International Journal of Wireless Information Networks (2021) 28:175–198

https://doi.org/10.1007/s10776-021-00508-9

A Comprehensive Review on Swarm Intelligence‑Based Routing

Protocols in Wireless Multimedia Sensor Networks
Fazilet Lemya Benmansour1 · Nabila Labraoui2

Received: 17 August 2020 / Revised: 24 October 2020 / Accepted: 11 February 2021 / Published online: 26 February 2021
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC part of Springer Nature 2021

Abstract
Wireless Multimedia Sensor Networks (WMSNs) have emerged as a new class of Wireless Sensor Networks (WSNs).
They are applicable in several applications specific Quality of Service (QoS) requirements, like area monitoring and video
surveillance. Recently, an intense research and considerable progress was conducted in solving numerous wireless sensor-
networking challenges. However, the problem of enabling real-time quality-aware video streaming or scalar data transmission
in WMSN is still open and largely unexplored. Unfortunately, transmitting multimedia data with reliability and in real time
is a challenging task. WMSNs are dedicated for critical and sensitive applications, and the QoS is required and constitutes
criterion to succeed the application. Routing protocols are then the building blocks of any WMSN transactions. Many WMSN
routing protocols have been proposed previously. The categorization of these protocols is related to the number and type of
QoS constraints. The eye-catching category in routing classification is the routing protocols based on Swarm Intelligence
(SI). Bio-inspired techniques are very attractive and consider biology as a source of inspiration by mimicking the dynamics
of natural species. The principle is to provide optimal routing solutions. In our work, we cover the details on how to define
and build smart routes to accommodate QoS-aware applications. In this paper, we present a comprehensive review that
specifically focuses on highlighting and describing all existing SI-based routing strategies for WMSNs. Moreover, detailed
classification and comparative analysis based on relevant metrics are presented. Design challenges and possible directions
for future research are also indicated.

Keywords Wireless multimedia sensor network (WMSN) · Swarm intelligence (SI) · Routing protocol · QoS

1 Introduction data (i.e. Temperature, pressure, humidity), usually via radio

transmitter to the sink either directly or through wireless
Technological advances in embedded microprocessors, multi-hops [2, 3]. WSN is deployed in numerous fields such
wireless communications and digital electronics have ena- as military, environmental monitoring, health care, security
bled the development of small and low-cost sensor nodes and surveillance in habitat and industry [4].
that made Wireless Sensor Networks (WSNs) one of the However, the integration of real time multimedia traffic
promising technologies during the past decade [1]. WSNs in WSN applications imposes novel requirements about sca-
[2] consist of a substantial number of interconnected and lar sensors capabilities. Fortunately, the rapid development
self-organized sensors with limited processing and energy and progress of sensors, in addition to the availability of
resources widely dispersed in a coverage area. The tasks of inexpensive Complementary Metal Oxide Semiconductor
these tiny sensors are to sense, process, and transmit scalar (CMOS) cameras and microphones, allowed for the emer-
gence of so-called Wireless Multimedia Sensor Networks
(WMSNs) [5]. WMSN is a new and emerging type of sensor
* Fazilet Lemya Benmansour networks that contain sensor nodes equipped with cameras,
fassilalamya.benmansour@univ-tlemcen.dz microphones, and other sensors producing multimedia con-
Nabila Labraoui tent (see Fig. 1). It is a network of wirelessly interconnected
nabila.labraoui@univ-tlemcen.dz devices that allow retrieving video and audio streams, still
1
LRIT Laboratory, University of Tlemcen, Tlemcen, Algeria images, and scalar sensor data [6]. The network imposes
2
new constraints related to the nature of the data captured and
STIC Laboratory, University of Tlemcen, Tlemcen, Algeria

13
Vol.:(0123456789)
176 International Journal of Wireless Information Networks (2021) 28:175–198

of the application in term of QoS. For all these reasons,

the design of a routing protocol should take into account
the requirements and resource constraints of the WMSN, in
order to meet the stringent QoS requirements.
In the literature, several research initiatives exist to
address the routing concerns in both WSNs and WMSNs.
The classification of the different routing schemes can be
done according to the classical approaches (reactive, proac-
tive, flat and hierarchical) or biological-inspired approaches.
In the last decade, bio-inspired routing emerged as an
alternative to the classical approaches and attracted many
researchers to improve the WSN/WMSN/Mobile ad hoc
networks (MANET)/Vehicular Ad-Hoc Network (VANET)/
Flying Ad-Hoc Networks (FANET) performance as well as
Fig. 1  WMSN architecture increasing the energy efficiency and communications effec-
tiveness. There are various bio-inspired and evolutionary
approaches, including genetic programming, neural net-
manipulated. Indeed, the multimedia data are very largely works, evolutionary programming, particle swarm optimi-
voluminous compared to the scalar data and having in addi- zation (PSO) and ant colony optimization (ACO) used for
tion a time constraint (real time). Furthermore, WMSN is routing protocols in ad hoc and sensor wireless networks
an exciting new technology that has not only extended the [9]. In this paper, we particularly focus on the Swarm Intel-
horizon of traditional sensor network applications, but also ligence (SI)-based routing protocols (ACO, PSO, ABC, etc.).
proliferated significantly to creating a series of new applica- SI [10–13], which is an Artificial Intelligence (AI) dis-
tions such as multimedia surveillance, traffic monitoring and cipline, is considered as an optimization technique used to
enforcement, gaming, personal and health care, environmen- solve various complex problems in diverse areas. It is a novel
tal and industrial [7]. field, originated from the study of the collective behavior of
In addition to the challenges raised by WSNs in terms decentralized, self-organized natural systems in interaction
of energy limitation, deployment, coverage, reliability, among themselves and with the environment. These multi-
WMSNs have more functionalities and exigency, such as agent intelligent systems take inspiration from the collective
high bandwidth demand, limited delay, real-time delivery, behaviors of insect societies, flocks of birds, and schools of
acceptable jitter and low packet loss ratio. These character- fishes, often observed in nature.
istics impose more resource constraints that involve energy Studies have shown that routing protocols based on such
consumption, memory, bandwidth and processing capabili- bio-inspired methods bring many improvements and offer
ties [8]. The objective is to be able to meet Quality of Ser- better performance in terms of energy efficiency, reliabil-
vice (QoS) requirements for multimedia data that aim to ity, scalability, adaptability and robustness [14]. In fact, the
confront the majority of the aspects described above. These design of most of these algorithms is inspired by the for-
mentioned constraints and requirements affect the design of aging behavior of insects to solve routing problems. They
WMSNs and lead to the need to develop protocols in order establish paths that can be used by individual insects to
to maximize the network lifetime while satisfying the QoS move back and forth efficiently from the nest of the colony
requirements of the various applications [8]. to the food sources. These paths are the result of synergis-
Similar to the WSNs, the main task of sensor nodes in tic interactions between a large number of individuals that
WMSN is to monitor the target area and transmit the col- inform others of their characteristics using various commu-
lected information to the sink node in order to perform spe- nication schemes [15].
cific processing according to the needs of each application. By going through the literature, most of existing rout-
In order to deliver data packets to the destination, sensor ing-related surveys are for WSNs. So far, many survey
nodes require the routing protocols. Routing protocols are papers that present and compare SI-based routing proto-
used for finding the best path to establish communication in cols for WSNs are published. However, over the past dec-
the networks. However, designing an efficient routing proto- ade, very little researches have been done on WMSNs and
col for WMSNs’ applications is a challenging task due to the in particular on WMSN routing protocols. Most of these
nature and abilities of sensor nodes such as energy supply, are limited in the context of reviewing the classical rout-
computation and communication abilities, deployment of ing protocols. Only few related surveys evoke SI-based
nodes, communication architecture, the unreliability of the routing protocols in their paper. For this purpose, and to
wireless links and the different performance requirements our knowledge, we have not discovered any review that

13
International Journal of Wireless Information Networks (2021) 28:175–198 177

exclusively treats SI-based routing protocols for WMSNs. The remainder of this paper is structured as follows (see
Therefore, and to bridge this gap, this research manuscript Fig. 2). In Sect. 2, we present the main factors to consider in
intends to present a comprehensive and state-of-the-art routing protocols design for WMSNs. After going through
research work on SI-based routing in WMSNs. the literature, we report, in Sect. 3, the different surveys on
The main contributions of this research are described SI-based routing protocols in WSNs and WMSNs. In Sect. 4,
as follows: we first discuss the SI principle. Then we classify all exist-
ing SI-based routing protocols proposed in the literature for
1. We carried out an in-depth research of various existing WMSNs and describe every protocol in detail. A compara-
classical and SI-based routing protocols for WSNs and tive summary of these protocols is also provided in a tabular
WMSNs since the appearance of the first to this day. form in Sect. 5. An overview of future research directions is
2. We give a taxonomy of the different SI-based routing given in Sect. 6. Finally, we conclude the paper in Sect. 7.
protocols designing for WMSNs by classifying them
according to their features. This categorization will
help readers to further comprehend and analyze these 2 Design Goals for WMSN Routing
approaches.
3. We provide a clear overall understanding of WMSNs The design of routing protocols must take into consideration
by presenting their characteristics and outlining some the performance of WMSNs in terms of QoS and energy
design requirements that routing protocols should con- efficient.
sider. Applications such as security surveillance, health sys-
4. An exhaustive survey of emerging WMSN routing tems and traffic management systems that involve multime-
techniques based on SI is highlighted since the appear- dia content must be delivered in real-time. The multimedia
ance of the first in 2008 until today. We explain in detail data collected are voluminous and require high transmission
operation, objectives, features, limitations, simulation rates and processing capabilities.
and experimental result of each algorithm. In WMSNs, requirements such as real-time multimedia
traffic and handling data volumes raise many challenges and
This contribution raises awareness among a large
resource constraints to the routing protocols design. These
audience about the existence and the performance
problems are addressed by many researchers as surveyed in
of the SI-based solutions.
[6, 16, 17].
5. After listing all SI-based WMSN routing protocols, we This section outlines the different challenges that should
first established a comparative analysis of these proto- be considered for improving the communication efficiency
cols based on prominent metrics such as network topol- in these networks.
ogy (flat or hierarchical), route selection, multipath,
location awareness, congestion control, classification • Energy efficiency: WMSNs deliver multimedia content
service, energy efficiency, load balancing and QoS consisting of high volumes of data, which require high
parameters considered. Then, we give a set of informa- transmission rates and processing capabilities. Sensors
tion on the simulation approach that was used for each in WMSNs usually consume even more energy than in
solution studied. WSNs. Therefore, energy consumption is one of the most
6. We provide an insight into the promising future research important performance metrics to consider when design-
directions that can improve the use of routing protocols ing the routing protocol in order to prolong the network
in WMSNs. This will contribute to the development of lifetime.
this area of research. • QOS requirement: Delivering QoS and considering
application-specific requirements remain significant
challenges in the design of WMSNs applications. In

Fig. 2  Document structure Review of SI-based Comparison of

Introduction Design goals for Literature
routing protocols in SI-based routing
WMSN routing Survey
WMSNs protocols in WMSNs

ASAR, M-IAR, AGRA, LBHR, Open research

Previous Previous
AntSensNet, ACOWMSN, issues
surveys on SI- surveys evoking
SI-based routing IEEABR, Enhenced AntNet,
based routing
protocols in protocols in [99], [109], IPACR, ICACR,
WMSNs AntQHSeN, [110], EAQHSeN,
WSNs Conclusion
EABC-EELB-PWDGR

