluster i.e. some clusters may contain enormous number of
protocols facilitates distributed control by the formation of sensor nodes, which result in outsized overhead of inter
hierarchical network topology [24, 25] to figure out and communication. To fix this problem, different improved
set-up the network management regulating the data K-clustering algorithms have been come up with in

123
344 Wireless Netw (2016) 22:335–350

succession. Youssef et al. put forward KOCA algorithm distribution, and considered connectivity density and
[28], an overlapping clustering algorithm appropriate for residual energy of sensor nodes to form clusters. In cycle
uniform distribution of sensor nodes, elaborating the phase, the process of reelecting cluster head is triggered.
overlapping k-hop clustering problem for wireless sensor The cluster head gathers the weight of all member nodes,
networks and presented a randomized distributed heuristic and then selects the node with highest weight as the next
algorithm for solving the problem. head node. In this way, the communication costs are
decreased.
5.2 Distributed energy—balanced unequal
clustering (DEBUC) [29] 5.4 Sink mobility based and energy balancing
unequal clustering protocol (SEBUCP) [31]
DEBUC protocol, proposed by Jiang et al. combines an
unequal clustering mechanism with an inter-cluster multi- In order to balance the energy consumption of a sensor
hop routing. DEBUC partitions all nodes into unequal node moderately, SEBUCP chooses those sensor nodes as
clusters, by deploying time based competitive clustering cluster heads having higher residual energy and powerful
method, in which the cluster heads can preserve more communication capabilities and by adopting the improved
energy for the inter-cluster relay communications to avoid SFLA (shuffled frog leaping algorithm), divides all sensor
the ‘hot-spots’ problem. DEBUC takes up an energy-aware nodes into clusters of different size. In order to reduce the
multi-hop routing system for inter-cluster traffic, to reduce cluster head replacement regularity, cluster head serves
and balance the energy overhead of the cluster heads. continuously and by comparing nodes weight to determine
However, unequal clusters may result in loss of informa- the cluster head exchange time. The greedy algorithm is
tion and full area coverage problem arises. introduced to select an optimum relay sensor node between
cluster head and base station. Simulation results demon-
5.3 Distributed self-organization balanced strate that SEBUCP has good performance on the network
clustering algorithm (DSBCA) [30] lifetime and energy balance.

DSBCA follows a distributed approach to establish hier- 5.5 Transmission-efficient clustering method
archical structure in self-organizing mode without central (TECM) [32]
control. The authors have divided the protocol into three
stages: cluster head (CH) selecting phase, clusters building The hybrid method of using compressive sensing (CS) was
phase and cycle phase. In CH selection phase, DSBCA proposed to reduce the number of transmissions in sensor
selects random sensor node (ut) to trigger clustering pro- networks. The authors proposed a clustering method that
cess. The sensor node with the highest weight in k-hop employs hybrid CS for sensor networks. The sensor nodes
neighbors of ut is elected as CH. Then, the trigger node ut are organized into clusters where in each cluster, nodes
calculates its connected density and distance from the base transmit data to cluster head (CH) without using CS and
station to determine cluster radius k and becomes the later CHs use CS to transmit data to the base station.
temporary CH. In cluster building phase, the authors cal- Significant reduction in number of transmissions is
culated the radius of cluster based on distance and observed from the simulation results where centralized

Table 5 Merits and demerits of clustering protocols


Protocol Merits Demerits

KOCA Constitute maximum k-hop non-overlapping clusters with partial Doesn’t consider cluster size and may form unbalanced
[28] networks topology information and saves energy cluster, no consideration for residual energy of sensor
nodes
DEBUC An energy-aware multi-hop routing is adopted for inter-cluster traffic, Unequal clusters may result in loss of information and full
[29] to reduce and balance the energy overhead of the cluster heads coverage problem
DSBCA Communication costs are decreased as cluster head gathers the weight Random selection of trigger node for cluster head
[30] of all member nodes, and then selects the node with highest weight as selection, Energy holes may be created
the next head node
SEBUCP Cluster heads are the sensor nodes having higher residual energy and Non uniform energy drainage across different sensors
[31] powerful communication capabilities
TECM Optimal size of clusters is based on number of transmission and Coverage problem as some nodes may not join any cluster
[32] compressive sensing is deployed to reduce redundancy

