Wireless Personal Communications

https://doi.org/10.1007/s11277-021-08745-0

A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo

Search Clustering Based Multipath Routing in Wireless
Sensor Networks

C. Mohanadevi1 · S. Selvakumar2

Accepted: 21 July 2021

© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2021

Abstract
Wireless sensor networks applications have been used in many areas that are difficult to
access by humans. However, Routing remains the key challenges in sensor networks as it
is important for the timely delivery of sensed datato the base station. In recent years, Mul-
tipath Routing has been used to ensure reliable and scalable data transmission in WSN.
Although many multipath routing algorithms have been proposed, very few protocols have
been focused on the Quality of Service (QoS) based routing. This paper proposes a QoS-
aware, multipath routing protocol in which sensor nodes are clustered using the hybrid Par-
ticle Swarm Optimization-Cuckoo Search Optimization algorithm. The proposed protocol
then chooses multiple stable paths (optimized network routing) using the Cluster Heads to
transmit data based on multi-hop communication. Unlike the existing protocols, it relies on
paths that do not affect QoS for rapid data transmission. It also extends the network life-
time by changing the Cluster Heads periodically based on the residual energy and uses the
optimal number of paths to data transmission unlike the existing QoS Centric protocols.
The performance of the proposed protocol is evaluated using NS-2 Simulator in different
scenarios. The proposed protocol outperforms current protocols in terms of QoS param-
eters such as throughput, packet delivery ratio, end-to-end delay, and network lifetime,
according to simulation results.

Keywords Wireless sensor network · Multipath routing · Quality of Service · Particle

Swarm Optimization · Cuckoo Search Optimization

* C. Mohanadevi
[email protected]
S. Selvakumar
[email protected]
1
Department of Electronics and Communication Engineering, Mailam Engineering College,
Villupuram, India
2
Department of Computer Science and Engineering, GKM College of Engineering & Technology,
Perungalathur, Chennai, India

13
Vol.:(0123456789)
 C. Mohanadevi, S. Selvakumar

1 Introduction

WSNs have been widely used in a variety of applications, including environmental control,
disaster management, and surveillance [3, 27]. They are most often used in holistic areas
where humans have difficulty in accessing. A WSN is a collection of a large number of tiny
sensor nodes that are connected wirelessly for seamless communication [1]. These sensors
have limited processing power, energy, memory and data transmission range. Moreover,
these sensor nodes are normally transmitting data using wireless radio communication [2].
Routing is one of the key issues in the WSN because of the dynamic topological changes
of the network and limited energy availability [4, 5]. It is critical when WSN is used in
delay-sensitive applications such as remote patient monitoring, gas leakage detection and
emergency response systems. To support such applications, it is utmost important to find
the optimal route that helps to meet all Quality of Service (QoS) metrics [6]. In WSN, sen-
sor nodes that are located close to the sink node is involved in many data transmission and
transfers huge amounts of data. Subsequently, the energy levels of such nodes are drained
rapidly. Hence, the network will be disconnected, which reduces the lifetime of the net-
work. This problem is known as hot spot or energy hole problem as shown in the Fig. 1.
Many routing techniques have been proposed to address the hot spot problem in the
WSN. These protocols have efficiently used the energy of nodes in the sink coverage to
improve the network lifetime. Clustering is one of the efficient techniques to resolve the hot
spot problem which extends the network lifetime by organizing sensor nodes into hierarchi-
cal topology. In the clustering based technique, the sensor nodes are divided into smaller
groups called clusters. Each cluster has a special node called Cluster Head (CH) which is
responsible for collecting data from other nodes in the cluster. CH is also responsible for
data aggression and redundant data elimination. Even though CHs are used to minimize
the energy depletion in the network, CHs are quickly drained due to their additional func-
tionalities. As CHs plays an important role to determine the network lifetime, it is essential
to select the best sensor nodes as CHs. Hence, CH selection is considered as an NP-Hard
Problem. In recent years, many swarm intelligent algorithms that are designed based on the
exploration and exploitation behavior of different birds and animals are effectively used to
solve optimization problems including hot spot problems in WSN [22–26]. Furthermore,
Multipath routing protocols have been recently used to improve the reliability of data trans-
mission in WSN [7–9, 27] In this kind of routing, sensor nodes send the data over multi-
ple paths to the sink node simultaneously. If one path fails due to the node failure or link