13
178 International Journal of Wireless Information Networks (2021) 28:175–198

WMSNs, most applications require real-time multi- 3.1 Previous Surveys on SI‑Based Routing Protocols
media data delivery. Therefore, different QOS met- in WSNs
rics like delay, bandwidth, jitter and reliability will be
imposed on WMSNs as a result according to the type In the WSN, many surveys have focused on the SI approach
of the application. and have only presented routing protocols based on SI [14,
• High bandwidth demand: Multimedia content requires 15, 49–61]. While in other reviews, both classical and SI
high transmission bandwidth. In order to optimize routing protocols have been studied [37, 62–65].
resource constraints of WMSN, new transmission
techniques are needed to provide the required band-
width with low power consumption. 3.2 Previous Surveys Evoking SI‑Based Routing
• Multimedia source coding techniques: Coding must be Protocols in WMSNs
taken into account as it is directly related to the size
of the data. In WMSNs, multimedia devices such as In WMSNs, most of surveys are limited in the context of
cameras generate voluminous multimedia content that reviewing the classical routing protocols [66–79]. Despite
is why multimedia transmission requires a multimedia the considerable research on SI-based routing protocols in
source coding techniques. One of the main aims of WMSNs, a very small number of surveys evokes some of
coding techniques is reducing the information amount these protocols in their investigation.
sent in the network by using efficient compression In Table 1, we discuss the reviews and surveys that
techniques. evoked SI-based routing protocols that have been done so
• Multimedia In-network Processing: In WMSNs, mul- far in the literature.
timedia in-network processing algorithms are applied Thus, there is no comprehensive paper that examines
to the raw data sent from the environment. New archi- all the SI-based WMSN routing protocols in an appropri-
tectures for collaborative, distributed, and resource- ate classified manner. Therefore, the main objective of our
constrained processing are required in in-network pro- review paper is to provide direct access and an initial reading
cessing for filtering and extracting information needed point for future beginning researchers.
to the periphery of the network. Thus, the scalability
of the system increases by decreasing redundant data
transmission. Therefore, to enable flexibility in-net-
work processing of multimedia content, it is important 4 Review of SI‑Based Routing Protocols
to develop application-independent architectures. in WMSNS

WMSNs require the delivery of multimedia content with a

certain level of QoS. However, multimedia content is dif-
3 Related Work ferent from scalar data, especially with regard to the size
of the data. Therefore, routing in WMSNs requires special
There are many ways to classify routing protocols depend- attention because the large size of the data reduces the effi-
ing on different parameters. The classes and subclasses ciency of data transmission. Some routing protocols have
are not mutually exclusive as many protocols belong to been developed according to the principles of SI. Thus, the
more than one class or a subclass. Therefore, the classi- majority of research in this area has focused on energy effi-
fication of the different protocols can be done according ciency and QoS. In this section, we list all the existing SI-
to the classical approaches (based on path establishment, based routing algorithms dedicated for WMSNs and offered
network structure or protocol operation) and based on SI. in the literature from the first in 2008 until today.
Concerning WSNs, the bulk of research in the domain Figure 3 illustrates the proposed taxonomy. Bio-inspired
of routing has been done following a non SI-based routing applied to WMSNs is divided into three main cat-
approach. So far, several surveys of classical routing egories: Evolutionary algorithms, Swarm intelligence algo-
protocols exist in the literature on a variety of different rithms and other biologically inspired algorithms.
research domains. The major important studies of routing The evolutionary algorithms [87] are computational
schemes existing are described in [18–33]. While other techniques inspired by natural evolution using iterative pro-
surveys are more specific to certain aspects, such as gress such as inheritance, growth, development, reproduc-
energy-efficient routing [34–38], hierarchical/clustered- tion, selection, and survival as seen in a population. They
based routing [39–43], multipath routing [44, 45], and include Genetic Algorithms (GAs), Evolutionary Program-
real time routing [46–48]. ming (EP) and Evolutionary Strategies (ES). These are pop-
ulation-based stochastic search algorithms operating with

13
International Journal of Wireless Information Networks (2021) 28:175–198 179

Table 1  Description of previous surveys evoking SI-based routing protocols in WMSN

Year of survey References Survey characteristic Survey description

2012 [80] Energy-efficient QOS routing Ehsan and Hamdaoui presented a classification and description of
some energy-efficient QOS routing techniques for WMSNs. This
paper was the first to have evoked the notion of routing protocols
based on SI for WMSN in its classification. In this category, only
three routing protocols were discussed briefly
2013 [17] Classical routing The paper published by Abazeed et al. proposes a classification and
discussion concerning different routing protocols for WMSNs
with their features and limitations. As well, a general comparison
between them is made. SI routing protocol is a category in which
four routing protocols have been defined (ASAR, ACOLBR,
M-IAR, and Ant-like Routing Algo)
2014 [81] ACO-based routing Another paper from Abazeed et al. focused on ACO routing
protocols designed for multimedia transmission. A description of
the ACO principle and an explanation of the ACO-based routing
protocols for WMSNs were done. Also, a comparison between
them was realized. This review was the first to be interested in
the SI approach and more particularly to ACO, but also to study a
certain number of routing protocols based on ACO
2015 [82] Secure routing approaches for next generation In a third paper, Abazeed et al. addressed the concept of secure
routing in WMSNs. A set of routing protocols based on SI (based
on ACO) and on security was identified for WMSNs. A compari-
son between the mentioned routing protocols is summarized in a
table according to various QoS parameters and others
2015 [83] QOS routing Aswale and Ghorpade proposed a classification of QoS routing
protocols in WMSNs against a number of routing metrics as a
categorization criterion and summarized the analysis in a com-
parative table by indicating the strengths and weaknesses of each
protocol. The SI aspect has not been considered in this article.
In the description of the different techniques, only five routing
protocols based on SI are reviewed
2016 [84] Classical and swarm intelligence routing A survey of various existing protocols for WMSNs is presented
by Bhandary et al. in which these are examined and categorized
based on different characteristics into three main categories. An
analytical comparison of protocols by category is carried out.
SI-based routing is one of the categories in which a significant
number of protocols have been listed and discussed, whose some
are not intended for WMSNs
2016 [85] Classical routing Shen and Bai present an exhaustive study on WMSNs routing pro-
tocols classified into five major categories based on their design
and optimization objectives. A detailed comparison of each solu-
tion is provided according to various parameters. SI-based routing
is a sub-sub-category of one of the above mentioned categories
called intelligent routing in which only two routing protocols have
been reported
2018 [86] Classical and swarm Deb and Choudhury present a study of existing routing approaches
intelligence routing used for WMSNs. The routing protocols cited were classified
according to four categories and compared according to some
characteristics. The biologically inspired category includes
evolutionary based and certain SI algorithms (ASAR, M-IAR,
ACOLBR, AntSensNet, AntHQSeN and AGRA)

best-to-survive criteria [88]. In WMSNs, A Crossover Game processes have served as an inspiration for various algo-
Routing Algorithm [89] has been proposed combining the rithms. Cell biology and bacterial foraging are two inter-
genetic algorithm (GA) with game theory. esting bio-inspired approaches used for WSN routing.
Besides swarm intelligence and evolutionary algo- Works that apply cell biology [90, 91] are inspired by
rithms, which represent the two main categories of the similarities that exist between organisms and networks:
bio-inspired algorithms, a number of other biological

13
180 International Journal of Wireless Information Networks (2021) 28:175–198

Fig. 3  Classification of SI-based routing protocols for WMSN

organisms consist of organs, tissues and cells, while a net- community members and the collaboration between them
work consists of clusters and nodes. brings out possibilities for representation, creation and learn-
Following the principle of swarm-based algorithms, the ing that are superior to those of isolated individuals. Swarm
Bacteria Foraging Optimization Algorithm (BFOA) [92] is is used to refer to a finite set of particles or interactive
inspired by the foraging behavior of a group of Escherichia agents. Thus, by imitating the social behavior of particles
coli swarms (E. coli). It is a bacterium that lives in the intes- forming swarms capable of self-organizing, several algo-
tines of some mammals and uses tumbling and swimming as rithms have been proposed in recent decades among them,
it travels to the food source. we find Ant Colony Optimization (ACO) [93] and Artificial
Our interest in this survey relates to the second category, Bee Colony (ABC) [94], the most widely used. These SI-
as shown in Fig. 3, which consists of listing all the WMSN based techniques are artificially simulated and used to solve
routing protocols inspired by SI such as Ants and Bees. Two routing problems in wireless networks.
sub-classes can be identified in this category, according to Therefore, SI proposes efficient and robust metaheuristic
the existing routing protocols; Ant Colony Optimization algorithms to solve several problems in WSNs, as reviewed
(ACO) and Artificial Bees Colony Optimization (ABC). We in [15, 62]. Routing constitutes an optimization problem,
established a taxonomy for ACO because the majority of the which, by analogy, can be solved with many swarm intel-
research work has been inspired by the behavior of ants. In ligence based algorithms.
this category, three sub-classes can be observed: A. Energy-
efficient based, B. QoS-aware based and C. QoS-enabled and 4.1.1 Ant Colony Optimization (ACO)
energy efficient based. Protocols fall under each category
have also been listed in the figure. Concerning ABC, only ACO [93, 95] is the most famous SI inspired algorithm. It
one sub-class based on QoS-enabled and energy efficient draws inspiration from the intelligent and foraging behav-
includes only one routing protocol taken inspiration from ior of real ant colonies. Ants use indirect communication
Bee behavior. through the secretion of chemical pheromones, which allows
them to find shortest paths from the food source to their
4.1 The Concept of Swarm Intelligence nest. In their natural environment, ants move surrounding
their nest in search of food. Once the food is found, the
Considered as bio-inspired algorithms, Swarm Intelligence- ant deposits pheromones, leaving behind traces of the path
based algorithms [10] are inspired by natural phenomena taken. The ants behind can smell these substances. When
such as the collective intelligence of communities, observed choosing its way, each ant tends to choose, in probability,
in particular in social insects (such as ants, bees, animals the path with the highest pheromones. Therefore, the path
evolving in groups such as migratory birds, schools of fish). likely to be taken by the majority of ants turns out to be the
Collective intelligence refers to the cognitive capacities of a shortest path to the food source. This reinforcement of the
community resulting from the multiple interactions between path is due to strong pheromone concentrations deposited

13
International Journal of Wireless Information Networks (2021) 28:175–198 181

with each passage of an ant. Nevertheless, less visited paths surpasses all protocols in term of low energy consump-
will see their pheromone trail gradually evaporate over time, tion. In summary, it outperforms its competitors in most
thereby reducing their attractive strength. of the assumed scenarios and metrics.
Most routing protocols intended for WMSNs are devel-
oped using ACO approach. These solutions are described 4.1.3 QoS‑Aware Routing Protocol
below divided into three categories.
• Enhanced AntNet Protocol for WMSNs
4.1.2 Energy‑Efficient Routing Protocol   Bennis et al. [98] propose an optimization of the basic
AntNet routing protocol to adapt it to WMSNs and pro-
• Improved Energy-Efficient Ant-Based Routing Algo- vide better QoS in terms of delay and Packet Delivery
rithm (IEEABR) Ratio (PDR). Several improvements of the initial version
  Zungeru et al. [96] have developed an enhanced ver- of the protocol were made such as the adaptation of the
sion of energy-efficient ant-based routing EEABR [97], protocol in a wireless and mobile environment in order
energy-aware, based on ACO and designed for Visual to make it more robust and more desirable for WMSNs.
Sensor Networks (VSNs). This protocol named IEEABR The goal is to minimize the number of hops, modify the
addresses the specific requirements of video and image way of choosing the next hop, make the protocol reactive
applications that require high bandwidth. It uses the and add lists that allow storing information such as an
self-organization, self-adaptability of ACO to determine ancestor list, destination list, etc.
multiple optimal paths while avoiding using the energy   The simulation results of the protocol with NS2
of nodes on the optimal path. It considers the available showed that compared to AODV, the proposed protocol
power of nodes and improves memory usage. The modi- has better packet delivery ratio, end-to-end delay and
fications improve energy consumption of nodes and net- overhead and these improvements are more readable
work performance. Therefore, the routing tables of each especially when the number of node increases.
node are intelligently initialized, a priority is given to   The disadvantage of this protocol is that it has not been
neighboring nodes of source, which must be the destina- compared to other routing protocols designed for WMSN
tion, and the congestion is controlled by reducing the to accurately evaluate its performance.
flooding ability of ants in the network. As in EEABR, • A Routing Optimization Based on Ant Colony for
the protocol uses Fants to find a path to the destination. WMSNs
Each ant only carries the address of the last visited nodes.   To transmit multimedia data, a routing optimization
Every node saves the previous node, forward node, ant approach has been proposed by Putra et al. [99] based
identification and a timeout value in a memory carried by on ACO to guarantee the bandwidth or QoS in WMSN
the ant. Among the neighbor, the node, which falls to be applications. The ACO proposed a technique that allows
the destination, is assigned the highest probability. The to determine best selected paths between sensor nodes
Bants are generated to update the pheromone trail of the and a sink. In spite of the variation of the bandwidth
path traversed by the Fant. The objective is that nodes channels of WSNs, the goal of the protocol is precisely
near the sink have more pheromone levels and therefore to adapt to the varying bandwidth conditions.
force the distant nodes to find better paths.   The performance parameters that were taken into
  IEEABR approach has been compared to its prede- account to ensure the QOS were throughput and packet
cessor EEABR and to some ant-based routing protocols delivery ratio (PDR).
such as Basic Ant-Based Routing (BABR) Algorithm,   The optimization process is done in relation to sev-
Sensor-driven and Cost-aware ant routing (SC), Flooded eral levels of problems encountered. The number of lay-
Forward ant routing (FF) and Flooded Piggybacked ant ers represents the number of variables of the problems
routing (FP). These algorithms use Routing Modeling encountered and the number of nodes of a layer corre-
Application Simulation Environment (RMASE) and the sponds to the number of discrete values allowed
​​ for the
obtained results in each static or dynamic scenario are linked variables. Thus, each node is associated with a
compared according to certain performance metrics like: possible discrete value for a variable.
latency, success rate, energy consumption and energy   Pheromone tables are always updated by forward and
efficiency. backward ants. Thus, the routing tables will be updated
  The results showed that IEEABR has the lowest according to the variable bandwidth conditions.
latency and a high success rate in packets delivery. In   The approach was simulated with the NS2 simula-
terms of energy consumption, EEABR consumes 31% tor. Two flows of UDP and TCP data were sent through
and 29.66% of energy than IEEABR for static and the WMSN and three routing protocols were simulated
dynamic scenario respectively. This means that IEEABR namely the proposed protocol ACO, Ad hoc On-Demand