123
Wireless Netw (2016) 22:335–350 345

clustering is done on basis of relationship between the size protocols find a route on demand by flooding the network
of clusters and number of transmissions in the hybrid CS with route request packets. However, it takes high latency
method intended for the optimal size of clusters. time in route discovery and excessive flooding can lead to
network congestion (AODV) [34]. Hybrid protocols are
Research challenges identified in clustering: blend of proactive and reactive protocols where path is
• Grouping adjacent sensor nodes and how many groups established with proactive approach and then provides the
should be there that could optimize some performance demand from additionally activated sensor nodes through
parameter. reactive flooding. The efficiency of hybrid protocol
• Determining optimal value of distance d that minimizes depends on number of other sensor nodes being activated.
overall energy consumption Nevertheless, reaction to traffic demand depends on
• Amount of data to be transferred within each cluster amount of traffic volume (ZRP) [35].
and between clusters
• Selection of cluster head (CH) of a cluster. 6.2 Network based routing protocols

Network based routing protocols depend on the strategy


6 Routing in structure-free and structured how the network is prearranged. Such protocols fall under
wireless sensor network three categories [36]: flat, hierarchical and location based.
Flat based routing is used when enormous sensor nodes are
Energy efficiency is a key issue while designing the net- required to impart same responsibility. The protocol fol-
work routing as sensor nodes meet head-on constraints lows data-centric routing approach in which base station
such as limited energy supply, low memory and reduced sends query to a group of specific sensor nodes in a region
processing power in structure-free as well as structured and waits for response for e.g. direct diffusion [37]. In
wireless sensor networks Table 6. hierarchical-based routing, high energy sensor nodes are
High sensor node density in wireless sensor networks arbitrarily selected as cluster heads for processing and
excludes them from being completely cut off from each transmitting data while low energy nodes are used for
other hence sensor nodes are anticipated to be highly sensing and sending sensed data to the cluster heads. For
connected. This feature incorporates multi-hop in wireless e.g. HPAR [38], TEEN [39]. Location-based routing pro-
sensor networks where each sensor node imparts a twin tocols perform routing based on the location of sensor
role as data sender as well as data router. The malfunc- nodes by using their position information instead of their
tioning of some sensor nodes due to energy drain out can links information for routing. They are located mostly by
cause topological changes, and may need rerouting of means of GPS. The distance between sensor nodes is
packets and reconfiguration of the wireless sensor network estimated by the signal strength received from these nodes
[33]. This may result in unstable network topology and and their coordinates are calculated by exchanging infor-
shrinkage in network coverage area. To restrain energy mation between neighboring sensor nodes. For e.g. GAF
consumption, protocols proposed in the literature for [40], GPSR [41]. The merits and demerits of network based
wireless sensor network exploit some well-defined routing routing protocols are listed in Table 7.
strategy. More or less all of the routing protocols can be
classified [46] according to the path based as proactive, 6.3 Operation based routing protocols
reactive or hybrid, network structure based as flat, hierar-
chical, or location-based, protocol operation based as As per the requirements, the applications of WSNs are
multipath, query, negotiation, QoS or coherent/non-coher- designed with different functionalities. Hence routing pro-
ent and next hop selection based as broadcast, location, tocols are classified according to their type of action taken in
probabilistic, content or hierarchical as in Fig. 7. order to meet these functionalities as: multipath, query-
based, negotiation-based, QoS-based, coherent and non-
6.1 Path based routing protocols coherent routing protocols. In case of multipath routing
protocols, multiple path selection is followed by a message to
Path establishment can be done while routing in three reach destination so that delay is reduced to improve the
ways, namely proactive, reactive or hybrid. Proactive network performance. However, consumption of energy is
protocols uphold fresh lists of destinations and their routes more due to augmented communication as network paths are
by periodically distributing routing tables throughout the to be kept alive by sending periodic messages For e.g.
wireless sensor network. However, they necessitate rele- MMSPEED [42]. Query based routing protocols works on
vant amount of data for maintenance and have slow reac- commencement of queries for data. The initiator node set off
tion on reorganization and failures (DSDV) [33]. Reactive query of interest from a node through network and node with

123
346 Wireless Netw (2016) 22:335–350

Table 6 Analysis of clustering protocols on basis of properties


Protocol Cluster Intra-cluster Inter-cluster Stability Cluster Clustering goal Algorithm
count topology connectivity balancing complexity

