Review Article
A Comprehensive Survey on Hierarchical-Based Routing
Protocols for Mobile Wireless Sensor Networks: Review,
Taxonomy, and Future Directions
Nabil Sabor,
1,2
Shigenobu Sasaki,
1
Mohammed Abo-Zahhad,
2
and Sabah M. Ahmed
2
1
Department of Electrical and Electronic Engineering, Niigata Un iversity, N iigata 950-2181, Japan
2
Electrical and Electronics Engineering Departmen t, Faculty of Engineering, Assiut University , Assiut 71516, Egypt
Correspondence should be addressed to Nabil Sabor; nabil
sabor@aun.edu.eg
Received June ; Accepted August ; Published January
Academic Editor: Pierre-Martin Tardif
Copyright © Nabil Sabor et al. is is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introducing mobility to Wireless Sensor Networks (WSNs) puts new challenges particularly in designing of routing protocols.
Mobility can be applied to the sensor nodes and/or the sink node in the network. Many routing protocols have been developed
to support the mobility of WSNs. ese protocols are divided depending on the routing structure into hierarchical-based, at-
based, and location-based routing protocols. However, the hierarchical-based routing protocols outperform the other routing
types in saving energy, scalability, and extending lifetime of Mobile WSNs (MWSNs). Selecting an appropr i ate hierarchical routing
protocol for specic applications is an important and dicult task. erefore, this paper focuses on reviewing some of the recently
hierarchical-based routing protocols that are developed in the last ve years for MWSNs. is survey divides the hierarchical-
based routing protocols into two broad groups, namely, classical-based and optimized-based routing protocols. Also, we present
a detailed classication of the reviewed protocols according to the routing approach, control manner, mobile element, mobility
pattern, network architecture, clustering attributes, protocol operation, path establishment, communication paradigm, energy
model, protocol objectives, and applications. Moreover, a comparison between the reviewed protocols is investigated in this survey
depending on delay, network size, energy-eciency, and scalability while mentioning the advantages and drawbacks of each
protocol. Finally, we summarize and conclude the paper with future directions.
1. Introduction
e importance of Mobile W ireless Sensor N etworks
(MWSNs) in monitoring and collecting data from the
environment has been increased in the recent years. A
MWSN typically consists of mobile sensor nodes and/or
mobile sink(s) that can move within the network. Mobility
canbeachievedbyattendingthemobileelementswith
mobilizers for controlling their locations or they can be
attached to transporters like vehicles, animals, r obots,
and so forth [, ]. Some real-world applications demand
environments composed of static and mobile sensor nodes
within t he same network, while other applications request
complete mobile sensors environment [–]. For instance,
in outdoor medical applications, patients use body sensor
networks (mobile nodes) to monitor their vital signs (e.g.,
heart rate and blood pressure) which are transmitted to
doctor (mobile sink) and monitoring room (static sink) for
tracking and monitoring patients’ health status, while in wild
life applications, sensors are equipped on animals (mobile
nodes) to observe their behavior within a spacious habitat
which are collected via a mobile sink (e.g., a car or a plane)
[]. e sensor nodes are small devices and can perform
many tasks, including event sensing, data processing, and
data communication. Each no de consists of four main parts
as shown in Figure . e rst part is a sensing unit that
senses a phenomenon and conv erts sensing data into a
digital form using a sensor and Analog-to-Digital Converter
(ADC). Processor unit is the second part that processes all
data and controls operations of the other parties. e third
part is a transceiver unit that is used to transmit and receive
data with a limited transmission range. Power unit is the last
Hindawi
Wireless Communications and Mobile Computing
Volume 2017, Article ID 2818542, 23 pages
https://doi.org/10.1155/2017/2818542
W ireless Communications and Mobile Computing
Additional units
GPS
Mobilizer
Power generator
Processing unit
Processor
Memory
Sensing unit
ADC
Sensor
Transceiver unit
Power unit
F : Structure of sensor node.
part that supplies power to all parties. Moreover, a sensor
node might additionally have some specic components,
such as the Global Positioning System (GPS), mobilizer, and
power generator units. e GPS unit can help a sensor node
to obtain its own location information while the mobilizer
unit oers the movement capability of a sensor node. e
power generator unit is responsible for power generation by
applying some specic techniques such as solar cell [].