13
182 International Journal of Wireless Information Networks (2021) 28:175–198

Distance Vector (AODV) and Destination Sequenced frequency of reverse ants to accelerate the convergence
Distance Vector (DSDV). The results of the simulation of ASAR.
show that routing optimization can provide better perfor-   ASAR considers the routing scheme between the CH
mance in terms of PDR and the throughputs and better and the sink. It is made to run in all CHs. For each ser-
QoS than DSDV and AODV routing protocols. There- vice, it searches three QoS routes by using a positive
fore, the ACO algorithm proposed allows to optimize feedback mechanism adopted in Ant Colony Optimiza-
routing techniques for multimedia applications on the tion (ACO) and the results are stored on CHs. For differ-
WMSNs. ent services, all CHs manage three optimal path tables,
three pheromone tables, the real time pheromone value
and transition probability for its next hop.
4.1.4 QoS‑Enabled & Energy Efficient Routing Protocols   A simulation with NS2 was performed to evaluate the
performance of ASAR with traditional ant-based algo-
• An ant-based Service-Aware Routing algorithm for rithm, Dijkstra and conventional directional diffusion
Multimedia Sensor Networks (ASAR) (DD) in WSN. The results show that the effectiveness of
  ASAR, proposed by Sun et al. [100], is a QoS hierar- the ASAR algorithm depends on differentiated services.
chical routing protocol for WMSNs based on the tradi- Also, it offers better convergence and better QoS for sev-
tional ant algorithm. Routing selection involves the dis- eral types of services in WMSN.
covery of three service-aware accessible paths to meet • Multimedia-Enabled Improved Adaptive Routing
different QoS requirements, thus optimizing network uti- (M-IAR)
lization, minimizing interference between the three types   M-IAR, proposed by Rahman et al. [101], represents
of services, balancing traffic distribution and improving an extension of the Improved Adaptive Routing (IAR)
network performance. It is a new service-oriented mul- algorithm [102], incorporating wireless multimedia sen-
tipath routing selection scheme. This protocol is intended sors and targeting multimedia applications in WSNs. It is
for three basic services provided by multimedia sensor a flat multi-hop routing protocol that uses the geographic
networks and who are: location of the sensor nodes, by assuming that each node
knows its own position as well as that of its neighbors
• Event-driven service mode: R service must meet the and the sink, in order to find the shortest path having the
requirements of real time and reliability. It is delay least number of nodes between the sending and receiving
and error intolerant. This requires less bandwidth, a nodes. It uses the ACO concept to optimize some QoS
path with little traffic and high signal-to-noise ratio. parameters like end-to-end delay and jitter.
• Data query service mode: the data received by ser-   The protocol considers two kinds of ants: Forward
vice D must always be as reliable as possible. It is ant (Fant) and Backward ant (Bant). At the beginning,
error intolerant but query-specific delay tolerant. Fant is used by the source node to explore the route
This requires a path with significant congestion and toward the sink. A probability is calculated for each of
a high signal-to-noise ratio on each link. its neighbors in order to choose its best neighbor node
• Stream query service mode: S services are delayed taking into account the distance to the sink and the sender
intolerant, but query-specific error tolerant. This node. Based on this information, the node with the high-
requires a path with less traffic and lower signal-to- est probability value is selected to become the next for-
noise ratio. warder and the routing table is updated. In its packet
header, every Fant has a set of global parameters like
  The proposed routing algorithm is designed based on the nodes visited and their neighbors, the corresponding
the cluster-based architecture. In the cluster, nodes are probability values, the total number of visited hops and
able to collect and process multimedia data. The cluster the distance for each link. Intermediate nodes follow the
head (CH) merges these multimedia data and then trans- same procedure until reaching the sink. Once the desti-
fers them upstream. The sink node manages the status nation is reached, Fant generates Bant and sends it via
of CHs and broadcasts signals in the network. The CH the same visited path with the same global header infor-
connects the sink node via multi-hop wireless links. mation to reinforce the visited nodes by increasing the
  For each kind of service (R/D/S service), every CH probability.
generates ants to discover three different QoS paths from   The proposed protocol was simulated and the results
each CH to the sink. The QoS parameters used for the demonstrate that it presents low jitter and low end-to-end
selection of the routing are latency, packet loss, energy delay what can be beneficial for multimedia traffic. More-
consumption and bandwidth. So, the pheromone value over, in 98% the cases, M-IAR can find the shortest path
is quantified on the sink in order to decrease the sending with only three route discovery attempts by consuming

13
International Journal of Wireless Information Networks (2021) 28:175–198 183

less energy, visiting less number of hops, and providing cost and some suboptimal paths from the source cluster
high packet success rate. Unfortunately, the simulation head to the sink. The suboptimal path is used by splitting
results obtained by M-IAR were not compared to any the stream when the amount of data exceeds the path flow
other routing protocols. threshold. This improves the efficiency of the transmis-
• Ant-like Game Routing Algorithm for WMSNs sion and guarantees the QoS of the network.
(AGRA)   When a node sends its data to the next node, Fant
  The ant-like game routing algorithm, namely AGRA updates the pheromone using the local pheromone update
[103] is a combination of ACO and game theory to solve rule. Once in the sink, the Bant releases more phero-
the QoS routing problem of WMSNs. It is an energy effi- mones on the selected path using the global pheromone
cient QoS routing algorithm that uses game theory to update rule in order to reinforce the current optimal path.
improve the performance of the ant colony algorithm. Thus, LBHR improves the ability to explore ants by
The goal is to find optimal routing path from source to introducing the maximum and minimum of pheromone
sink node providing the QoS guarantee. density values. The same process is repeated until the
  Game theory uses three concepts: a set of players, second and third suboptimal paths are found. During the
strategies of each player and the payoff function of each transmission, in the event of a node failure, the neighbor-
one. The assumption is that each player tries to maximize ing node will set the pheromone value to zero and send
his payoffs with minimum cost. The model uses mixed an error message to the source node. Then the source
strategies game. node will stop the transmission on this path and activate
  In the proposed routing algorithm, the node uses local alternate path for transmission. However, if the next node
information to find a path according to the game result. finds that the end-to-end delay of a path from the source
When sending a packet from a sensor node, Fants are sent exceeds a certain threshold, it will send a congestion
to find the path to the sink. Based on the pheromone trails message to the source. Then the source node will reduce
left by ants, the probability is calculated using residual the data amount transmitted in this path and activate an
energy, delay and bandwidth to define payoffs of game alternate path to ensure the reliability.
players. All ants select next hop with the probabilities of   Finally, the intra-cluster routing is constructed using
mixed strategy Nash equilibrium of routing game. the minimal spanning tree (MST) algorithm with the
  Once the Fants successfully reach the sink, a Bant is cluster head as the root. Therefore, a hierarchical routing
generated and takes the path found by Fants towards the tree is established and the ability of each cluster member
source. decreases from top to bottom.
  A soon as, the Bants return to the source node, Fants   Thus, the cluster member sends the collected data after
are generated again, and each ant selects the next hop aggregation to cluster head through the intra-cluster hier-
with the result of routing game until all ants select the archical routing tree, while the cluster head sends the
same routing. aggregated data to the sink using the inter-cluster routing
  The algorithm has not been simulated to be compared scheme.
to others.   The algorithm was simulated using the Network
• Load balancing-based hierarchical routing algorithm Simulator NS2 and was compared to clustered-control
for WMSNs (LBHR) (CC) and M-IAR algorithms according to the follow-
  Li and Wang proposed LBHR protocol [104] to ing parameters: end-to-end delay, network lifetime, the
improve the QoS of data transmission in WMSNs. The transmission success rate and communication overhead.
algorithm is inspired by ACOLBR (Ant Colony Opti- The results obtained demonstrate that LBHR reduces the
mization-based Load Balancing Routing Algorithm for end-to-end delay, increases the transmission success rate
WMSNs) [105]. and achieves load balancing. Thus, the protocol has better
  LBHR is a bio-inspired hierarchical routing protocol adaptability and scalability, can in fact prolong the net-
based on QoS used for load balancing and includes the work lifetime and guarantee the QoS of data transmission
clustering algorithm, the inter-cluster routing and the in WMSNs.
intra-cluster routing. • Ant-based multi-QoS routing metric for WMSNs
  First, the hierarchical architecture of the network is (AntSensNet)
constructed from a new clustering algorithm, which   AntSensNet protocol proposed in [106] by Cobo et al.
allows both the selection of cluster heads and cluster is a QoS routing model specially designed for WMSN
formation. Each node within a cluster can become the that combines the principles of ACO metaheuristic with
cluster head if its weight is the smallest. hierarchical architecture, thereby improving network
  Then, the inter-cluster routing is built by the improved performance and optimizing its use. Moreover, it holds
ACO algorithm to find an optimal path with minimum a power efficient packet scheduling scheme to achieve

13
184 International Journal of Wireless Information Networks (2021) 28:175–198

minimum video distortion during transmission through Forward ants (Fant). Along the way, the Fants collect
multiple paths. The proposed algorithm is a hybrid rout- traversed nodes’ IDs together with the four QoS metrics
ing protocol that takes into account reactive and pro- of the nodes passed by. When a CH relay node receives
active components for efficient route establishment. It a Fant, its information is updated before sending it to the
chooses the suitable path based on a combination of four next hop. Once Fant arrives at the sink, the routes found
QoS routing metrics: packet loss rate, available memory, are evaluated. When the route meets the application
queuing delay, and normalized remaining energy in order requirements, a Bant is generated and forwarded through
to meet the QoS requirements requested by the appli- the reverse path, updating the pheromone values. In case
cations while minimizing the consumption of limited of congestion or link loss issues, the protocol generates a
resources. maintenance ant to inform neighborhood CHs to update
  The authors take into account the network classifica- their pheromone tables and to find alternative routes.
tion and believe that each class of traffic should be treated   Data Ants (D ant) assist in route discovery and mainte-
with its own appropriate QoS metrics. Thus, the aim of nance, their behavior is similar to that of FANTs. A Dant
this algorithm is to determine accessible routes for each is assigned to transport urgent or real-time data from a
class of traffic from a transmitting CH node to the sink node to the sink and is processed before all other traffic
that satisfy various QoS requirements by introducing a classes in each node.
packet scheduling policy which considers different pri-   Several simulations were performed with NS2 and
orities for each traffic class. the results showed that AntSensNet converges better and
  The algorithm works in three steps. The first is to cre- improves QoS for multiple types of services in WMSNs
ate clusters in the network, drawing inspiration from the compared to ad hoc on-demand distance vector AODV
behavior of ants. Next, search for network routes between and T-ant. Regarding the transmission of video packets,
clusters satisfying the criteria of each application using the video quality is much better with AntSensNet than
ants. Finally, transfer different traffic using the routes pre- with TPGF or ASAR because these do not handle video
viously discovered by the ants. content correctly.
  The clustering approach is used to ensure the scal- • An Ant Colony Optimization-Based QoS Routing
ability, improve network data aggregation mechanisms, Algorithm for WMSNs (ACOWMSN)
thereby reducing node workloads, saving energy and   ACOWMSN is a routing algorithm based on ACO,
increasing the network lifetime. presented by Yu et al. [108], for the transmission of video
  AntSensNet includes three operating phases: cluster- and imagery data in WMSNs based on QoS and energy
ing process, route discovery and data transmission. efficiency. The algorithm is designed to find an optimal
  The clustering process is based on an improved ant path that meets application-specific QoS constraints
colony algorithm T-ant [107] and is made up of rounds. and aims to prolong the network lifetime. It is a reactive
Each round is divided into cluster setup phase and a routing protocol using packet loss rate, queuing delay,
steady phase. In the steady phase of the algorithm, data bandwidth and remaining energy as QoS constraints
transmission takes place between sources and the sink. to determine the path with minimum routing cost from
In the cluster setup phase, cluster ant (Cant) controls the the source to the destination. In routing modeling, the
selection of CHs in a totally distributed manner. A node algorithm involves two phases: routing discovery and
with a Cant become a CH and the rest of the nodes join routing confirm, thus including two types of artificial
the cluster. An information update phase is carried out ants: forward and backward ants. In routing discovery
by the sensors before the cluster setup phase and con- phase, Fants are sent to the destination node using uni-
sists of broadcasting a HELLO packet with the node’s cast or broadcast. Broadcast approach is used only when
ID, its clustering pheromone value, and its state to its current node does not have information about the sink.
neighbors. On arrival, the stored information in a table Fants collect information like minimum residual energy,
of neighborhood is used then to decide how to join the cumulative queue delay, packet loss ratio and available
cluster and route discovery. The suitability of a node to memory of each node visited in order to find the next
become CH depends on the clustering pheromone value. optimal neighbor node towards the destination. For each
An adjustable factor called “cluster radius” determines node, next forwarding node is selected by calculating the
the minimum distance between two CH nodes. probability that combines the pheromone value with the
  After clusters formation, the CH begins the route dis- residual energy. In routing confirm phase, when the Fant
covery process. Each CH maintains a pheromone table reaches the destination, it will be converted into Bant
for its neighbors in each traffic class according to the and return to the source node in the opposite direction
QoS parameters. To find an appropriate path to the sink along the same path. Bant updates the local node status
node for a specific traffic class, the CH source broadcasts and the pheromone concentration value of every visited