KOCA Fixed Fixed (k-Hop) Direct link/multi- Presumed Low Overlapping and Constant O(1)
[28] hop connectivity
DEBUC Fixed Adaptive (multi- Direct link/multi- Prerequisite Moderate Scalability and fault Constant O(1)
[29] level) hop tolerance
DSBCA Variable Adaptive (multi- Direct Link/multi- Prerequisite Moderate Load balancing and Constant O(1)
[30] level) hop bandwidth gain
SEBUCP Variable Fixed (k-Hop) Direct link Presumed Moderate Save energy and coverage Constant O(1)
[31]
TECM Variable Fixed (k-Hop) Direct link/Multi- Prerequisite High Save energy and load Constant O(1)
[32] hop balancing

Fig. 7 Classification of energy-


aware routing protocols in
wireless sensor networks

Table 7 Merits and demerits of network based routing protocols


Protocol Category Merits Demerits

DD [38] Flat Resolve the numerous mobile sinks in large scale Large latency as matching data to queries might require
WSNs overhead
Suitable to event detecting WSNs among irregular Low energy efficiency
data traffic
TEEN Hierarchical Data transmission can be controlled by varying two If thresholds are not meet, the sensor node will not
[39] thresholds communicate
Well suited for time critical applications Data may be lost if CHs are not able to communicate with
each other
GAF Location Increase the network lifetime by saving energy Large traffic injection and delay is not predictable
[40] based Routing fidelity is maintained

123
Wireless Netw (2016) 22:335–350 347

Table 8 Merits and demerits of operation based routing protocols


Protocol Category Merits Demerits

PRTR [53] Multipath Minimizes delay for real-time traffic and alleviate possible Not suitable for time varying topologies
congestions
COUGAR Query based Uses queries to abstract query processing Query may add extra overhead
[43] Utilizes in-network data aggregation to obtain more energy Synchronization among nodes is required
savings
SPIN [11] Negotiation Negotiates before transmitting data to avoid redundancy Data advertise mechanism do not guarantee
based Topological changes are localized since each node knows delivery of data
only single-hop neighbors Need to maintain global network topology
SAR [44] QoS based Routing metric: energy resources, QoS on each path, and the Suffers from the overhead of maintaining the
priority level of each packet routing tables and states at each sensor node
Ensures fault tolerance and easy recovery
SWE, Data- Election of a sensor node is based on energy reserves and Has longer delay, higher overhead, and lower
MWE processing computational capability scalability
[45] based A minimum-hop spanning tree will completely cover the
network that yields the minimum energy consumption

this interest matches the query and drives back to the node 6.4 Next-hop selection based routing protocols
which initiated the query. For e.g. COUGAR [43]. Negoti-
ation based routing protocols employ negotiation to elimi- Next-hop selection based routing protocols can be cate-
nate redundant data transmission by exploiting high level gorized under following: broadcast based, location based,
data descriptors that make intelligent decisions for commu- hierarchy based, content-based and probability based. In
nication and other actions on basis of some fact like avail- broadcast-based routing protocols, each node in the net-
ability of resources. For e.g. SPIN [11]. work decides individually whether to forward a message or
In QoS based routing protocols, various parameters have not. If a node decides to forward, it simply re-broadcasts
to be considered during routing to provide a sense of bal- the message. If it declines to forward, the message will be
ance between energy consumption and data quality. The dropped. For e.g. Minimal cost forwarding algorithm
parameters like delay, reliability, jitter and bandwidth are (MCFA) [47]. Location-based routing protocols make a
determined as QoS metric in the wireless sensor network decision on the next-hop en routed for the destination on
when the sensor nodes communicate to the base station in the basis of known location of its neighbors and the des-
addition with energy consumption as another metric. tination designated as centroid of a region. Such protocols
Routing decision depends on three factors: available can avoid the communication overhead imposed by
energy resources, QoS on each path, and the priority level flooding, but the calculation of the positions of neighbors
of each packet. To avoid single route failure, a multipath may incur extra overhead. For e.g. Geographical and
approach and localized path restoration schemes are used. energy aware routing (GEAR) [48]. In case of hierarchical-
Examples of such routing are: SPEED [44], Sequential based routing protocols, all nodes forward a message to a
assignment routing (SAR) [44]. Coherent and non-coherent sensor node (also called aggregator). Each sensor node
processing [45] involves data processing as a major con- aggregates the incoming data in order to reduce commu-
stituent in the operation of wireless sensor networks at the nication overload and conserve more energy thereby
time of routing that adopt different data processing tech- resulting in increase in network lifetime. This routing is
niques. In non-coherent data processing routing, sensor primarily bi-layer routing where one layer is assigned for
nodes in the surrounding area process the raw data before selection of cluster heads and the other layer is dedicated
being sent to other sensor nodes for further processing. The for routing. For e.g. TSC [50]. Content-based routing
sensor nodes that carry out further processing are called the protocols determine the next-hop on the route purely based
aggregators. In case of coherent data-processing routing, on the query content and are compatible to the architecture
the data is forwarded to aggregators after minimum pro- of wireless sensor networks as the base station requests
cessing that typically includes the tasks like time-stamping only for data regardless of its origin. For e.g. Gradient
and redundancy elimination. For e.g. Single winner algo- based routing (GBR) [50]. Probabilistic routing protocols
rithm (SWE) and Multiple winner algorithm (MWE) [45]. assume that all sensor nodes are homogeneous and ran-
Further, merits and demerits of operation based routing domly deployed. Using this routing protocol, sensor nodes
protocols are summarized in Table 8. arbitrarily select the next-hop neighbor for each message to