Fig. 1  The hot spot problem

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

failures, alternative paths are used to continue data transmission. Multipath routing is used
to establish fault tolerant and reliable paths in the WSN. Multihop routing is a significant
assistance required for WSN. MANET routing strategies neglect to perform well in WSN.
Web steering expects to be dependable wired associations and thus there are inconsistent
packet blunders, yet in WSN this isn’t the situation. Additionally, MANET directing con-
ventions accept symmetric connections between two associations which isn’t the situation
with WSN as the greater part of the traffic is unidirectional way.
This paper proposes a navel clustering based multipath routing protocol which
uses hybrid Particle Swarm Optimization – Cuckoo Search Optimization algorithm to form
clusters in WSN. It chooses the best Cluster Heads based on the residual energy and dis-
tance with the neighboring nodes to determine the optimal paths to the sink node. Perhaps,
it ensures that the desired QoS metric values and energy efficiency of application never
violated. It enables the source node to transmit data on multiple paths simultaneously to
reduce data transmission time. Hence, it significantly conserves the residual energy of sen-
sors and reduces the network traffic when compared to the existing protocols. Simulation
results reveal that the proposed protocol improves various quality metrics such as through-
put, packet loss, latency and network lifetime.
The rest of the paper is structured as follows. Section 2 discusses the principles of vari-
ous multipath routing protocols and Sect. 3 presents the assumptions behind the System
model. Section 4 describes the various phases of the proposed hybrid Particle Swarm Opti-
mization-Cuckoo Search Optimization algorithm based multipath routing protocol. Sec-
tion 5 accesses the performance of the proposed protocol and compares the performance
with the existing protocols. Finally, Sect. 6 concludes the paper.

2 Related Work

Many routing protocols have been designed for WSNs depending on the application, archi-
tecture and objectives [1, 6, 7]. Multipath routing protocols have received more attention
due to its ability to support scalability, fault tolerance and energy conservation. This sec-
tion discusses the working principle of the existing multipath routing protocols.
Deepak et al. has designed a braided multipath algorithm which uses sensor nodes’
information to establish two types of paths [10]. This protocol recovers from failures with
less energy when compared to the earlier protocols. At the same time, it does not improve
the network lifetime significantly. Swades De et al. have presented a meshed multipath
routing algorithm for WSN [11]. Unlike the disjoint multipath protocol in [10], it improves
the network throughput and utilizes network resources including channel bandwidth and
battery power more efficiently by selective forwarding of packets and end-to-end Forward
Error Correction (FEC) coding technique. However, this method needs sensor nodes with
Global Positioning System(GPS).
Younis and Fahmy designed a clustering protocol called Hybrid Energy Efficient Distrib-
uted (HEED) method for multipath routing [12]. This protocol has organized sensor nodes
into clusters. It then selects Cluster Head (CH) for each cluster based on the node degree and
residual energy periodically. Unlike the earlier approach in [11], it did not need location-aware
nodes. This protocol improves network lifetime, scalability and load balancing. In [13], Mul-
tipath Routing Protocol (MRBCH) has used credible cluster heads to ensure secure commu-
nication over multipath. In this method, CHs are selected based on the node’s creditability
and residual energy. Perhaps, the credit value which is calculated using trust mechanisms are