13
International Journal of Wireless Information Networks (2021) 28:175–198 185

node on the traversed path based on the path quality and   The simulation results state that increasing the num-
the end-to-end delay. ber of ants increases the probability of finding the opti-
  The simulation of the proposed algorithm with NS2 mal path, but results in increasing the computation time.
shows that by comparing it to M-IAR and AODV accord- Increasing the pheromone weight means increasing the
ing to the end-to-end delay, network lifetime and packet importance of the following pheromone trails and would
delivery ratio, the algorithm offers better performance lead in unsatisfactory path. Increasing the heuristic
than the other algorithms and increases the network life- weight means increasing the importance of the heuris-
time. In addition, it is relatively more reliable and scal- tic value which is the real cost of the path. In conclu-
able than the other ACO-based protocols. Notwithstand- sion, to find the optimal path, the heuristic value must be
ing, it is not suitable for large-scale WMSNs because more important than the pheromone value. An increase
determining parameters is not easy and therefore the in evaporation rate would result in faster evaporation of
algorithm does not converge quickly. pheromone trails and reduce the probability of finding
• Optimal routing protocol based on ant colony optimi- the optimal path.
zation in MWSNs   An increase in the transmission bit rate results in a
  An optimal routing protocol [109] that is energy-aware decrease in the average queuing delay per node, end-to-
and QoS-aware based on ACO for WMSNs allows to end delay, delay jitter and loss percentage. On the other
find the optimal routing path between sensor nodes and hand, an increase in the event generation rate leads to an
the sink. The proposed algorithm minimizes energy con- increase in the queuing delay per node, end-to-end delay,
sumption and prolongs the lifetime of the network while delay jitter and loss percentage. In addition, good quality
maximizing quality and reliability link. video requires a high frame encoding rate. Increasing this
  The cost function of the link depends on the costs of rate increases the percentage of loss. Thus to reduce this
energy consumption link, quality link and reliability link. percentage, a high transmission bit rate is required for
However, each metric cited can be attributed to an impor- high frame coding rates.
tance that varies depending on the multimedia applica- • QoS Enabled Probabilistic Routing for Heterogeneous
tion requirements. Wireless Sensor Networks
  Thus, the optimal path is one that has the minimum   Kumar et al. [110] presented a QoS distributed rout-
cost in terms of energy consumption, link reliability and ing algorithm based on ACO in Heterogeneous Wireless
quality in addition to defining the delay as a constraint. Sensor Networks (HWSNs). It takes into consideration
  The energy consumption cost is calculated in term of heterogeneous characteristics of nodes and allows better
the communication energy, transmission energy and a packet delivery. The protocol was designed to meet vari-
receiver node energy level. The quality link is the bit ous QoS requirements posed by different types of traffic
error rate on the link and the reliability link cost is the generated by heterogeneous nodes, thereby maximizing
percentage of time that the link is up and functioning network performance and usage. The network considers
properly. scalar and multimedia sensor nodes supporting scalar and
  The Fants move from a node to its neighbor according multimedia data traffic respectively.
to a transition probability that depends on the pheromone   Each node maintains two tables: a pheromone table
value deposited on the link and the heuristic value of the containing the neighbor information such as pheromone
link, which is the inverse of the cost of the link. At first, value required to reach a destination and an average hop
the initial pheromone value is the same for all links. The count and a neighbor table in which QoS metrics are
pheromone value is updated only on the links found by stored as available bandwidth and remaining energy of
all ants. The pheromone value evaporates at a certain rate the node. When selecting QoS aware route, the protocol
on all links in the paths constructed by all ants. uses three ants: Hello ants, Forward ants (FA) and back-
  Knowing that the simulation is run with different num- ward ants (BA). Hello ants are short messages broad-
bers of nodes. The parameters involved in the proposed cast by all the nodes at each interval and are used for
algorithm are number of ants, pheromone value impor- neighbor discovery, pheromone diffusion and link failure
tance, the heuristic value importance and the evaporation detection. To find a path from source to destination, the
rate. The purpose of the effected simulation is to see the source node generates control packet FA. A FA stores
impact of varying values of these parameters on the per- intermediate nodes encountered on its way and network
formance of the algorithm to determine the time taken status information for determining the QoS-aware routes.
to find the optimal path. The impact of the transmission On receiving the FA, every intermediate node updates
bit rate, the event generation rate and the frame coding the residual bandwidth and energy. At each intermediate
rate on various performance metrics are also taken into node, depending on whether the routing information for
consideration. destination is available or not, FA is either unicast or

13
186 International Journal of Wireless Information Networks (2021) 28:175–198

broadcast. In the first case, the next hop is selected based tocol. To enhance the scalability of the network, ICACR
on a probability calculation. uses the "divide and conquer" principle. The network
  Upon arrival at destination, each FA is converted into is divided into clusters and each can be considered as a
BA. The destination computes the pheromone released sub-networks. The additional sub-network is the back-
by BA. This pheromone value is determined by the route bone composed of the sink node and all the CHs. The
status information carried by the ant, hop count and cluster head can communicate with the sink node, while
delay. The BA is unicast, thus taking the same path in the the other nodes communicate with each other within the
opposite direction to the source. When the source node cluster.
receives BA, it begins to send data packets stochastically   IPACR protocol first searches for the local optimal
on different paths. The probability of selecting the next path set inside each cluster then searches for the local
hop is determined according to the heuristic function and optimal path set which is connected through each clus-
the pheromone value deposited on each node for the des- ter heads until the sink node. Finally, ICACR protocol
tination. For multimedia traffic, the protocol considers searches for the approximate global optimal path from
residual bandwidth as heuristic factor. For scalar traffic, paths found by IPACR.
residual energy is considered as the heuristic factor.   Both algorithms were simulated with NS2 and the
  The simulation results of the proposed protocol with results show that by increasing the network scale, the
NS2 show that in terms of packet delivery fraction, convergence rate of IPACR is better than the standard
end-to-end delay, jitter and percentage packets loss, the ant colony algorithm. Moreover, ICACR outperforms the
proposed solution reveals a significant improvement in IPACR and Dijsktra in all routing performance metrics
performance compared to AODV standard in HWSNs and in network lifetime, so it is more scalable. Nonethe-
with dynamic topology. However, in terms of normalized less, if the source node is located near to the sink, ICACR
routing load, the proposed protocol has higher routing is less efficient. For the transmission of the real video
overhead than AODV. data, the quality of the video in ICACR is good.
• A QOS-aware routing algorithm based on ant-cluster   The two algorithms are not applicable in the case of
in WMSNs real-time data transmission.
  Huang et al. [111] offer two multi-path routing algo- • ACO Based QoS Aware Routing for WSN with Het-
rithms based on ACO in WMSNs and propose a general- erogeneous Nodes (AntQHSeN)
ized QoS-aware routing model that integrate priorities of   Based on their previous work [110], Kumar et al.
packets and multiple routing metrics such as end-to-end [112] present a reactive ant-based QoS routing proto-
delay, packet loss rate, bandwidth and energy consump- col for HWSN called AntQHSeN working according
tion. The first, called an improved ant-based routing to two phases: route discovery phase and route mainte-
algorithm IPACR, a flat routing protocol that consists nance phase. Fants are created by the source node to find
in improving the standard ant colony routing algorithm multiple paths to the destination node. Each ant collects
(SACR) through optimizing the initial pheromone distri- information to assess the quality of the path by using
bution thus aiming to accelerate the convergence rate of some QoS parameters of the intermediate nodes lying
the algorithm. While SACR is based on the principle that on the path. The pheromone concentration is calculated
each link has the same initial pheromone value between using residual bandwidth, minimum residual energy and
nodes and their neighbors, IPACR establishes a neighbor- route cost. The route cost is computed according to hop
list, which stores information about adjacent nodes and count and end-to-end delay. In AntHQSeN, hello ants
guarantees at least one feasible path. The optimal path is allow the discovery of immediate neighbor nodes and
chosen according to the first hello packet, arriving at the carry the bandwidth, timestamp, energy, and pheromone
node. Log SEQ also allows to define the optimal path. concentration information. When a link failure occurs,
Hence, the initial pheromone value of each link depends route maintenance phase is activated.
on log SEQ. Each Fant has a table that contains all the   While in previous work, on receiving the Fant, each
visited nodes in order to prevent looping. The probabil- intermediate node updates the residual bandwidth, here
ity equation of selecting the next node is based on the the update of residual bandwidth is made depending upon
pheromone value and the heuristic function that depends the status of a flag bit and its residual bandwidth, thus
largely on the remaining energy. Local and global phero- either it forwards the ant or drops it. When the destination
mone values are updated by Fants and Bants. node receives Fant, it sends a Bant to the source node,
  The second, called ICACR is a QoS-aware, hierarchi- containing the status information of the route carried by
cal and ant based routing protocol that uses IPACR and Fant. To improve reliability, AntQHSeN used another
adapts it to the clustering topology to support large-scale method to calculate the pheromone value deposed on the
WMSNs. Clusters are formed according to the Leach pro- node.