123
348 Wireless Netw (2016) 22:335–350

Table 9 Merits and demerits of next-hop selection based routing protocols


Protocol Category Merits Demerits

MCFA Broadcast Each node know the least cost path estimate from Long delays cause a node to become bottleneck due to
[46] based itself to the base station rebroadcasting
GEAR Location Each node keeps an estimated cost and a learning A hole occurs when a node does not have any closer neighbor
[47] based cost of reaching the destination Requires a location service to map locations and node
identifiers
GBR Content Obtain balanced distribution of the traffic in the Extra overhead in terms of energy consumption and memory
[48] based network, thus increasing the network lifetime storage
Memorize the number of hops when interest is
diffused through the whole network
EAR Probability Maintains a set of paths instead of maintaining one Needs location information and setting up the address
[49] based optimal path at higher rates mechanism for the nodes, which complicate route setup
TSC Hierarchical Constructs more balanced network in term of energy Lot of overhead due to global information for communication
[50] based consumption and communication load Single hop communication consume lot of energy
Uses dynamic sizing of clusters
ALBA- Content Relay selection and load balancing Requires a location service to map locations and node
R [51] based Detect and route around connectivity holes identifiers
EESOR Opportunistic Reduces the size of forwarder list on basis of Long delays cause a node to become bottleneck
[52] proximity to destination

be forwarded. A set of paths are maintained and selected by summarized and compared different proposed design and
means of a specific probability whose value depends on protocols. Specific consideration has been given to a sys-
how low the energy consumption for each path can be tematic and comprehensive classification of the protocols
achieved. For e.g. Energy aware routing (EAR) [51]. that are being proposed in the literature for energy con-
Subsequently, the next-hop selection based protocols are servation. We observed that the field of energy conserva-
summarized with their merits and demerits in Table 9. We tion targeted for data collection and aggregation has to be
observed that there are some hybrid protocols that fall explored with collaboration of clustering and communica-
under more than one category. tion. There is a big room available for developing energy-
efficient techniques with different approaches (aggregation,
Research challenges identified in routing: clustering and routing) into a single off-the-shelf feasible
• To develop energy-efficient data routing protocol from solution for data collection. Clustering protocols presented
sensor nodes to cluster head and from cluster head to in this paper have focus on minimization of energy con-
base station via normal/traffic nodes handling frequent sumption associated with the cluster head selection pro-
change in network topology and node hardware cess, distribution of cluster heads, re-selection of cluster
condition. heads and re-clustering. Further enhancements should be
• Exploit spatial diversity to provide spatial communi- able to get accustomed to clustering mechanism to main-
cation between sensor nodes and density of sensor/ tain wireless sensor network connectivity by reducing
actuator nodes should allow the network to adapt to an energy consumption activities associated with re-election
unpredictable environment. of CHs and re-clustering. The comprehensive review of
• To provide localization, self-configuration and recon- routing protocols convinced us that a routing protocol is
figuration in wireless sensor nodes being unattended in said to be energy-efficient if it fulfills one or more aspects
a dynamic and energy constrained environment to of the following: minimizing total energy spent in the
enhance the network lifetime. network, minimizing the number of data transmissions,
maximizing the number of alive nodes over time or bal-
ancing the energy dissipation among the sensor nodes in
7 Conclusion the WSN. We are persuaded that this class of approach will
get an even greater importance and attention within the
In this paper we had surveyed and examined current state research community in the coming years and can reduce
of the main approaches for energy conservation in WSNs the gap between technology and application by reducing
with issues on three major activities (1) data collection and energy consumption over time and extending network
aggregation, (2) clustering and (3) routing. We have lifetime.