13
 C. Mohanadevi, S. Selvakumar

exchanged among the CHs. The CHs then formed the route to the sink node. Most of the ear-
lier multipath protocols have tried to optimize the energy consumption for data transmission.
However, many QoS based Routing protocols have been proposed to improve the reliability of
critical applications while optimizing the energy consumption in the recent years.
Sohrabi et at have presented a Sequential Assignment Routing(SAR) algorithm to deter-
mine multiple paths for data transmission [14]. It organizes the nodes in tree form and selects
paths to the sink based on QoS metrics, residual energy and priority level of data packets.
Although SAR is energy efficient, it has to rediscover routes whenever a sensor node fails.
He et al. proposed a location-aware, QoS centric protocol called SPEED [15]. This protocol
ensures end-to-end delivery guarantee with desired level of QoS parameters. Felmban et al.
have proposed a Multipath and Multispeed Routing Algorithm (MMSPEED) for WSN[16].
Unlike SPEED, this protocol minimizes the packet loss and network congestion by compro-
mising QoS guarantee. Although this protocol is scalable, it does not focus on the energy
consumption. The Multi-Constrained QoS Multipath Routing Protocol(MCMP) used braided
multipath to transmit data over the optimal paths. It has used linear integer programming to
limit the end-to-end delay in data transmission. Though it reduces the energy consumption, it
does not guarantee for optimal energy consumption in all circumstances. Bagheri et al. have
presented a clustering based multipath routing which uses CHs for data transmission [17].
It defines stable paths and selects cluster heads for each cluster based on the residual energy
of sensor nodes in the cluster. Due to control packet overhead, it is less energy efficient and
allows nodes to be fitted with a GPS system.
Mazaheri et al. suggested multipath hierarchical routing, in which sensors within a given
range are clustered, and the Cluster Head is chosen from among the clusters with the highest
residual energy[18].The path to the sink node is then established by within the range deter-
mined depends on the distance between Signal to Noise (SNR) ratio and buffer size. It consid-
erably improves the reliability of paths. However, it fails to improve the latency and network
lifetime. Almalkavi et al. developed another cluster based multipath protocol in which the sink
initiates the cluster formation process[19]. In this algorithm, the sink node broadcasts the con-
trol packets to all sensor nodes and it selects the cluster heads based on the signal strength of
node’s reply. It then arranges the cluster heads into different levels and initiates data transmis-
sion from CHs in the upper level. But, it does not ensure reliability due to the existence of
improper paths. SVM cluster based secure and effective routing protocol is studied in [28].
Alwan et al. developed the multipath QoSRouting (MQoSR) protocol. In this protocol, the
paths are found using heuristics for neighbor selection based on geographic routing mecha-
nism. This protocol can be used to meet the QoS requirements of various WSN applications.
Deepa and Suguna have used Particle Swarm Optimization(PSO) for establishing cluster
based multipath routing protocol. It uses alternate path along with primary path to transmit
data. Shahbaz et al. have proposed a multipath routing algorithm which uses firefly algorithm
to cluster the sensor nodes. The protocol has used fuzzy rules to establish the paths between
the sources and sink node. Although it discovers two paths, it uses only one path for data
transmission [9].

3 Network Model

The following assumptions about the WSN are made in the proposed work. The WSN is mod-
eled as a weighted, undirected graph G (V,E,W). The graph model contains vertex, Edges
and weight function. Here the V represents vertex set, E denotes Edge set and W represents

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

the weight value of communication sink. Each vertex in the Vertex Set V represents a sen-
sor node whereas an edge in the Edge Set E represents the bidirectional communication link
between the sensor nodes. A weight value is associated with each communication link.
Moreover,sensor nodes are heterogeneous and have the same capabilities in terms processing
power, storage, communication range and battery power. It also assumes that each sensor has
a unique ID in the network.However, sensor nodes can change their transmitting range. Hence,
Cluster Heads (CHs) can send data to the sink through other cluster heads and all other nodes
can send data to the CH through single hop manner as shown in the Fig. 2.
With sensor node, cluster head, inter and intra cluster communication route is established
in this Figure. source and sink is identified and route establishment is done properly.

3.1 Energy Model

Energy is an important in WSN as it decides the usability of sensor network. The energy
depletion for transferring k bits of data over a distance d by a node i is computed as in the fol-
lowing Eq. 1.
{
k.Eelec + k𝜀fs + d2 if d < do
ET (k, d) =
k.Eelec + k𝜀mp + d4 if d ≥ do (1)

where Eelec is the energy consumed by the transmitter or receiver circuitry in the sensor
nodes, 𝜀fs is the energy consumed by the transmitter amplifier and 𝜀mp represents the energy
consumed by the transmitter amplifier which depends on the transmission distance in a
multipath model. The threshold transmission distance do is computed as in Eq. 2.

Fig. 2  System model

13
 C. Mohanadevi, S. Selvakumar

√
do = 𝜀fs ∕𝜀mp (2)

Similarly, the energy required to receive k bits of data is defined as in Eq. 3.

ER (k) = k ⋅ Eelec (3)
The residual energy is computed by subtracting the energy transmitted or received from
the current energy in the sensor node.

4 The Proposed Method

The proposed routing protocol can find multiple reliable paths that support energy-efficient
and satisfying the QoS requirements of the applications. It comprises the following three
stages.

1. Cluster Formation Phase WSN is divided into many smaller clusters using the hybridi-
zation of Particle Swarm Optimization and Cuckoo Search Optimization algorithm.
2. Route Discovery Phase Reliable multiple paths are established and data is transmitted
over multiple paths simultaneously.
3. Route Maintenance Rerouting or Re-clustering in case of node failures.

In the proposed method, clusters are formed using the hybrid Particle-Swarm
Optimization-Cuckoo Search Optimization Algorithm in the first stage while energy
efficient,optimal paths that meet the QoS requirements of application are established in the
second phase[25]. The routes are maintained and reestablished in the last phase. The fol-
lowing subsection elaborates the processes of each phase.