13
International Journal of Wireless Information Networks (2021) 28:175–198 187

  Simulating AntQHSeN with NS-2 and comparing it maximize the network lifetime, apart from Enhanced
with EEABR and AODV protocols, the results showed AntNet protocol [98] and [99]. Also, with the excep-
that in terms of packet delivery fraction, end-to-end delay tion of the IEEBR [96] protocol, all studied protocols
and routing overhead, the solution outperforms its com- employ at least one or multiple metrics to meet QoS
petitors for dynamic topology environments. requirements. Moreover, most use multipath routing to
  However, AntQHSeN initially lacks sufficient informa- distribute the network traffic load between the different
tion to determine paths. The algorithm converges after a sensor nodes.
certain time, discovering better quality routes than those   Knowing that for a protocol intended for WMSN,
of AODV. AntQHSeN also provides consistent perfor- real time represents a primary constraint that most of
mance due to its reactive route discovery mechanism as routing protocols have not taken into account. The same
compared to EEABR. goes for the bandwidth, which is an essential measure
• Enhanced ant-based QoS-aware routing protocol for of QoS.
heterogeneous WSNs (EAQHSeN)   In summary, ACO is the most widely adopted SI
  EAQHSeN [113] is an improved QoS-aware routing approach by researchers because it provides QoS sup-
protocol for heterogeneous WSNs based on ACO sup- port to heterogeneous traffic load through multipath and
porting multimedia and scalar data traffic. This proposal multi-constraint routing.
meets the differentiated QoS requirements defined by
heterogeneous traffic generated by the nodes. The rout-
ing path is established according to each type of traffic 4.1.5 Artificial Bee Colony (ABC)
(control traffic, scalar traffic and multimedia traffic) and
its QoS constraints. This protocol is based on the work Artificial Bee Colony (ABC) algorithm [94, 114] is an opti-
done in [110] with some improvements. Before updating mization algorithm based on SI that simulates the intelli-
the bandwidth by the intermediate node and on receiv- gent foraging behavior of a honeybee swarm living inside
ing several ants of the same generation, it must compare a hive and possessing individual cognitive capacities and
the bandwidth field of each ant to that of the previously self-organization. Depending on the model, foraging bees
received ants of the current generation to decide to accept can be organized into employed and unemployed forager
or reject the ant packet using an acceptable factor. This bees. The employed forager bee is associated with a specific
improvement reduces routing overhead while keeping food source that it is currently exploiting. An unemployed
only the best paths. In addition, each Fant received at forager bee searches for a food source to exploit. It could be
the destination is not necessarily converted to Bant. Only an onlooker trying to find a food source using the informa-
Fants received within a stipulated time interval are con- tion providing by the employed bee or a scout who searches
verted to Bant to limit end-to-end delay and avoid extra the environment randomly for food. The position of a nour-
routing overhead while selecting a route. In addition, for ishment source depicts a possible solution to the multi-con-
multimedia traffic, EAQHSeN protocol considers both strained optimization issue, and its nectar amount is related
residual bandwidth and end-to-end delay as heuristic fac- to the quality of the associated solution. The employed for-
tors. ager bee exploits a nourishment source (solution) based on
  A series of simulations were carried out with NS2 to local information and the nectar amount (fitness cost). These
assess performance of the proposed EAQHSeN protocol bees will remember the new position and neglect the old one
and thus compare it to AODV and EEABR protocols. if the nectar amount associated with the new food source
In the two scenarios taking into account the variation (new solution) is better than the previous one. Once they
of pause time and maximum node speed respectively, have all completed the search process, they start sharing the
the results showed that significant optimizations were nectar information of the nourishing source like: direction,
observed compared to EEABR and AODV concerning distance, and profitability with the onlooker bees through
packet delivery fraction, end-to-end delay, routing over- a waggle dance. Therefore, an onlooker bee evaluates the
head and minimum remaining energy. nectar amount information provided by each employed bee
  EAQHSeN improves minimum residual energy by 4% and selects a food source with a higher probability to find the
compared to EEABR, which indicates an extended life- nectar. After few existing food sources have been abandoned
time. by the bees after many forages, scout bees begin to randomly
  After the comprehensive analysis of reviewed ACO- search for new food sources around the hive. Accordingly,
based routing protocols, it can be seen that most of the ABC algorithm achieves global optimization through inves-
routing protocols developed for WMSNs use ACO as tigation carried out by artificial scouts, whereas local opti-
a bio-inspired technique. We note that the majority of mization is obtained through exploitation, which is executed
works try to reduce energy consumption in order to by onlookers and employed bees [61].

13
188 International Journal of Wireless Information Networks (2021) 28:175–198

4.1.6 QoS‑Enabled & Energy Efficient Routing Protocol in network routing and the natural communication of spe-
cies.
• An Enhanced Artificial Bee Colony Based EELB-
PWDGR (EABC-EELB-PWDGR)
  Based on EELB-PWDGR protocol, which is itself 5 Performance Comparison
based on EE-PWDGR [115] created to overcome the
limitations of PWDGR [116], Al-Ariki et al. [117] first In this section, we establish two comparative analytical
proposed the ABC-EELB-PWDGR protocol based on tables that contain all these protocols and their main char-
the Artificial Bee Colony (ABC) to reduce the time acteristics. Tables 2 and 3 contain the 15 SI-based routing
consumed during the path selection process in order protocols for WMSN listed in the previous section.
to select the optimal path satisfying QoS constraints. Firstly, in Table 2, we choose various parameters to com-
However, the ABC algorithm presents unsatisfac- pare these surveyed protocols, according to their network
tory performance. Therefore, the use of an enhanced topology (flat or hierarchical), route selection, multipath,
ABC algorithm has enabled the development of An location awareness, congestion control, classification ser-
Enhanced Artificial Bee Colony Based EELB-PWDGR vice, energy efficiency, load balancing and QoS parameters
protocol called EABC-EELB-PWDGR for optimized considered.
route selection in WMSNs. Then, in Table 3, we give for each routing protocol cited
  EABC-EELB-PWDGR reduces both time consumed above, its advantages and disadvantages, but also an over-
during the classification of the routing paths based on view of the simulation approach that was used such as: the
QoS priority and energy consumption based on QoS- type of simulator used, the protocols with which they were
based load balancing between paths using two new compared, the performance parameters taken into account
improved solution search equations such as employed in the simulation and finally the results of the comparison.
bee and onlooker bee. In addition, to reduce the search
time and increase the convergence of ABC, the proto-
col uses the chaotic search method and the opposition- 6 Open Research Issue
based learning mechanism in population initialization.
Thus, an initial population is constructed in order to WMSN stills a constantly evolving area of research. It is
enhance the solution search. important to note that, in spite of the abundant research
  The proposed EABC-EELB-PWDGR protocol and undertaken in the field of routing protocols in recent years,
its predecessor, ABC-EELB-PWDGR were evaluated several open issues remain unresolved and require special
using NS2 and compared with PWDGR, EE-PWDGR, attention. In this section, these research issues are high-
and EELB-PWDGR in terms of end-to-end delay, Peak lighted for current and future explorations [71, 75, 80,
signal-to-noise ratio (PSNR), energy per-packet, hop 82–84].
count and network lifetime. The outcomes show that Energy efficiency: Energy efficiency plays an important
the proposed solution provides better delay with 20% role in the design of routing protocol for WMSNs. Thus,
less, reduced energy consumption by 60%, longer life- the main challenge of routing protocols is to reduce network
time by 17%, a PSNR higher by about 8db and a num- power consumption in order to achieve reliability in data
ber of hop reduced by 30%. delivery packets, extend network lifetime and ensure load
  In this category, an energy-efficient multipath rout- balancing in the network.
ing protocol has been developed guaranteeing the QoS Multi-constrained QoS Guarantee: In WMSNs, real-time
through several metrics. In principle, the bio-inspired multimedia data transmission requires some QoS parameters
ABC algorithm is designed to optimize multivariable to be taken into account. So, to ensure this, a routing proto-
and multimodal continuous functions. It is best suited col must support different application-specific QoS require-
for implementing multi-objective clustering in WMSN. ments such as end-to-end delay, delay jitter, bandwidth con-
The relevant drawbacks of these algorithms are the sumption, reliability, energy efficiency.
slow convergence and the local optimum [60]. Security: WMSNs are usually deployed in unattended or
  In summary, designing an efficient routing protocol hostile environments. Therefore, they are vulnerable to vari-
for the WMSN consists of taking into account all the ous attacks. Current routing protocols focus essentially on
criteria, or at least the most relevant for multimedia the QoS and energy efficiency and do not consider security.
needs. In this respect, henceforth, secure routing requires addi-
  To conclude, the main motivation behind the deploy- tional attention to protect against interceptions and malicious
ment of bio-inspired networking techniques stems from behavior. Ensuring network security while saving energy is
the analogy that exists between communication scenarios another area of ​​WMSN research [82].

13
 Table 2  Comparison of SI-based routing protocols for WMSNs
SI-routing protocols Architecture Route selection Multipath Location Conges- Clas- Energy QoS parameters Others Load balancing
aware- tion sification effi-
Flat Hierarchical ness control service ciency

ASAR ✓ Proactive ✓ ✓ ✓ Delay, bandwidth ACO ✓

2008 Packet loss ratio
M-IAR ✓ Proactive ✓ ✓ Delay and jitter ACO
2008
AGRA​ ✓ ✓ Delay ACO
2008 Bandwidth Game theory
LBHR ✓ Hybrid ✓ ✓ ✓ Bandwidth ACO ✓
2010 Delay The minimum
spanning tree
algorithm
AntSensNet ✓ Hybrid ✓ ✓ ✓ ✓ Delay ACO ✓
2010 Packet loss rate
Available memory
International Journal of Wireless Information Networks (2021) 28:175–198

ACOWMSN ✓ Reactive ✓ ✓ Delay, bandwidth, packet loss ACO ✓

2011 rate
[109] ✓ ✓ ✓ ✓ Link quality ACO
2011 Link reliability
IEEABR ✓ Proactive ✓ ✓ ✓ ACO
2011
Enhanced AntNet ✓ Reactive Delay, packet delivery ratio ACO
2013 (PDR)
[110] ✓ Reactive ✓ ✓ Bandwidth ACO ✓
2013
IPACR, ICACR​ ✓ ✓ ✓ Delay, bandwidth ACO
2014 Packet loss rate
AntQHSeN 2014 ✓ Reactive ✓ ✓ Bandwidth, end to end delay ACO ✓
[99] ✓ ✓ Bandwidth ACO
2016
EAQHSeN ✓ Reactive ✓ ✓ Bandwidth, end-to-end delay ACO ✓
2016
EABC-EELB-PWDGR ✓ ✓ ✓ Average delay, Link ABC ✓
2018 Quality, packet loss rate, path
reliability

13
189
 Table 3  Simulation scenario of SI-based routing protocols for WMSNs
190

Routing protocol Simulation Simulator type Performance metrics Compared to Advantages Disadvantages Comparison results

13
ASAR ✓ NS-2 Queuing delay, received Traditional ant-based, Fast convergence Performances degrada- ASAR has better conver-
2008 packet rate, dropped Dijkstra, Directional Better QoS for multi- tion due to bottleneck gence then Dijkstra and
packet rate diffusion routing algo- ple types of WMSN problem and repetitive traditional ant-based
rithm DD services usage of optimal paths algorithms
It outperforms Dijkstra
and DD
It select the optimal paths
to meet their individual
QoS requirements, thus
improving network
performance
M-IAR ✓ Java Delay, Jitter Not compared Self-organized, fault Limited scalability M-IAR shows good
2008 tolerant, adaptive No load balancing performance, which
achieves acceptable
delay, jitter and energy
consumption
AGRA 2008 Not simulated
LBHR ✓ NS-2 End-to-end delay, A novel clustered-control Prolonged network Complex calculations to LBHR reduces the end-
2010 network lifetime, algorithm CC lifetime find optimal route to-end delay, increases
Transmission success M-IAR Efficient load balancing Heavy load results in a the transmission success
rate, communication Congestion control bottleneck situation rate and achieves load
overhead support balancing
Thus, it has better adapt-
ability and scalability,
can in fact prolong the
network lifetime and
guarantee the QoS of
data transmission in
WMSNs
AntSensNet ✓ NS2 End to end delay, Packet Traditional ant-based Supports energy efficient High complexity AntSensNet outperforms
2010 delivery ratio, routing (T-ANT), AODV, video transmission AODV in terms of
overhead ASAR, TPGF Prevents network delivery ratio, end-to-
Congestion end delay and routing
overhead
The quality video is better
than ASAR and TPGF
It has better conver-
gence and provides
significantly better QoS
for multiple types of
services in WMSN
International Journal of Wireless Information Networks (2021) 28:175–198
 Table 3  (continued)
Routing protocol Simulation Simulator type Performance metrics Compared to Advantages Disadvantages Comparison results