123
Wireless Netw (2016) 22:335–350 349

References 21. Ding, M., Cheng, X., & Xue, G. (2003). Aggregation tree con-
struction in sensor networks. 2003 IEEE 58th vehicular tech-
1. News of bridge collapse at Minnesota, USA. http://www.nytimes. nology conference (Vol. 4, no. 4, pp. 2168–2172).
com/2007/08/02/us/02bridge.html. 22. Vaidhyanathan, K., Sur, S., Narravula, S., & Sinha, P. (2004).
2. News of fire break at forests of California. http://www.huffington Data aggregation techniques sensor networks. Technical report,
post.com/2013/06/30/arizona-wildfire-yarnell_n_3526934.html. OSU-CISRC-11/04-TR60, Ohio State University.
3. Akyildiz, I. F., Su, W., Sankarasubramaniam, Y., & Cayirci, E. 23. Comeau, F., Sivakumar, S. C., Robertson,W., & Phillips,W. J.
(2002). A survey on sensor networks. IEEE Communications (2006) Energy conserving architectures and algorithms for
Magazine, 40(8), 102–114. wireless sensor networks. 39th hawaii international conference on
4. Tilak, S., Abhu-Gazhaleh, N., & Heinzelman, W. R. (2002). A system sciences (HICSS-39 2006).
taxanomy of wireless micro-sensor network models. ACM SIG- 24. Tilak, S., Abu-Ghazaleh, N. B., & Heinzelman, W. (2002). A
MOBILE Mobile Computing Communication Review, 6(2), 28–36. taxonomy of wireless micro-sensor network models. Mobile
5. Shen, C. C., Srisathapornphat, C., & Jaikaeo, C. (2001). Sensor Computing Communications Review, 6(2), 28–36.
information networking architecture and applications. IEEE 25. Lee, J. S., & Cheng, W. L. (2012). Fuzzy-logic-based clustering
Personal communications, 8(4), 52–59. approach for wireless sensor networks using energy predication.
6. Chong, C.-Y., & Kumar, S. P. (2003). Sensor networks: Evolu- IEEE Sensors Journal, 11(9), 2891–2897.
tion, opportunities, and challenges. Proceedings of the IEEE, 26. Attea, B. A., & Khalil, E. A. (2011). A new evolutionary based
91(8), 1247–1256. routing protocol for clustered heterogeneous wireless sensor
7. Yick, J., Mukherjee, B., & Ghosal, D. (2008). Wireless sensor networks. Applied Soft Computing, 12(7), 1950–1957.
network survey. Elsevier Journal on Computer Networks, 52, 27. Fernandess, Y., & Malkhi, D. (2002). K-clustering in wireless ad-
2292–2330. hoc networks. In Proceedings 2nd ACM workshop principles
8. Halgamuge, M. N., Zukerman, M., Vu, H. L., & Ramamoha- mobile computing conference (pp. 31–37).
narao, K. (2009). An estimation of sensor energy consumption. 28. Youssef, M., Youssef, A., & Younis, M. (2009). Overlapping
Progress In Electromagnetics Research B, 12, 259–295. multi-hop clustering for wireless sensor networks. IEEE Trans-
9. Ramesh, R., & Varshney, P. K. (2006). Data aggregation tech- actions on Parallel and Distributed Systems, 20(12), 1844–1856.
niques in sensor networks: A survey. Electrical Engineering and 29. Jiang, C. J., Shi, W. R., Tang, X. L., Wang, P., & Xiang, M.
Computer Science. Paper 22. http://surface.syr.edu/eecs/22. (2012). Energy balanced unequal clustering routing protocol for
10. Fan, K. W., Liu, S., & Sinha, P. (2007). Structure-free data wireless sensor networks. Journal of Software, 23(5), 1222–1232.
aggregation in sensor networks. IEEE Transactions on Mobile 30. Liao, Ying, Qi, Huan, & Li, Weiqun. (2013). Load-balanced
Computing, 6(8), 929–942. clustering algorithm with distributed self-organization for wire-
11. Kulik, J., Heinzelman, W. R., & Balakrishnan, H. (2002). less sensor networks. IEEE Sensors Journal, 13(5), 1498–1506.
Negotiation-based protocols for disseminating information in 31. Chen, K. (2013). Unequal cluster-based routing protocol in wire-