Asensornodehaslimitedon-boardstorage,process-
ing, power, and radio capabilities due to its small size.
erefore, MWSNs require eective mechanisms to uti-
lizeandresolvethelimitedresources.Routingisoneof
these mechanisms that prolongs the lifetime of network
by reducing the energy consumption in communication.
e routing protocol should consider the network structure,
data sending methods, node and link heterogeneity, node
mobility, consumption energy, coverage, connectivity, data
aggregation, and quality of service issues in order to be
an ecient and reliable protocol. e developed routing
protocols can be grouped based on the routing structure into
hierarchical-based, at-based, and location-based routing
protocols. According to literature [–], the hierarchical-
based routing protocols outperform the other routing types in
saving energy, extending a lifetime of WSNs, and scalability.
e hierarchical-based routing partitions the network into
multiplegroups.Eachgroupcontainsoneheadnodeand
many member nodes (MNs). MN only senses and delivers
itssenseddatatoitsrelatedheadnode,whileaheadis
responsible for collecting and aggregating data of its MN s and
then transfers the aggregated data to sink.
Hierarchical-based routing is a feasible solution for
reducing energy consumption in WSNs due to reduction of
redundant data transmission. In addition, it can eectively
balance the load among the sensor nodes via assigning
dierent tasks for each sensor node according to its capa-
bilities. Finally, it can easily achieve collision-free network
byusingaproperMACprotocol.Basedontheapproachof
data routing, t he hierarchical-based routing can be broadly
divided into classical hierarchical-based routing and opti-
mized hierarchical-based routing. is survey will specialize
in reviewing some of the recently classical and optimized
hierarchical-based routing protocols that are developed in the
last ve years for the MWSNs. Also, a detailed classication
of the reviewed protocols is presented in this survey based on
many dierent metrics. ese metrics are routing approach,
control manner, mobile element, mobility pattern, network
architecture, clustering attributes, protocol operation, path
establishment, communication paradigm, energy model,
protocol objectives, and applications. Moreover, we compare
between the surveyed protocols on the basis of delay, network
size, energy-eciency, and scalability while analyzing the
advantages and drawbacks of each protocol.
e paper is organized as follows. Section is a literature
survey about various clustering protocols of MWSNs. Taxon-
omy criteria of the hierarchical-based routing protocols are
explained in Section . Section discusses the metrics that
are used for evaluation of the hierarchical-based routing pro-
tocols. Reviewing t he hierarchical-based routing protocols of
MWSNs is given in Section . In S ection , the classication
and comparison of the reviewed protocols are described.
Section oers some conclusions and future directions.
2. Related Works
is section summarizes t he previous surveys [–] of the
routing protocols that are designed for MWSNs in literature.
A survey of routing techniques that are developed for wireless
ad-hoc or WSN networks was presented in []. e aim of this
survey is to identify the routing protocol that can support the
mobility of the sensor nodes in mixed WSNs, which consist
of both static and mobile sensor nodes. Moreover, authors
highlighted the advantages/disadvantages and performance
issues of each routing technique. Aut hors in [] gave a
comprehensive taxonomy of the routing protocols based on
discovery, data transfer, routing, and motion control. is
survey classied the routing protocols into at routing and
proxy-based routing. In [], authors reviewed LEACH-M,
LEACH-ME, CBR-M, ECBR-MWSN, ER, L-LEACH-M,
FTCPMWSN, LFCP-MWSN protocols that are developed
for the MWSNs. In addition, they presented a comparison
between the reviewed protocols based on assumptions, CH
selection, location awareness, scalability, and complexity.
In [], the authors categorized the at- and hierarchical-
based routing protocols of MWSNs depending on their net-
work structure, state of information, energy-eciency, and
mobility. is classication summarizes the advantages and
disadvantages of the reviewed routing protocols. Moreover,
this survey gives an insight into the enhancements that
can be done to imp ro ve the existing routing protocols. A
survey of the location-based routing protocols that support
the sink mobilit y is presented in []. is survey classied
the location-based routing approaches into backbone-based
and rendezvous-based approaches according to their network
structures w hile highlighting the pros and cons of each type.