ACOWMSN ✓ NS2 End-to-end delay, net- M-IAR Low delay Not suitable for large- ACOWMSN increases
2011 work lifetime, packet AODV Improve network lifetime scale networks due to network lifetime, ensur-
delivery ratio heavy traffic ing a lower delay and
Slow convergence high packet delivery
ratio when compared to
AODV and M-IAR
[109] ✓ Not specified Number of ants, phero- Not compared Prolonged network No load balancing The simulation studies
2011 mone value impor- lifetime Good quality video the effect of changing
tance, heuristic value Maximized quality and increases the percent- the number of ants,
importance, evapora- reliability link age of loss pheromone value impor-
tion rate tance, the heuristic value
importance, and the
evaporation rate on the
probability of finding the
optimal path
The effects of transmission
bit rate, event generation
rate and frame encoding
rate on various perfor-
mance metrics are also
International Journal of Wireless Information Networks (2021) 28:175–198

examined
IEEABR ✓ RMASE Latency, success rate, BABR, SC, FF, FP, Energy efficient QoS parameters not IEEABR has the lowest
2011 energy consumption, EEABR Congestion control considered end-to-end delay fol-
energy efficiency support lowed by EEABR
It outperforms all pro-
tocols in terms of low
energy consumption
It has a high success rate
in packets delivery and
surpasses all the routing
protocols in terms of
energy efficiency
[110] ✓ NS-2 Packet delivery fraction, AODV Better packet delivery High routing overhead The proposed algorithm
2013 average end to end Maximization of network achieves a higher
delay, Jitter, % Packets performance and use packet delivery ratio
lost, normalized routing than AODV protocol,
load especially when the traf-
fic load is high. It has a
lower jitter compared to
AODV for highly loaded
networks
It offers better end-to-end
delay and lesser packet
loss rate than AODV

13
191
 Table 3  (continued)
192

Routing protocol Simulation Simulator type Performance metrics Compared to Advantages Disadvantages Comparison results

13
Enhanced AntNet ✓ NS-2 Packet delivery ratio AODV Protocol adapted to a No comparison with The protocol has better
2013 (PDR), overhead, end- wireless and mobile WMSN routing pro- packet delivery ratio,
to-end delay environment tocols end-to-end delay and
Energy no considered overhead than AODV
IPACR, ✓ NS-2 Average delay, energy Standard ant colony rout- Scalable Real-time data transmis- In a large-scale network,
ICACR​ cost, packet loss rate, ing SACR, Dijkstra Good quality of video sion not applicable IPACR converges faster
2014 throughput, number of data transmission than SACR​
hops ICACR outperforms the
IPACR and Dijkstra in
all the considered per-
formance metrics and in
network lifetime, so it is
more scalable. It is less
efficient when the source
node is located near to
the sink,
For the transmission of
the real video data, the
quality of the video in
ICACR is good
AntQHSeN 2014 ✓ NS-2 Packet delivery fraction EEABR Work fine in both static Long setup time The PDF of AntQHSeN
(PDF) AODV and dynamic topolo- Lacks sufficient informa- is significantly higher
end-to-end delay (EED), gies tion at the beginning to compared with AODV
routing overhead Improve reliability find routes and EEABR
It has considerably lower
average EED than
AODV
It also provides consistent
performance due to its
reactive route discovery
mechanism as compared
to EEABR
International Journal of Wireless Information Networks (2021) 28:175–198
 Table 3  (continued)
Routing protocol Simulation Simulator type Performance metrics Compared to Advantages Disadvantages Comparison results

[99] ✓ NS-2 Received packets, packet AODV Guaranteed bandwidth Energy no considered The routing algorithm
2016 delivery ratio PDR, DSDV Transmit multimedia can deliver multimedia
throughput data with guaranteed data within an average
QoS throughput of 164.65
kbps
It provides better perfor-
mance in terms of PDR
and the throughputs than
the DSDV and AODV
Thus, it has an ability to
optimize routing tech-
niques for multimedia
applications over the
WMSNs
EAQHSeN ✓ NS-2 Packet delivery fraction EEABR Prolonged network Scalability cannot be EAQHSeN performs very
2016 end-to-end delay, routing AODV lifetime assured well irrespective of the
overhead, minimum Reduce routing and node’s pause time and
remaining energy control overheads outperforms AODV
and EEABR in terms of
packet delivery fraction,
International Journal of Wireless Information Networks (2021) 28:175–198

average end to-end delay

and routing overhead
It has better lifetimes than
EEABR
EABC-EELB-PWDGR ✓ NS2 End-to-end delay, PSNR, PWDGR, EE-PWDGR, Less delay Node failure event EABC-EELB-PWDGR
2018 energy per-packet, hop EELB-PWDGR, ABC- Energy efficient provides better routing
count, Network lifetime EELB-PWDGR Load balancing performance than ABC-
Improve ABC conver- EELB-PWDGR, which
gence in turn outperforms
EELB-PWDGR
EABC-EELB-PWDGR
offers 17% longer
lifetime, 20% less delay,
60% lesser energy con-
sumption, approximately
8 dB higher PSNR and
reduced number of hop
by 30%

13
193
194 International Journal of Wireless Information Networks (2021) 28:175–198

Multiple Sources and Base Stations: Most WMSN routing In a wireless communication system, routing still a big
protocols are designed to transmit data from a single source challenge problem. The SI techniques are applied to solve
to a unique base station. Some situations involve using routing problems in sensor networks due to their fast con-
multiple sources and sinks, for example, to simultaneously vergence, simplicity, high-quality and optimized solution.
obtain different information about events. Therefore, multi- Thus, these techniques have widened the scope in the area
sink and/or multi-source oriented routing may be considered of WSN. Studies have proved that the SI-based routing
for achieving higher performance in WMSNs. protocols perform better in terms of energy efficiency, reli-
Support for Node Mobility: Originally, in WMSNs, sen- ability, scalability, robustness and can enhance the QoS
sor nodes are immobile. Recent advances have enabled the performances of routing.
integration of mobile sensors to improve the performance, This paper has elucidated that research is very active
including coverage and energy efficiency [70]. In addition, in WSNs field, as opposed to WMSNs, where there is still
the mobility of sensors and sinks allows real-time delivery. much to discover. Therefore, a comprehensive analysis of
In WSNs, there are already works that support the mobility all existing SI-based routing protocols applied to WMSNs
of sinks such as [118, 119]. However, the majority of exist- has been provided. A classification of these protocols has
ing WMSN routing solutions in the literature do not con- been developed according to the existing SI algorithms,
sider mobility. Hence, routing that supports node mobility which include ACO and ABC. The basic concepts of these
is a relatively unexplored field and can be considered as an SI techniques have also been described. Then, a compari-
interesting area for future research and investigation. son of these routing protocols has been offered according
New Class of Algorithms: QoS is an important constraint to some primary metrics.
in the design of routing protocols in WMSNs. QoS rout- According to this review, we note that the majority of
ing is a multi-objective problem. It is therefore necessary proposed works have used the ACO approach in their rout-
to design and develop routing protocols that can increas- ing protocols and very few have used the bee colony algo-
ingly focus on several metrics. Routing approaches based rithm. The PSO-based technique was ignored. This is due
on computational intelligence as biologically inspired such to the fact that routing protocols based on ACO and ABC
as swarm intelligence (ACO, PSO, ABC, etc.), evolutionary are very efficient in various parameters such as energy,
algorithms (EA such as genetic algorithm), machine learning robustness, scalability and therefore perform best in com-
(ML), reinforcement learning (RL), fuzzy logic (FL), and plex transmission environments. Therefore, significant
artificial neural networks began to emerge and have shown efforts have been made to develop effective and efficient
promising results as they are adaptive and multi-objective. routing protocols for WMSNs. Thus, we conclude that SI
Consequently, these techniques remain very little exploited being a novel and bio-inspired field has contributed a lot
and require more investment in order to achieve further tech- to the improvement of routing issues of sensor networks.
nological advances in WMSN. For this first attempt, the goal of our contribution is to
Also, another concept that the future research should con- provide, on the one hand, a single document gathering all
sider and not overlook is hybrid intelligence algorithms. This existing SI-based routing protocols designed for WMSNs
involves using more than one algorithm to solve a problem. to encourage research in this direction and to allow direct
The interest of this concept is to be able to use the advan- access as an initial reading point for researchers. On the
tages of one solution to fill the gaps of another to solve or other hand, providing beginners or researchers with a ref-
improve some optimization problems in WMSNs. erence support that includes all previously published rout-
Cross-layer design: Very few protocols are designed ing surveys for WMSNs by avoiding time-consuming and
according to the cross-layer functionality. To improve QoS spending less effort.
routing in WMSNs, one of the solutions to be considered
is the use of a cross-layer approach for an efficient routing.

References
7 Conclusion
1. I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci,
A survey on sensor networks, IEEE Communications Maga-
Over the last decade, many research works about WMSNs zine, Vol. 40, No. 8, pp. 102–114, 2002.
have focused on improving the network performance as well 2. I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci,
as increasing the energy efficiency and communications Wireless sensor networks: a survey, Computer Networks, Vol.
effectiveness. SI is one of these new and promising tech- 38, No. 4, pp. 393–422, 2002.
3. R. K. Verma, S. Bharti, and K. K. Pattanaik, GDA: Gravita-
niques that enables a considerable increase in performance, tional data aggregation mechanism for periodic wireless sensor
scalability, robustness and efficiency by providing solutions networks, in 2018 IEEE Sensors, IEEE, pp. 1–4, 2018.
to complex problems.