wireless sensor networks. Wireless Networks, 8, 169–185. less sensor networks. Journal of Networks, 8(11), 2656–2662.
12. Chao, C. M., Hsiao, T. Y. (2009). Design of structure-free and 32. Xie, Ruitao, & Jia, Xiaohua. (2014). Transmission-efficient
energy-balanced data aggregation in wireless sensor networks. In clustering method for wireless sensor networks using compres-
11th IEEE international conference on high performance com- sive sensing. IEEE Transactions onParallel and Distributed
puting and communications, HPCC 2009 (pp. 222–229). Systems, 25(3), 806–815.
13. Sardouk, A., Mansouri, M., Merghem-Boulahia, L., Gaiti, D., 33. Boukerche, A. (2009). Algorithms and protocols for wireless
Rahim-Amoud, R. (2010) Multi-agent system based wireless sensor networks. Mississauga: Wiley.
sensor network for crisis management. In 2010 IEEE global 34. Dargie, W., & Poellabauer, C. (2010). Wireless sensor networks
telecommunications conference (GLOBECOM 2010) (pp. 1–6). theory and practice (1st ed.). Wiley: Hoboken.
14. Boughanmi, Najet, Esseghir, Moez, Merghem-Boulahia, L., & 35. Rahman, J., Hasan, M. A. M., Islam, M. K. B., (2012) Compar-
Khoukhi, L. (2013). Energy efficient aggregation in wireless ative analysis the performance of AODV, DSDV and DSR
sensor networks, ICD/ERA, UMR 6279, Troyes University of routing protocols in wireless sensor network. Electrical and
Technology, 12 rue Marie Curie, 10000 Troyes. Berlin: Springer. computer engineering (ICECE), 2012 7th international confer-
15. Perkins, C., Belding-Royer, E., & Das, S. (2003). Ad hoc on- ence on (pp. 283, 286).
demand distance vector (AODV) routing IETF RFC 3561. 36. Royer, E. M., & Toh, C. K. (1999). A review of current routing
16. Xue, Y, Cui, Y., & Nahrstedt, K. (2005). Maximizing lifetime for data protocols for ad hoc mobile wireless networks. IEEE Personal
aggregation in wireless sensor networks. ACM/Kluwer mobile net- Communications, 6(2), 46–55.
works and applications (MONET) special issue on energy constraints 37. Intanagonwiwat, C., Govindan, R., & Estrin, D. (2000, August).
and lifetime performance in wireless sensor networks (pp. 853–864). Directed diffusion: A scalable and robust communication para-
17. Tan, H. O., & Korpeoglu, I. (2003). Power efficient data gath- digm for sensor networks. In Proceedings of the 6th annual
ering and aggregation in wireless sensor networks. Sigmod international conference on ACM Mobile computing 2000 (Bos-
Record, 32(4), 66–71. ton, MA, pp. 56–67).
18. Heinzelman, W. B., Chandrakasan, A. P., & Balakrishnan, H. 38. Li, Q., Aslam, J., & Rus, D. (2001). Hierarchical power-aware
(2002). An application-specific protocol architecture for wireless routing in sensor networks. In Proceedings of the DIMACS
microsensor networks. IEEE Transactions on Wireless Commu- workshop on pervasive networking.
nications, 1(4), 660–670. 39. Manjeshwar, A., & Agarwal, D. P. (2001). TEEN: A routing
19. Du, K., Wu, J.,& Zhou, D. (2003). Chain-based protocols for data protocol for enhanced efficiency in wireless sensor networks. 1st
broadcasting and gathering in sensor networks. International international workshop on parallel and distributions of comput-
parallel and distributed processing symposium. ing issues in wireless networks and mobile computing.
20. Lindsey, S., Raghavendra, C., & Sivalingam, K. M. (2002). Data 40. Xu, Y., Heidemann, J., & Estrin, D. (2001). Geography-informed
gathering algorithms in sensor networks using energy metrics. energy conservation for ad hoc routing. In Proceedings of the 7th
IEEE Transactions on Parallel and Distributed Systems, 13(9), annual international conference on Mobile computing and net-
924–935. working (pp. 70–84).