4.1 Cluster Formation Phase

The sink (base station) node initiates the cluster formation process by sending Seek_Info_
Msg to all nodes in its coverage area. Upon receiving the message, each node sends a reply
message Source_Info_Msg containing its ID, position in two dimensional space, veloc-
ity and residual energy at that moment to the sink node. Hence, all information’s about
the neighboring nodes of sink node are maintained and updated in the sink node. Then,
it invokes the HPSOCS algorithm for clustering the nodes. The main aspect of the hybrid
PSO-CSO is that it uses the Lévy flight method to find the optimal solution in the search
space [22–24, 26] as shown in Fig. 3.
Velocity of Optimization calculation is one of its significant feature, as it is the instru-
ment used to move (advance) the situation of a molecule to look for ideal arrangements.
This velocity guideline intends to accomplish a harmony among investigation and misuse.
It is more efficient than the random walk method which is used in the PSO. The key steps
of the HPSOCS algorithm are as follows.
Initially, a swarm with n particles that represent the sensor nodes in the sink coverage
area are selected. Then the fitness value which determines the best position of the particle
is calculated as in the equation.

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

Fig. 3  Clustering using Hybrid Particle Swarm Optimization and Cuckoo Search Algorithm

Step 1:In the HPSOCS search space, each particle which represents the sensor node in
the sink’s coverage area has two dimensions, namely: particle position and velocity.
Step 2: Then the fitness ∑ value( of each( particle using the following )) equa-
n
fitness(current_ particle) = 𝛼1 distance current_particle, membernodei ∕n
i=1
∑n
on.it
( )
+ 𝛼2 (Energy membernodei ∕Energy(current_particle)
i=1
+ (1 − 𝛼1 − 𝛼2 )1∕n
where,α1and α2 are weighing parameters with values ranging from 0 to 1, and n
denotes the number of sensor nodes in the cluster led by the sink node. The fitness

13
 C. Mohanadevi, S. Selvakumar

function is used to maximise the average distance to ensure that the best sensor
node’s energy is higher than the average energy of all nodes.The particle’s velocity is
initialized randomly. At the end of each generation, the algorithm records the best
position in the current generation pbest and best position of the entire swarm, gbest
and current best position lbest of a particle.
Step 3.The particles of next iteration are derived from the initial generation as fol-
lows.
Step 3.1. The velocity of the particle is updated to move towards with the best parti-
cle.It is calculated as follows:

velocitynew = 𝜔 ∗ velocityold + r1 c1 (pbest − lbest) + r2 c2 (gbest − lbest)

where ω is inertia weight and the random variables ­r1 and ­r2 are uniformly distributed
within range [0, 1]. The acceleration coefficients ­c1 and ­c2are positive constants.
Step 3.2. The new position of the particle is computed using the previous position of
the particle and new velocity value.
positionnew = positionold + velocitynew

Step 4.Once the particles positions are updated, the fitness values of new particles are
computed as in Step 2.
Step 5.After that, the fitness values of the old and new particles are compared, and the
better one is advanced to the next iteration.
Step 6.In each iteration, one best solution is selected as albest solution.The particle
which has maximum fitness value up to the current iteration is selected as gbest solu-
tion.
// Cuckoo Search Optimization with Levy.
Step 7: Population with gbset of PSO is taken as initial population of n nests Xi where
i = 1 to n.
Step 8: Fitness of each cuckoo is calculated.
Step 9: While (iteration t < Maximum Generation) or (stop criterion).

Step 10.Once the algorithm terminates, the best nest with the maximum fittest cuckoo
egg is considered as the optimal solution and Cluster Heads (CHs) are generated from
the optimal solution.

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

Lévy Flight, which depends on the flight conduct of creatures and bugs, has been dif-
ferently concentrated in literary works. It is characterized as a sort of non-Gauss stochas-
tic cycle with an arbitrary walk whose progression sizes depend on Lévy stable disper-
sion. There are bunches of renditions of Lévy dispersion. Mantegna Algorithm has been
tried to be the most proficient technique to produce Lévy circulation esteems. It has been
received in numerous development calculations to escape from neighborhood least. The
CHs are then broadcast thejoin_Cluster message containing their location to other sen-
sor nodes in its communication range. When a sensor node receives the message, it sends
member_Cluster to the CH. Upon receiving reply messages to all its request message, CH
includes the nodes in its cluster and sends ack_CH message to the node. When the sensor
node receives multiple join_Cluster messages, it sends member_Cluster message to the CH
which is nearest in terms of Euclidean distance and sends reject_Cluster messages to other
CHs.