Asurveyin[]classiedtheexistingroutingprotocols
of MWSN s into delay-sensitive routing protocols and delay-
tolerant routing protocols based on delay, into centralized
and distributed protocols based on the routing decisions, into
discretely and continuously based routing protocols based on
mobility pattern, and into location-bas ed and topology-based
protocols depending on the routing information needed.
Also, this survey explained how each protocol works and
mentioned the advantages and disadvantages of each one.
Authors in [] surveyed the distributed routing protocols
that support mobile sinks. In addition, they explained the
challenges and the design requirements of a mobile sink-
based routing protocol. e surveyed protocols are compared
Wireless Communications and Mobile C omput ing
based on position awareness, sink mobility pattern, virtual
structure type, data aggregation, multisink support, protocol
overhead, structure accessibility, and hotspot mitigation. A
survey of eleven mobile sink-based routing protocols was
adoptedin[].eseprotocolsarecomparedbasedon
number of mobile sinks, type of protocol, and sink mobility
pattern. In [], a brief overview of the cluster-based protocols
for static WSNs, where the sensor nodes are xed, and for
MWSNs, where some or all the sensor nodes are mobile
in nature, was presented. Moreover, the surveyed protocols
are compared based on the basic assumptions, working
environment, advantages, limitations, and working style.
Asshowninthepreviousdiscussion,therearenoworks
focusing on reviewing the hierarchical routing protocols that
support the mobility of sensor nodes and/or sink node.
erefore, this survey paper will provide a comprehen-
sive review and ne-grained classication of the recently
hierarchical-based routing protocols that were developed in
the last ve years for MWSNs. e reviewed protocols are
classied based on routing approach, control manner, mobile
element, mobility pattern, network architecture, clustering
attributes, protocol operation, path establishment, commu-
nication paradigm, energy model, protocol objectives, and
applications. Based on delay, network size, energy-eciency,
and scalability, a comprehensive comparison between the
surveyed protocols will be considered in this paper while
analyzing the advantages and drawbacks of each protocol.
3. Taxonomy Criteria of Hierarchical-Based
Routing Protocols
e main aim of hierarchical-based routing in MWSNs is
saving the residual energy of each s ensor node, extending
the network lifetime, and ensuring the connectivity among
the sensor nodes. Here, we present a detailed taxonomy
and classication of the hierarchical-based routing proto-
cols based on dierent metrics. ese metrics are routing
approach, control manner, mobile element, mobility pattern,
network architecture, clustering attributes, protocol opera-
tion, path establishment, communication paradigm, energy
model, protocol objectives, and applications as shown in
Figure .
3.1. Routing Approach. e challenge of data routing can
have two broad methods, namely, classical-based method and
optimized-based method.
(a) Classical-Based Routing. In the classical-based routing,
head nodes are selected randomly using the timer function,
which causes an uneven trac ow in dierent head nodes.
Although they are suitable for applications of WSNs, they
stillhavemanychallengessuchasscalability,loadbalancing,
connectivity, coverage, and robustness.
(b) Optimized-Based Routing. Routing in WSNs puts new
challenges for researchers as various classical protocols lack
fault tolerance, energy-eciency, connectivity, robust ness,
and scalability. Researchers have developed some robust
routing protocols based on optimization algorithms such
as Particle Swarm Optimization (PSO), Fuzzy L o gic (FL),
Genetic Algorithm (GA), and Articial Bee Colony (ABC),
which provide opt imal solutions to the mentioned problems.
us in the optimized-based routing, the head nodes are
determined based on multicriteria to achieve the require-
ments of QoS.
3.2. Control Manner. On the basis of control manner, routing
approaches of MWSNs can b e centralized, distributed, or
hybrid approaches.
(a) Centralized Approaches. In the centralized approaches,
a sink/head node requires global information (e.g., energy
level, geogra phical position) of the network/group to control
the network/group. ese approaches are used to organize the
network into clusters and assign a head node for each one.