13
International Journal of Wireless Information Networks (2021) 28:175–198 195

4. J. Yick, B. Mukherjee, and D. Ghosal, Wireless sensor network 25. S. Taghizadeh and S. Mohammadi, A survey on routing pro-
survey, Computer Networks, Vol. 52, No. 12, pp. 2292–2330, tocols of wireless sensor networks, International Journal of
2008. Advanced Research in Computer Science, Vol. 2, No. 5, pp.
5. I. F. Akyildiz, T. Melodia, and K. R. Chowdury, Wireless mul- 234–244, 2011.
timedia sensor networks: A survey, IEEE Wireless Communica- 26. D. Goyal and M. R. Tripathy: Routing protocols in wireless sen-
tions, Vol. 14, No. 6, pp. 32–39, 2007. sor networks: A survey, in: Second International Conference on
6. I. F. Akyildiz, T. Melodia, and K. R. Chowdury, A survey on Advanced Computing & Communication Technologies, IEEE,
wireless multimedia sensor networks, Computer Networks, Vol. pp. 474–480, 2012.
51, No. 4, pp. 921–960, 2007. 27. A. Anwar and D. Sridharan, A survey on routing protocols for
7. I. F. Akyildiz, T. Melodia, and K. R. Chowdury, Wireless multi- wireless sensor networks in various environments, International
media sensor networks: Applications and testbeds, Proceedings Journal of Computer Applications, Vol. 112, No. 5, pp. 13–29,
of the IEEE, Vol. 96, No. 10, pp. 1588–1605, 2008. 2015.
8. I. T. Almalkawi, M. Guerrero Zapata, J. N. Al-Karaki, and J. 28. H. Singh and D. Singh, Taxonomy of routing protocols in wire-
Morillo-Pozo, Wireless multimedia sensor networks: current less sensor networks: A survey, in International Conference on
trends and future directions, Sensors, Vol. 10, No. 7, pp. 6662– Contemporary Computing and Informatics (IC3I), IEEE, pp.
6717, 2010. 822–830, 2016.
9. Z. Ali and W. Shahzad, Critical analysis of swarm intelligence 29. A. Sarkar and T. S. Murugan, Routing protocols for wireless sen-
based routing protocols in adhoc and sensor wireless networks, sor networks: What the literature says?, Alexandria Engineering
in International Conference on Computer Networks and Informa- Journal, Vol. 55, No. 4, pp. 3173–3183, 2016.
tion Technology IEEE, pp. 287–292, 2011. 30. N. Shabbir and S. R. Hassan, Routing protocols for wireless sen-
10. E. Bonabeau, M. Dorigo, and G. Théraulaz, Swarm Intelligence: sor networks (WSNs), in: Wireless Sensor Networks: Insights and
From Natural to Artificial Systems, Oxford University Press, New Innovations. P. Sallis (ed.), pp. 21, IntechOpen, London, 2017.
York, 1999. 31. A. Kochhar, P. Kaur, P. Singh, and S. Sharma, Protocols for wire-
11. R. C. Eberhart, Y. Shi, and J. Kennedy, Swarm Intelligence, Else- less sensor networks: a survey, Journal of Telecommunications
vier, London, 2001. and Information Technology, 2018. https​://doi.org/10.26636​/
12. A. P. Engelbrecht, Computational Intelligence: An Introduction, jtit.2018.11741​7.
Wiley, Hoboken, 2007. 32. B. Bhushan and G. Sahoo, Routing protocols in wireless sensor
13. C. Blum and X. Li, Swarm intelligence in optimization, in Swarm networks, in Computational Intelligence in Sensor Networks,
Intelligence, Springer, Berlin, pp. 43–85, 2008. Springer, Berlin, pp. 215–248, 2019.
14. B. Velusamy and S. C. Pushpan, A review on swarm intelligence 33. A. Mehto, S. Tapaswi, and K. K. Pattanaik, A review on ren-
based routing approaches, International Journal of Engineering dezvous based data acquisition methods in wireless sensor net-
and Technology Innovation, Vol. 9, No. 3, pp. 182–195, 2019. works with mobile sink, Wireless Networks, Vol. 26, No. 4, pp.
15. M. Saleem, G. A. Di Caro, and M. Farooq, Swarm intelligence 2639–2663, 2020.
based routing protocol for wireless sensor networks: Survey and 34. M. R. Mundada, S. Kiran, S. Khobanna, R. N. Varsha, and S. A.
future directions, Information Sciences, Vol. 181, No. 20, pp. George, A study on energy efficient routing protocols in wireless
4597–4624, 2011. sensor networks, International Journal of Distributed and Paral-
16. R. Chiwariro, Quality of service aware routing protocols in wire- lel Systems, Vol. 3, No. 3, pp. 311, 2012.
less multimedia sensor networks: survey, International Journal 35. N. A. Pantazis, S. A. Nikolidakis, and D. D. Vergados, Energy-
of Information Technology, 2020. https​://doi.org/10.1007/s1265​ efficient routing protocols in wireless sensor networks: A survey,
2-020-02307​-1. IEEE Communications Surveys and Tutorials, Vol. 15, No. 2, pp.
17. M. Abazeed, N. Faisal, S. Zubair, and A. Ali, Routing protocols 551–591, 2012.
for wireless multimedia sensor network: A survey, Journal of 36. O. Ogundile and A. Alfa, A survey on an energy-efficient and
Sensors, 2013. https​://doi.org/10.1155/2013/46982​4. energy-balanced routing protocol for wireless sensor networks,
18. J. N. Al-Karaki and A. E. Kamal, Routing techniques in wireless Sensors, Vol. 17, No. 5, pp. 1084, 2017.
sensor networks: a survey, IEEE Wireless Communications, Vol. 37. L. K. Ketshabetswe, A. M. Zungeru, M. Mangwala, J. M. Chuma,
11, No. 6, pp. 6–28, 2004. and B. Sigweni, Communication protocols for wireless sensor
19. I. F. Akyildiz, M. C. Vuran, O. B. Akan, and W. Su, Wireless networks: A survey and comparison, Heliyon, Vol. 5, No. 5, pp.
sensor networks: a survey revisited, Computer Networks Journal, e01591, 2019.
Vol. 45, No. 3, pp. 393–422, 2004. 38. P. Maratha and K. Gupta, A comprehensive and systematized
20. K. Akkaya and M. Younis, A survey on routing protocols for review of energy-efficient routing protocols in wireless sensor
wireless sensor networks, Ad Hoc Networks, Vol. 3, No. 3, pp. networks, International Journal of Computers and Applications,
325–349, 2005. 2019. https​://doi.org/10.1080/12062​12X.2019.16975​13.
21. G. Acs and L. Buttyán, A Taxonomy of Routing Protocols for 39. C. Wei, J. Yang, Y. Gao, and Z. Zhang, Cluster-based routing
Wireless Sensor Networks, BUTE Telecommunication Depart- protocols in wireless sensor networks: a survey, in Proceedings
ment, pp. 32–40, 2007. of 2011 International Conference on Computer Science and Net-
22. A. Boukerche, M. Ahmad, B. Turgut, and D. Turgut, A taxonomy work Technology, IEEE, vol. 3, pp. 1659–1663, 2011.
of routing protocols in sensor networks, in: Algorithms and Pro- 40. X. Liu, A survey on clustering routing protocols in wireless sen-
tocols for Wireless Sensor Networks. A. Boukerche (ed.), pp. sor networks, Sensors, Vol. 12, No. 8, pp. 11113–11153, 2012.
129–160, Wiley, Hoboken, 2008. 41. N. Sabor, S. Sasaki, M. Abo-Zahhad, and S. M. Ahmed, A com-
23. L. García Villalba, A. Sandoval Orozco, A. Trivino Cabrera, prehensive survey on hierarchical-based routing protocols for
and C. Barenco Abbas, Routing protocols in wireless sensor mobile wireless sensor networks: review, taxonomy, and future
networks, Sensors, Vol. 9, No. 11, pp. 8399–8421, 2009. directions, Wireless Communications and Mobile Computing,
24. S. K. Singh, M. P. Singh, and D. K. Singh, Routing protocols in 2017. https​://doi.org/10.1155/2017/28185​42.
wireless sensor networks—A survey, International Journal of 42. C. Gherbi, Z. Aliouat, and M. Benmohammed, A survey on
Computer Science & Engineering Survey (IJCSES), Vol. 1, No. clustering routing protocols in wireless sensor networks, Sensor
2, pp. 63–83, 2010. Review, 2017. https​://doi.org/10.1108/SR-06-2016-0104.

13
196 International Journal of Wireless Information Networks (2021) 28:175–198

43. C. Narmadha, P. D. Marichamy, and D. A. Narayanan, A Survey 60. T. Kaur and D. Kumar, A survey on QoS mechanisms in WSN
on hierarchical-based routing protocols for wireless sensor net- for computational intelligence based routing protocols, Wireless
works, International Journal of Pure and Applied Mathematics, Networks, Vol. 26, No. 4, pp. 2465–2486, 2019.
Vol. 119, No. 16, pp. 3663–3676, 2018. 61. T. Kaur and D. Kumar, Computational intelligence-based energy
44. M. Radi, B. Dezfouli, K. A. Bakar, and M. Lee, Multipath rout- efficient routing protocols with QoS assurance for wireless sen-
ing in wireless sensor networks: Survey and research challenges, sor networks: A survey, International Journal of Wireless and
Sensors, Vol. 12, No. 1, pp. 650–685, 2012. Mobile Computing, Vol. 16, No. 2, pp. 172–193, 2019.
45. M. Masdari, M. Tanabi, and W. Azarbayjan, Multipath routing 62. M. Zungeru, L. M. Ang, and K. P. Seng, Classical and swarm
protocols in wireless sensor networks: A survey and analysis, intelligence based routing protocols for wireless sensor networks:
International Journal of Future Generation Communication and A survey and comparison, Journal of Network and Computer
Networking, Vol. 6, No. 6, pp. 181–192, 2013. Applications, Vol. 35, No. 5, pp. 1508–1536, 2012.
46. Y. Li, C. S. Chen, Y. Q. Song, and Z. Wang, Real-time QoS sup- 63. J. Yan, M. Zhou, and Z. Ding, Recent advances in energy-effi-
port in wireless sensor networks: A survey, IFAC Proceedings cient routing protocols for wireless sensor networks: a review,
Volumes, Vol. 40, No. 22, pp. 373–380, 2007. IEEE Access, Vol. 4, pp. 5673–5686, 2016.
47. A. D. Zhan, T. Y. Xu, H. G. Chen, B. L. Ye, and S. L. Lu, A 64. W. Rehan, S. Fischer, and M. Rehan, A critical review of surveys
survey on real-time routing protocols for wireless sensor net- emphasizing on routing in wireless sensor networks—an anato-
works, Chinese Journal of Computer Science, Vol. 3, No. 11, mization under general survey design framework, Sensors, Vol.
pp. 234–238, 2008. 17, No. 8, pp. 1713, 2017.
48. P. Chennakesavula, J. Ebenezer, and S. S. Murty, Real-time rout- 65. K. Guleria and A. K. Verma, Comprehensive review for energy
ing protocols for wireless sensor networks: A survey, in Proceed- efficient hierarchical routing protocols on wireless sensor net-
ings of the Fourth International Workshop on Wireless & Mobile works, Wireless Networks, Vol. 25, No. 3, pp. 1159–1183, 2018.
Networks (WIMo), Coimbatore, India, pp. 141–158, 2012. 66. S. Misra, M. Reisslein, and G. Sue, A survey of multimedia
49. F. Çelik, A. Zengin, and S. Tuncel, A survey on swarm intel- streaming in wireless sensor networks, IEEE Communications
ligence based routing protocols in wireless sensor networks, Surveys & Tutorials, Vol. 10, No. 4, pp. 18–39, 2008.
International Journal of Physical Sciences, Vol. 5, No. 14, pp. 67. A. Jayashree, G. S. Biradar, and V. Mytri, Review of multipath
2118–2126, 2010. routing protocols in wireless multimedia sensor network—a sur-
50. Z. Ali and W. Shahzad, Analysis of routing protocols in ad hoc vey, International Journal of Scientific & Engineering Research,
and sensor wireless networks based on swarm intelligence, Inter- Vol. 3, No. 7, pp. 1–9, 2012.
national Journal of Networks and Communications, Vol. 3, No. 68. R. R. Singh, R. K. Makhijani, and P. Singh, Overview of different
1, pp. 1–11, 2013. routing techniques for wireless multimedia sensor network, Inter-
51. A. M. Shamsan-Saleh, B. M. Ali, M. F. A. Rasid, and A. Ismail, national Journal of Innovations in Engineering and Technology,
A survey on energy awareness mechanisms in routing protocols Vol. 2, No. 1, pp. 164–169, 2013.
for wireless sensor networks using optimization methods, Trans- 69. M. Macit, V. C. Gungor, and G. Tuna, Comparison of QoS aware
actions on Emerging Telecommunications Technologies, Vol. 25, single-path vs. multi-path routing protocols for image transmis-
No. 12, pp. 1184–1207, 2013. sion in wireless multimedia sensor networks, Ad Hoc Networks,
52. W. Guo and W. Zhang, A survey on intelligent routing protocols Vol. 19, pp. 132–141, 2014.
in wireless sensor networks, Journal of Network and Computer 70. A. Alanazi and K. Elleithy, Real-time QoS routing protocols in
Applications, Vol. 38, No. 1, pp. 185–201, 2013. wireless multimedia sensor networks: Study and analysis, Sen-
53. A. Nayyar and R. Singh, A comprehensive review of ant colony sors, Vol. 15, No. 9, pp. 22209–22233, 2015.
optimization (ACO) based energy-efficient routing protocols for 71. M. Kumhar and V. Ukani, Survey on QoS aware routing pro-
wireless sensor networks, International Journal of Wireless Net- tocols for wireless multimedia sensor networks, International
works and Broadband Technologies (IJWNBT), Vol. 3, No. 3, pp. Journal of Computer Science & Communication, Vol. 6, No. 1,
33–55, 2014. pp. 121–128, 2015.
54. P. Krömer and P. Musilek, Bio-inspired routing strategies for 72. A. A. Anasane and R. A. Satao, A survey on various multipath
wireless sensor networks, in Propagation Phenomena in Real routing protocols in wireless sensor networks, Procedia Com-
World Networks, Springer, Cham, pp. 155–181, 2015. puter Science, Vol. 79, pp. 610–615, 2016.
55. S. Sendra, L. Parra, J. Lloret, and S. Khan, Systems and algo- 73. H. D. E. Al-Ariki and M. S. Swamy, A survey and analysis of
rithms for wireless sensor networks based on animal and natural multipath routing protocols in wireless multimedia sensor net-
behavior, International Journal of Distributed Sensor Networks, works, Wireless Networks, Vol. 23, No. 6, pp. 1823–1835, 2017.
Vol. 11, No. 3, pp. 625972, 2015. 74. Z. M. Hasan, H. Al-Rizzo, and F. Al-Turjman, A survey on mul-
56. T. Gui, C. Ma, F. Wang,, and D. E. Wilkins, Survey on swarm tipath routing protocols for QoS assurances in real-time wireless
intelligence based routing protocols for wireless sensor networks: multimedia sensor networks, IEEE Communications Surveys &
An extensive study, in 2016 IEEE International Conference on Tutorials, Vol. 19, No. 3, pp. 1424–1456, 2017.
Industrial Technology (ICIT), IEEE, pp. 1944–1949, 2016. 75. M. Doorwar and P. Malathi,s Comparative analysis of different
57. J. Kumar, S. Tripathi, and R. K. Tiwari, A survey on routing architecture based protocol in wireless multimedia sensor net-
protocols for wireless sensor networks using swarm intelligence, work, in 2017 International Conference on Emerging Trends &
International Journal of Internet Technology and Secured Trans- Innovation in ICT (ICEI), IEEE, pp. 92–97, 2017.
actions, Vol. 6, No. 2, pp. 79–102, 2016. 76. N. Kusuma and A. Subramanyam, A survey of multipath routing
58. A. Nayyar and R. Singh, Ant colony optimization (ACO) based protocols in wireless multimedia sensor networks, International
routing protocols for wireless sensor networks (WSN): a survey, Journal of Advance Research, Ideas and Innovations in Technol-
International Journal of Advanced Computer Science and Appli- ogy, Vol. 3, No. 3, 2017.
cations, Vol. 8, No. 2, pp. 148–155, 2017. 77. S. A. Jayadhas and S. E. Roslin, An analysis on routing proce-
59. X. Liu, Routing protocols based on ant colony optimization in dures in multimedia wireless sensor network, in 2018 Interna-
wireless sensor networks: A survey, IEEE Access, Vol. 5, pp. tional Conference on Communication, Computing and Internet
26303–26317, 2017. of Things (IC3IoT), IEEE, pp. 347–351, 2018.