4.2 Route discovery

In this step, the routes between the CHs and the path between the CHs and the sink node
are established. At first, each CH calculates the score using three factors, namely Residual
Energy, Distance and traffic support. The traffic support of CH is calculated as the ratio
between thenumber of packets in the buffer to the total capacity of the CH buffer. After
that, CH broadcasts a message containing its score to the nearby CHs which is subse-
quently stored in the Nbr-CH table in the CH. This process is repeated until all the CHs
have complete information about all neighboring CHs.When a CH receives a request
to transfer data to the base station, it sends its application specific requirements of QoS
parameters such as Reliability, Delay and Energy to the CH through route_ Request mes-
sage. Upon receiving the message, CH looks for appropriate CHs in the Nbr-CH table and
forwards the request to the CHs that can fulfill the requirements. This process continues
until the CH establishes the multiple paths that satisfy the QoS requirements of the sensor
nodes to the sink node as shown in Fig. 4. When there is a path failure and CH replacement
due to low residual energy, the route discovery process is initiated again. When the base
station receives the route_ Request message, it sends the route_ Reply message through the
established path. Each CH waits for the route_ Request message from their neighboring
CH for a specific time ­Twait. It uses all the paths established within the time frame to trans-
mit the data simultaneously.
Once the paths are established, all paths are used by the source node to deliver packets
about events to the sink nodes. The data transmission algorithm is shown below:

13
 C. Mohanadevi, S. Selvakumar

procedure Data_Transmission
begin
while ( there is a data about a event)
begin
if( number of path == 1) then
Send the date through the single path
else

Send the data over the multiple paths in round robin manner
Records the end-to-end energy consumption of the path
Sort the paths based on end-to-end energy consumption
end

end

The proposed protocol uses a path for a while and selects another path for the next time
period when there is more than one path to the sink. One of the notable improvements
in the proposed protocol is that it utilizes all paths established to deliver packets. This is
because of utilizing the energy- efficient paths for some time longer than the other paths.
Since it changes the paths and uses them orderly, it ensures the reliability of the paths and
network lifetime.

4.3 Route Maintenance

The sink node of the network initiates the rerouting process when the QoS parameter
such as throughput drops below the threshold required by the application. Moreover,
the cluster heads maintain the residual energy of all sensor nodes within the cluster
and the sink node maintains the residual energy of all cluster heads. Initially, the sink

Fig. 4  Route establishment process

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

node initiates the re-clustering process when the residual energy of the cluster head falls
below the average residual energy of CHs which is computed initially. In this manner,
the proposed protocol will increase the network lifetime and utilize the energy in the
sensor nodes more efficiently.

5 Performance Evaluation

The proposed method is implemented using NS2.35 simulator and its performance is
compared with the existing protocols based on EE-LEACH, EPSO-CEO, OQoS-CMRP.
The simulation parameters are shown in Table.1. The sensor nodes are placed randomly
on the 300 m × 200 m and the number of nodes is varied from 50 to 250 during the
simulation. IEEE 802.15.3 is used to define the physical and MAC sub layer for data
packet transmission. The size of data packets is 128 bites and the initial energy of each
sensor node is set at 3 J. The rectangular field is used to represent the network coverage
area and the network has supported bandwidth of 250 kbps.

Table 1  Simulation parameters Radio propagation model Two-ray ground reflection model

MAC Type IEEE 802.15.4

Antenna Type Omnidirectional
Simulation area 300×200 m
Link bandwidth/data rate 250 kbps
Radio frequency 2.4 GHz
Number of nodes 50, 100, 150, 200, 250
Radius (r) 40 m
Simulation time 500 s
Channel type Channel/wireless
Queue type Priority queue
Data packet size 128 bytes
Control packet size 32 bytes
Initial energy 3J
Transmission power consumption 0.002*dist J
Reception power consumption 0.02 J
Eelec 50 nJ/bit
𝜀fs 1011 J/bit/m2
𝜀mp 1.3 1015 J/bit/m4
do 87 m
Acceleration coefficient ­C1,C2 2
Maximum iteration of PSO 200
ω 0.5
Pa 0.25
Levy component L 1.5
Maximum iteration 300
Traffic source CBR

13
 C. Mohanadevi, S. Selvakumar

The performance of the algorithm is evaluated using various QoS parametes such as
throughput, Packet Delivery Ratio, end-to-end delay, network lifetime and energy con-
sumption as defined in [20].