(b) Distributed Approaches. In the distributed approaches,
the sensor nodes collaborate with each other to build routes
without need for the global information of the network. Each
sensor node can execute its own algorithm and takes the
decision of becoming a head node or not. ese approaches
are used for coordination between the head nodes.
(c) Hybrid Approaches. Hybrid approaches combine the
features of being centralized and distributed.
3.3. Mobile Element. Since the network consists of a large
number of sensor nodes and sink nodes, the mobility can
apply to the sensor nodes and/or the sink node(s) depending
on the applications. erefore, the routing protocols can be
classied based on the mobile elements into protocols sup-
porting sink node(s) mobility, protocols supporting sensor
nodes, and protocols supporting mobility of both sensor
nodes and sink node(s).
3.4. Mobility Pattern. One of the main challenges in routing
of MWSNs is determining the moving pattern of the mobile
element (i.e., sensor nodes or sink node). Depending on the
application and network size, there are dierent mobility
patterns.
(a) Predened Mobility Pattern.Inthispattern,themobile
element moves along a predened path wi thin the sensor eld
andstopsatpredenedpositionstoperformaspecictask.
ispatterncanbeusedforthemobilesink.
(b) Random Mobility Pattern.erandommobilitypattern
canbeusedforthemobilesensornodesandthemobilesink.
In this pattern, the mobile element moves randomly within
the sensor eld. e synthetic mobility models are used for
simulating the random mobility of a mobile element such
as Random WayPoint mobility model (RWP) and Reference
Point Group Mobility Model (RPGM) [–].
(c) Contr olled Mobility Pattern. In the controlled mobility, the
mobile element is guided based on the control of the routing
protocol. e movement of the mobile element is controlled
based on some factors such as energy level, avoiding energy
W ireless Communications and Mobile Computing
Taxonomy criteria of hierarchical-based routing protocols in MWSNs
Routing approach
Classical-based
Optimized-based
Control manner
Centralized
Distributed
Hybrid
Mobile element
Sink
Sensor node
Sink and sensor node
Mobility pattern
Predened
Random
Controlled
Network architecture
Block-based
Tree-based
Chain-based
Clustering attributes
Cluster properties
Cluster size
Cluster density
Intracluster routing
Intercluster routing
Stability
Sensor capabilities
Homogeneous
Heterogeneous
Protocol operation
Negotiation-based
Multipath-based
Query-based
QoS-based
Coherent-based
Path establishment
Proactive
Reactive
Hybrid
Communication paradigm
Data centric
Node centric
Position centric
Radio model
First order
Realistic
Protocol objectives
Lifetime max.
Load balancing
Others
Applications
Time-driven
Event-driven
On-demand
Tracking-based
F : Taxonomy criteria of hierarchical-based routing protocols in MWSNs.
hole or hotspot problem [], and connectivity, to achieve
better results.
3.5. Network Architecture. e underlying network architec-
ture plays an important role in the function of the hierar-
chical routing protocols in MWSNs. Based on the network
architecture, the hierarchal routing protocols can be classied
into three categories, which are block-based hierarchical
routing, tree-based hierarchical rout ing, and chain-based
hierarchical rout ing.
(a) Block-Based Hierarchical Routing. In the block-based
hierarchica l routing protocols, the network is divided into
Wireless Communications and Mobile C omput ing
groups called clusters. Each one has a head node cal led
Cluster Head (CH) node that is responsible for collecting
andaggregatingthedataofitsMNsandthentransfersthe
aggregated data to the sink node. e main problem with
these protocols is how to select CH and the limited range of
sensor nodes to connect directly with sink.
(b) Tree-Based H ierarchal Routing. In the tree-based hierar-
chical routing protocols, a routing tree is formed among all
sensor nodes and the sink is the root of this tree. Leaf nodes
in the routing tree send data to their parent. Each parent node
aggregates the received data and sends it to the next level
parent node towards the sink.