13
International Journal of Wireless Information Networks (2021) 28:175–198 197

78. A. Chikh and M. Lehsaini, Multipath routing protocols for wire- 95. M. Dorigo and C. Blum, Ant colony optimization theory: A
less multimedia sensor networks: A survey, International Jour- survey, Theoretical Computer Science, Vol. 344, No. 2–3, pp.
nal of Communication Networks and Distributed Systems, Vol. 243–278, 2005.
20, No. 1, pp. 60–81, 2018. 96. A. M. Zungeru, L. M. Ang, S. R. S. Prabaharan, and K.P. Seng,
79. S. Li, S., J. G. Kim, D. Han, and K. S. Lee, A survey of energy- Ant based routing protocol for visual sensors, in International
efficient communication protocols with QoS guarantees in wire- Conference on Informatics Engineering and Information Sci-
less multimedia sensor networks, Sensors, Vol. 19, No. 1, p. 199, ence, Springer, Berlin, pp. 250–264, 2011.
2019. 97. T. C. Camilo, C. Carreto, J. S. Silva, and F. Boavida, An
80. S. Ehsan and B. Hamdaoui, A survey on energy-efficient routing energy-efficient ant based routing algorithm for wireless sen-
techniques with QoS assurances for wireless multimedia sensor sor networks, in Proceedings of 5th International Workshop
networks, IEEE Communications Surveys and Tutorials, Vol. 14, on Ant Colony Optimization and Swarm Intelligence, Springer,
No. 2, pp. 265–278, 2012. Berlin, pp. 49–59, 2006.
81. M. Abazeed, N. Faisal, A. Ali, and S. Zubair, A review of ant 98. I. Bennis, O. Zytoune, and D. Aboutajdine, Enhanced AntNet
colony optimization routing protocols for wireless multimedia protocol for wireless multimedia sensor networks, in Interna-
sensor network, Journal of Theoretical and Applied Information tional Conference on Networked Systems, Springer, Berlin, pp.
Technology, Vol. 70, No. 2, pp. 241–251, 2014. 316–320, 2013.
82. M. Abazeed, K. Saleem, A. Derhab, M. A. Orgun, N. Fisal, J. Al- 99. E. H. Putra, R. Hidayat, W. Widyawan, and I. W. Mustika, A
Muhtadi, and S. Zubair, A review of secure routing approaches routing optimization based on ant colony for wireless multime-
for current and next-generation wireless multimedia sensor net- dia sensor networks (WMSNs), International Journal of Intel-
works, International Journal of Distributed Sensor Networks, ligent Engineering and Systems, Vol. 9, No. 4, pp. 179–184,
Vol. 11, No. 10, pp. 524038, 2015. 2016.
83. S. Aswale and V. R. Ghorpade, Survey of QoS routing protocols 100. Y. Sun, H. Ma, L. Liu, and Y. E. Zheng, ASAR: An ant-based
in wireless multimedia sensor networks, Journal of Computer service-aware routing algorithm for multimedia sensor networks,
Networks and Communications, Vol. 2015, pp. 1–29, 2015. Frontiers of Electrical and Electronic Engineering in China, Vol.
84. V. Bhandary, A. Malik, and S. Kumar, Routing in wireless mul- 3, No. 1, pp. 25–33, 2008.
timedia sensor networks: A survey of existing protocols and 101. M. A. Rahman, R. GhasemAghaei, A. El Saddik, and W.
open research issues, Journal of Engineering, 2016. https​://doi. Gueaieb, M-IAR: biologically inspired routing protocol for wire-
org/10.1155/2016/96087​57. less multimedia sensor networks, in 2008 IEEE Instrumentation
85. H. Shen and G. Bai, Routing in wireless multimedia sensor net- and Measurement Technology Conference, IEEE, pp. 1823–1827,
works: A survey and challenges ahead, Journal of Network and 2008.
Computer Applications, Vol. 71, pp. 30–49, 2016. 102. GhasemAghaei, R., Rahman, M.A., Gueaieb, W., El Saddik, A.:
86. M. Deb and A. Choudhury, A review of wireless multimedia Ant colony-based reinforcement learning algorithm for routing
sensor network and existing routing protocols, International in wireless sensor networks. In: 2007 IEEE Instrumentation &
Journal of Computer Sciences and Engineering, Vol. 6, No. 3, Measurement Technology Conference IMTC 2007. IEEE, pp.
pp. 343–358, 2018. 1–6 (2007)
87. T. Back, Evolutionary Algorithms in Theory and Practice: Evolu- 103. Z. Ke, L. Li, Q. Sun, and N. Chen, Ant-like game routing algo-
tion Strategies, Evolutionary Programming, Genetic Algorithms, rithm for wireless multimedia sensor networks, in 2008 4th Inter-
Oxford University Press, Oxford, 1996. national Conference on Wireless Communications, Networking
88. S. Binitha and S. S. Sathya, A survey of bio inspired optimiza- and Mobile Computing, IEEE, pp. 1–4, 2008.
tion algorithms, International Journal of Soft Computing and 104. Z. Li and R. C. Wang, Load balancing-based hierarchical routing
Engineering, Vol. 2, No. 2, pp. 137–151, 2012. algorithm for wireless multimedia sensor networks, The Journal
89. K. Ke, L. Li, and N. Chen, A crossover game routing algorithm of China Universities of Posts and Telecommunications, Vol. 17,
for wireless multimedia sensor networks, in 2008 Ninth ACIS pp. 51–59, 2010.
International Conference on Software Engineering, Artificial 105. Bi, J., Li, Z., Wang, R.: An ant colony optimization-based load
Intelligence, Networking, and Parallel/Distributed Computing, balancing routing algorithm for wireless multimedia sensor net-
IEEE, pp. 595–599, 2008. works. In: 2010 IEEE 12th International Conference on Com-
90. K. Leibnitz, N. Wakamiya, and M. Murata, A bio-inspired robust munication Technology, IEEE, pp. 584–587 (2010)
routing protocol for mobile ad hoc networks, in 2007 16th Inter- 106. L. Cobo, A. Quintero, and S. Pierre, Ant-based routing for wire-
national Conference on Computer Communications and Net- less multimedia sensor networks using multiple qos metrics,
works, IEEE, pp. 321–326, 2007. Computer networks, Vol. 54, No. 17, pp. 2991–3010, 2010.
91. M. Paone, A. Cucinotta, A. L. Minnolo, A. Paladina, L. Puliafito, 107. S. Selvakennedy, S. Sinnappan, and Y. Shang, A biologically-
and A. Zaia, A bio-inspired distributed routing protocol for wire- inspired clustering protocol for wireless sensor networks, Com-
less sensor networks: Performance evaluation, in 2010 IEEE 30th puter Communications, Vol. 30, No. 14–15, pp. 2786–2801,
International Conference on Distributed Computing Systems 2007.
Workshops, IEEE, pp. 247–255, 2010. 108. X. Yu, J. Luo, and J. Huang, An ant colony optimization based
92. S. Das, A. Biswas, S. Dasgupta, and A. Abraham, Bacterial for- QoS routing algorithm for wireless multimedia sensor networks,
aging optimization algorithm: theoretical foundations, analysis, in Proceedings of the IEEE 3rd International Conference on
and applications, in Foundations of Computational Intelligence, Communication Software and Networks (ICCSN), pp. 37–41,
vol. 3. Springer, Berlin, pp. 23–55, 2009. 2011.
93. M. Dorigo, M. Birattari, and T. Sẗutzle, Ant colony optimization, 109. H. Al-Zurba, T. Landolsi, M. Hassan, and F. Abdelaziz, on the
IEEE Computational Intelligence Magazine, Vol. 1, No. 4, pp. suitability of using ant colony optimization for routing multime-
28–39, 2006. dia content over wireless sensor networks, International Journal
94. D. Karaboga and B. Basturk, A powerful and efficient algorithm On Applications Of Graph Theory In Wireless Ad Hoc Networks
for numerical function optimization: Artificial bee colony (ABC) And Sensor Networks, Vol. 3, No. 2, pp. 15–35, 2011.
algorithm, Journal of Global Optimization, Vol. 39, No. 3, pp. 110. S. Kumar, M. Dave, and S. Dahiya, QoS enabled probabilistic
459–471, 2007. routing for heterogeneous wireless sensor networks, International

13
198 International Journal of Wireless Information Networks (2021) 28:175–198

Journal of Computer Network and Information Security, Vol. 5, Publisher’s Note Springer Nature remains neutral with regard to
No. 4, pp. 31–39, 2013. jurisdictional claims in published maps and institutional affiliations.
111. H. Huang, X. Cao, R. Wang, and Y. Wen, A qos aware routing
algorithm based on ant-cluster in wireless multimedia sensor
networks, Science China Information Sciences, Vol. 57, No. 10,
pp. 1–16, 2014. Fa z i l e t Le mya B e n m a n ‑
112. S. Kumar, M. Dave, and S. Dahiya, ACO based QoS aware rout- sour received her Magister
ing for wireless sensor networks with heterogeneous nodes, in degree in Computer Science from
Emerging Trends in Computing and Communication, Springer, the University of Tlemcen, Algeria
New Delhi, pp. 157–168, 2014. in 2010. She is an assistant lec-
113. S. K. Malik, M. Dave, S. K. Dhurandher, I. Woungang, and L. turer in the Department of Com-
Barolli, An ant-based QoS-aware routing protocol for heteroge- puter Science at the University of
neous wireless sensor networks, Soft computing, Vol. 21, No. 21, Tlemcen since 2010. She is a
pp. 6225–6236, 2016. Ph.D. student. Currently, she is
114. D. Karaboga and B. Akay, A comparative study of artificial bee a member of the research Labora-
colony algorithm, Applied Mathematics and Computation, Vol. tory LRIT, Department of Com-
214, No. 1, pp. 108–132, 2009. puter Science, Tlemcen Univer-
115. M. S. Swamy, Energy-efficient geographical multi-path routing sity. Her research interests include
protocol with adaptive load balancing for wireless multimedia Wireless Multimedia Sensor Net-
sensor networks, Journal of Adv Research in Dynamical & Con- works and Swarm Intelligence.
trol Systems, Vol. 9, No. 9, pp. 148–166, 2017.
116. J. Wang, Y. Zhang, J. Wang, Y. Ma, and M. Chen, PWDGR: pair-
wise directional geographical routing based on wireless sensor
Nabila Labraoui is an associate
network, IEEE Internet of Things Journal, Vol. 2, No. 1, pp.
professor in computer engineering
14–22, 2014.
at the University of Tlemcen,
117. H. D. E. Al-Ariki, M. A. Alareqi, and S. Swamy, An enhanced
Algeria. She received her Dr. Eng.
artificial bee colony based EELB-PWDGR for optimized route
Degree in computer science from
selection in wireless multimedia sensor networks, Pertanika
ESI (Ecole Nationale Supérieure
Journal of Science & Technology, Vol. 26, No. 4, 2018.
d’ Informatique), Algiers and her
118. S. Jain, S. Sharma, and N. Bagga, A vertical and horizontal segre-
Ph.D. Degree from the University
gation based data dissemination protocol, in Emerging Research
of Tlemcen. Her current research
in Computing, Information, Communication and Applications,
interests include Wireless Ad hoc
Springer, New Delhi, pp. 401-412, 2016.
Sensor Networks, Vehicular Net-
119. S. Jain, K. K. Pattanaik, R. K. Verma, and A. Shukla, QRRP: A
works, Drones, E-health, Network
query-driven ring routing protocol for mobile sink based wire-
Security, Cognitive Radio, Cloud
less sensor networks, in TENCON 2019–2019 IEEE Region 10
and Fog computing.
Conference (TENCON), IEEE, pp. 1986–1991, 2019.

13