5.1 Network Life Time

The network’s lifespan is determined by the number of active sensor nodes and the WSN’s
route sustainability. The sensor node will be active as long as its energy is not drained
fully. The proposed protocol is compared with the existing protocols in terms of network
lifetime. As shown in the Fig. 5, the network life time increases with the number of nodes
in the network irrespective of protocols. However, the proposed protocol has significantly
improved the lifetime when the number of nodes compared to the EE-LEACH,EPSO-CEO
and OQoS-CMRP protocols due to the selection and replacement of cluster heads.

5.2 Energy Consumption

The energy consumption of node is computed using the following equation. The total
energy consumption is computed as a sum of all nodes energy consumption in the network.
Econsumption = Einitial − Eresidual

The energy consumption is steadily increased as the number of nodes is increased in the
network. As shown in Fig. 6, the energy consumption of proposed algorithm is decreased
by 40.06%, 32.4%, and18.21% compared with EE-LEACH, EPSO-CEO and OQoS-CMRP
protocols.

5.3 Throughput

It is defined as the number of data packets received by the sink node from the source node
at a particular time period. It is found that all four protocols have almost the same through-
put when the simulation time is less than 200 s as in Fig. 7. However, the proposed proto-
col significantly improves the throughput as the simulation time increases when compared
to the other three protocols when the simulation time is increased.

30000

25000
Network Lifeme(Sec)

20000
EE-LEACH
15000
EPSO-CEO
10000
OQoS-CMRP
5000 Proposed
0
50 100 150 200 250
Number of Nodes

Fig. 5  Comparison of network lifetime

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

Energy Consumpon
25

Energy Consumpon(J)
20

15 EE-LEACH

10 EPSO-CEO
OQoS-CMRP
5
Proposed
0
50 100 150 200 250
Number of Nodes

Fig. 6  Comparison of energy consumption

5.4 Packet Delivery Ratio (PDR)

It is the ratio between the numbers of data packets delivered to the sink node and number
of packets sent by the source node. It is found that the delivery ratio of protocol is more
than 5.55% higher than the OQoS-CMRP protocol as shown in Fig. 8.

5.5 End‑to‑End delay

That’s the cumulative time it takes to send a packet from a source node to a sink node.The
proposed protocol has optimal end to end delay as compared to the the EE-LEACH, EPSO-
CEO and OQoS-CMRP protocols.
As shown in the Fig. 9 the EE-LEACH protocol has maximum end to end delay as com-
pared to other three routing protocols. But, the proposed protocol considerably reduces the
end to end delay as the simulation time is increased when compared to the OQoS-CMRP
protocol. This is because of selection routes with less traffic during route establishment
phase. QoS adopted performance of proposed method achieves the result by comparing
above results. Depends on the performance of delay, PDR, throughput and network life-
time, the QoS is analyzed. Three different models are compared to proposed work, which
obtains the better outcome.

Fig. 7  Comparison of throughput Throughput

80
Throughput(Kbps)

60
EE-LEACH
40
EPSO-CEO
20 OQoS-CMRP

0 Proposed
100 200 300 400 500
Simula on Time

13
 C. Mohanadevi, S. Selvakumar

Fig. 8  Comparison of packet Packet Delivery Rao

delivery ratio 100

PDR In Percentage
90
EE-LEACH
80
EPSO-CEO
70
OQoS-CMRP
60 Proposed
100 200 300 400 500
Simulaon Time

6 Conclusion

The proposed QoS-aware Multipath Routing (QMR) protocol has established multiple
optimal paths between the source and sink node in the wireless sensor network (WSN). In
the protocol, the hybrid Cuckoo Search-Particle Swarm Optimization algorithm is used for
clustering the sensor nodes and chooses the cluster heads to ensure the reliable data deliv-
ery. The proposed protocol uses several paths for delivering data packets and has excellent
control over data traffic in the network. Simulation results show that the proposed protocol
outperforms the existing QoS centric protocols like EE-LEACH, EPSO-CEO and OQoS-
CMRP in terms of throughput, end-to-end delay, Packet Delivery Ratio, network lifetime
and energy conservation. Hence, the protocol can be used in various delay sensitive appli-
cations. In future, the efforts should be made to improve the security aspects of the proto-
col. Moreover, the performance of the protocol will be assessed in Wireless multimedia
sensor networks. In future, the work algorithm is improved with other deep learning and
swarm intelligent techniques.

End-to-End Delay
0.06

0.05
End-to-End Delay (ms)

0.04
EE-LEACH
0.03
EPSO-CEO
0.02
OQoS-CMRP
0.01 Proposed
0
100 200 300 400 500
Simula on Time

Fig. 9  Comparison of end to end delay

Declarations

Conflict of interest Conflict of interest is not applicable in this work.