(c) Chain-Based Hierarchical Routing.Inthechain-based
hierarchical routing protocols, one or more chains are con-
structed to connect the nodes for data transmission. A head
node for each chain ca lled leader is chosen to collect data
from the chain members. Data is delivered from the farthest
node from sink along the chain until the leader node forwards
thenalpackettowardthesink.However,thedatapacket
reaches the sink via a large number of hops, which increases
the packet delay. Moreover, the chain-based routing has less
ability of robustness because the failure of one node breaks
the chain and data will be dropped.
3.6. Clustering Attributes. e attributes of the clustering
process have an important eect on the performance of
hierarchical-based routing. e cluster properties and the
sensorcapabilitiesarethetwomainissuesintheclustering
attributes [, ].
3.6.1. Cluster Properties. In the hierarchical-based routing
protocols, the characteristics of the formed clusters are used
to dierentiate between these clusters in terms of saving
energy, load balancing, and lifetime.
(a) Cluster Size.Fromthepointofclustersize,the
hierarchica l-based routing protocols in MWSNs can be
grouped into equal and unequal clustering. In the equal
clustering protocols, all clusters have the same size, while
in the unequal clustering protocols, clusters have dierent
sizes. In general, unequal clustering protocols are used to
balancetheloadamongthenodesandsolvetheenergyhole
problem.
(b) Clu ster Density. Cluster density is the number of cluster
members. Density of cluster aects the energy consumption
ofCH.Clusteringprotocolscanbestaticclusteringprotocols
and dynamic clustering protocols. In the former one, clusters
have xed density, but the cluster density in the second one is
variable.
(c) Intracluster Routing.Intraclusterroutingisthecommuni-
cation between MNs and CH. is communication can be a
single-hop or a multihop routing. In the single-hop routing,
MNs directly deliver their data to CH. However in the multi-
hop routing, member nodes transmit their data to CH via
relay nodes.
(d) Intercluster Routing.Interclusterroutingisthecommu-
nication between the sensor nodes/CHs and sink node. e
intercluster communication can be a single-hop or a mul-
tihop routing. In the single-hop routing, sensor nodes/CHs
send their data directly to sink. In contrast to it, s ensor
nodes/CHs transmit their data to sink using intermediate
nodes in the multihop routing.
(e) Stability. e stability of routing process depends on the
cluster density. If the cluster density is xed, the stability of
routing is said to be xed. Otherwise, the routing stability
is considered variable because the cluster density varies
throughout the routing process.
3.6.2. Sensor Capabilities. Based on the resources of the
sensor nodes, MWSNs can be classied into homogeneous
and heterogeneous networks.
(a) Homogeneous Network. In the homogenous network,
all sensor nodes have the same energy, computation, and
communication resources. In this typ e of networks, CHs are
assigned according to a random manner or other criteria.
(b) Heterogeneous Network. In the heterogeneous network,
sensor nodes have unequal capabilities in heterogeneous
environment. erefore, the role of CHs is specied to sensor
nodes that have more capabilities.
3.7. Protocol Operation. Depending on the protocol opera-
tion, the hierarchical-based routing protocols are divided into
negotiation-based, query-based, multipath-based, coherent-
based, and QoS-based routing.
(a) Negotiation-Based Routing. In this type of routing, a high
level of descriptors is utilized for the negotiation between the
sensor nodes to minimize the duplicated information and
avoid the redundan t data. Generally, this negotiation should
be done before real data transmission between the source and
the relay node or the sink node.
(b) Query-Based Routing. is type of routing depends upon
queries from a destination. e source node transmits its data
in response to the received query from the destination node.
(c) Multipath-Based Routing. In the multipath routing , multi-
ple paths are constructed between a source and a destination
to increase the fault tolerance and enhance the network
performance.
(d) Coherent-Based Routing. Dierent data processing mech-
anisms are presented to save the processing computations
thatconsumeasignicantpartofthenodeenergy.Coher-
ent and noncoherent are the two main data processing
approaches that are used to save the consumption energy
in data computations. In the noncoherent data processing
technique, the sensor node processes the data lo cally and
then forwards it to the aggregator. Aggregator is a node,
which aggregates the received data from many sensor nodes
and forwards the aggregated packets to sink. In the coherent
method, a sensor node performs minimum data processing