123
350 Wireless Netw (2016) 22:335–350

41. Karp, B., & Kung, H. T. (2000). GPSR: Greedy perimeter Recent trends in computer networks and distributed systems
stateless routing for wireless sensor networks. In Proceedings of security communications in computer and information science
mobile computing (Boston, MA). (vol. 420, pp. 16–31).
42. Felemban, E., Lee, C. G., & Ekici, E. (2006). MMSPEED:
Multipath Multi-SPEED protocol for QoS guarantee of reliability
and, timeliness in wireless sensor networks. IEEE Transactions Sarika Yadav is presently pur-
onMobile Computing, 5(6), 738–754. suing a Ph.D. degree in Com-
43. Yao, Y., & Gehrke, J. (2002). The cougar approach to in-network puter Science and Engineering
query processing in sensor networks. Sept: SIGMOD Record. from Motilal Nehru National
44. He, T., Stankovic, J., Lu, C., & Abdelzaher, T. (2003). SPEED: A Institute of Technology
stateless protocol for real-time communication in sensor net- (MNNIT), Allahabad, India.
works. In Proceedings of the23rd international conference on She received her M. Tech.
distributed computing systems, providence, RI. degree in Computer Science and
45. Sohrabi, K., & Pottie, J. (2000). Protocols for self-organization of Engineering from Uttar Pradesh
a wireless sensor network. IEEE Personal Communications, 7(5), Technical University, Lucknow,
16–27. India. Her research interests are
46. Kumar, D. (2014). Performance analysis of energy efficient in the field of wireless sensor
clustering protocols for maximising lifetime of wireless sensor networks, network survivability,
networks. Wireless Sensor Systems, IET, 4(1), 9–16. fault tolerance systems, and
47. Ye, F., Chen, A., Lu, S., & Zhang, L. (2001). A scalable solution artificial intelligence.
to minimum cost forwarding in large sensor networks. In Pro-
ceedings of the 10th international conference on computer
communications and networks (pp. 304–309). Rama Shankar Yadav is cur-
48. Yu, Y., Estrin, D., & Govindan, R. (2001). Geographical and rently a professor at Motilal
energy-aware routing: A recursive data dissemination protocol Nehru National Institute of
for wireless sensor networks, UCLA Computer Science Depart- Technology, Allahabad, India.
ment technical report, UCLA-CSD TR-010023. He received his Ph.D. degree
49. Gautam, N., Lee, W. I., & Pyun, J. Y. (2009). Track-sector from the Indian Institute of
clustering for energy efficient routing in wireless sensor net- Technology (IIT), M.S. degree
works. In Proceedings of the 9th IEEE international conference from Birla Institute of Technol-
on computer and information technology (pp. 116–121). ogy and Science (BITS) Pilani,
50. Schurgers, C., & Srivastava, M. B. (2001). Energy efficient and B. Tech. degree from the
routing in wireless sensor networks. In Military communications Institute of Engineering and
conference, 2001. MILCOM 2001. Communications for network- Technology (I.E.T.), Lucknow,
centric operations: Creating the information. Force, McLean, India. Dr. Yadav has extensive
VA. research and academic experi-
51. Shah, R. C., & Rabaey, J. (2002). Energy aware routing for low ence. He has worked in leading
energy ad hoc sensor networks. IEEE WCNC, Orlando, FL (pp. institutions such as Govind Ballabh Pant Engineering College
17–21). (GBPEC), Pauri, Garhwal, and Birla Technical Training Institute
52. Petrioli, Chiara, Nati, Michele, Casari, Paolo, Zorzi, Michele, & (BTTI), Pilani. He has authored more than 50 research papers in
Basagni, Stefano. (2014). ALBA-R: Load-balancing geographic national/international conferences, refereed journals, and book chap-
routing around connectivity holes in wireless sensor networks. ters. Dr. Yadav’s areas of interest are real time systems, embedded
IEEE Transactions on Parallel and Distributed Systems, 25(3), systems, fault-tolerant systems, energy aware scheduling, network
529–539. survivability, computer architecture, distributed computing, and
53. Yamuna Devi, C. R., Shivaraj, B., Manjula, S. H., Venugopal, K. cryptography.
R., Patnaik, L. M. (2014). EESOR: Energy efficient selective
opportunistic routing in wireless sensor networks. Springer:

123

You might also like