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

Ethics Approval and Consent to Participate No participation of humans takes place in this implementation
process.

Human and Animal Rights This article does not contain any studies with human participants or animals per-
formed by any of the authors. No violation of Human and animal rights is involved.

References
1. Barati, H., Movaghar, A., & Rahmani, A. M. (2015). EACHP: Energy Aware clustering hierarchy protocol
for large scale wireless sensor networks. Wireless Personal Communication, 85(3), 765–789.
2. Adu-Manu, K. S., Adam, N., Tapparello, C., Ayatollahi, H., & Heinzelman, W. (2018). Energy harvest-
ing wireless sensor networks (EH-WSNs): A review. ACM Trans Sensor Networks (TOSN), 14(2), 1–50.
https://​doi.​org/​10.​1145/​31833​38
3. Sohraby, K., Minoli, D., & Znati, T. (2007). Wireless sensor networks: Technology, protocols, and applica-
tions. John Wiley & Sons.
4. Villalba, L. J. G., OrozcoAna, L. S., Cabrera, A. T., & Abbas, C. J. B. (2007). Routing protocols in wireless
sensor networks. IEEE Transactions on Parallel and Distributed Systems, 1, 919–931.
5. Sohrab, K., Gao, J., Ailawadh, V., & Pottie, G. J. (2000). Protocols for self-organization of a Wireless Sensor
Network. Personal Communication Journal, 7(5), 16–27.
6. Heinelman, W. R., Chandrakasan, A., Balakrishnan, H. (2002). Energy-Efficient Communication Protocol for
Wireless Microsensor Networks. In Proceedings of the 33rd Annual Hawaii International Conference on
System Sciences (pp. 4–7). USA.
7. Haque, M., Ahmad, T., & Imran, M. (2018). Review of hierarchical routing protocols for wireless sensor net-
works. In Intelligent communication and computational technologies (pp. 237–246). Springer.
8. Yan, J., Zhou, M., & Ding, Z. (2016). Recent advances in energy-efficient routing protocols for wireless sen-
sor networks: A review. IEEE Access, 4, 5673–5686. https://​doi.​org/​10.​1109/​ACCESS.​2016.​25987​19
9. Robinson, Y. H., Julie, E., Kumar, R., et al. (2019). Probability-based cluster head selection and fuzzy mul-
tipath routing for prolonging lifetime of wireless sensor networks. Peer-to-Peer Network Applications, 12,
1061–1075.
10. Ganesan, D., Govindan, R., Shenker, S., & Estrin, D. (2001). Highly-resilient, energy-efficient multipath
routing in wireless sensor networks. In Proceedings of the 2nd ACM international symposium on Mobile
ad hoc networking & computing (MobiHoc ’01). Association for Computing Machinery, New York, NY,
USA, 251–254. https://​doi.​org/​10.​1145/​501449.​501452
11. De, S., Qiao, C., & Wu, H. (2003). Meshed multipath routing with selective forwarding: an efficient
strategy in wireless sensor networks. Computer Network, 43(4), 481–497. https://​doi.​org/​10.​1016/​S1389-​
1286(03)​00355-4
12. Younis, O., & Fahmy, S. (2004). Distributed clustering in ad-hoc sensor networks: a hybrid, energy-effi-
cient approach. IEEE INFOCOM 2004 (p. 640). March 7–11, 2004, Hong Kong, China. https://​doi.​org/​10.​
1109/​INFCOM.​2004.​13545​34

13
 C. Mohanadevi, S. Selvakumar

13. Yang, Y., Bai, E., Hu, J., & Wu, W. (2010). MRBCH: A multi-path routing protocol based on credible
cluster heads for wireless sensor networks. International Journal on Communications Network and System
Sciences, 3, 689–696. https://​doi.​org/​10.​4236/​ijcns.​2010.​38092
14. Sohrabi, K., Gao, J., Ailawadhi, V., & Pottie, G. J. (2000). Protocols for self-organization of a wireless sen-
sor network. IEEE Personal Communications, 7(5), 16–27.
15. He, T., et al. (2003). SPEED: A stateless protocol for real-time communication insensor networks. In Pro-
ceedings of the IEEE international conference on distributed computing systems (pp. 46–55)
16. Felemban, E., Chang-Gun, L., & Ekici, E. (2006). MMSPEED: Multipath multi speed protocol for QoS
guarantee of reliability and timelines in Wireless SensorNetwork. IEEE Transactions on Mobile Comput-
ing, 5(6), 738–754.
17. Bagheri, T., Ghaffari, A., (2011). RECM: Reliable and energy effective clustering based multipath routing
algorithm for Wireless Sensor Networks. In Proceedings of IEEE world congress on information and com-
munication technologies (pp 1340–1345)
18. Mazaheri, M. R., Homayounfar, B., & Mazinani, S. M. (2012). QoS based and energyaware multipath
hierarchical routing algorithms in WSNs. Wireless SensorNetw., 4, 31–39.
19. Almalkawi, I. T., Zapata, M. G., & Al-Karaki, J. N. (2012). A cross-layer-based clustered multipath rout-
ing with QoS-aware scheduling for Wireless Multimedia Sensor Networks. Int. J. Distrib. Sensor Netw., 1,
1–11.
20. Deepa, O., & Suguna, J. (2020). An optimized QoS-based clustering with multipath routing protocol for
Wireless Sensor Networks. Journal of King Saud University - Computer and Information Sciences, 32(7),
763–774.
21. Shahbaz, A. N., et al. (2021). Multipath routing through the firefly algorithm and fuzzy logic in wireless
sensor networks. Peer PeerNetw. Appl., 14(2021), 541–558.
22. Kshirsagar, P., Balakrishnan, N., & Yadav, A. D. (2020). Modelling of optimised neural network for classi-
fication and prediction of benchmark datasets. Computer Methods in Biomechanics and Biomedical Engi-
neering: Imaging & Visualization, 8(4), 426–435. https://​doi.​org/​10.​1080/​21681​163.​2019.​17114​57
23. Kshirsagar, P., Akojwar, S., & Bajaj, N. (2020). A hybridised neural network and optimisation algorithms
for prediction and classification of neurological disorders. International Journal of Biomedical Engineer-
ing and Technology, 28(4), 1. https://​doi.​org/​10.​1504/​IJBET.​2018.​095981
24. Kshirsagar, P., & Akojwar, S. (2016). “Optimization of BPNN parameters using PSO for EEG signals”,
ICCASP/ICMMD-2016. Advances in Intelligent Systems Research., 137, 385–394.
25. Kshirsagar, P., & Akojwar, S. (2015). Classification & Detection of Neurological Disorders using ICA &
AR as Feature Extractor. International Journal of Series Engineering Scientific IJSES, 1(1), 1–6.
26. Akojwar, S. G., & Kshirsagar, P. R. (2016). A novel probabilistic-PSO based learning algorithm for opti-
mization of neural networks for benchmark problems. In WSEAS International conference on Neural Net-
work-2016, Rome, Italy.
27. Kshirsagar, P. R., Manoharan, H., Al-Turjman, F., & Kumar, K. (2020). DESIGN AND TESTING OF
AUTOMATED SMOKE MONITORING SENSORS IN VEHICLES. IEEE Sensors Journal. https://​doi.​
org/​10.​1109/​JSEN.​2020.​30446​04
28. Joseph, S., & Rajaram, A. (2017). A Novel Enhanced SVM cluster based secure and effective routing pro-
tocol for node authentication in Mobile Ad Hoc Networks. International Journal of Computer Technology
and Applications (IJCTA), 10(19), 25–39.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

13
A Qos‑Aware, Hybrid Particle Swarm Optimization‑Cuckoo Search…

C. Mohanadevi pursuing Doctor of Philosophy in the Faculty of Infor-

mation and Communication Engineering from the Anna University.
She received the Master of Engineering in Applied Electronics from
Thanthai Periyar Government Institute of Technology affliated to Anna
University. She is working as Associate Professor in Mailam Engineer-
ing College, Tindivanam, Villupuram. Her current research interests
includes MANETS, Wireless sensor Networks and Embedded sys-
tems. She has authored papers in International journal and conferences
in her research areas and also has published a patent.

S. Selvakumar received Doctor of Philosophy in Computer Science

and Engineering from the Anna University. He received the Master of
Engineering in Computer Science & Engineering from the Madurai
Kamaraj University. He received the Master of Business Administra-
tion from the same University. He is working as Professor in GKM
College of Engineering & Technology, Chennai. His current research
interests include Data Analytics, Software Engineering, and Networks.
He has authored more than 130 referred papers in international jour-
nals and conferences in his research areas. He has published patents
and has written over 10 book chapters in the emerging areas of Com-
puter Science. He is a Senior Member in Computer Society of India,
Member in IEEE, ACM, Institution of Engineers and ISTE. He is
reviewer in various International journals and IEEE conferences.

13