Enhancing Congestion Control using a Load-Balanced Routing Algorithm for Distributed Networks

International Journal of Computer Networks & Communications (IJCNC) Vol.17, No.1, January 2025
DOI: 10.5121/ijcnc.2025.17102 11
ENHANCING CONGESTION CONTROL USING A
LOAD-BALANCED ROUTING ALGORITHM FOR
DISTRIBUTED NETWORKS
Jogendra Kumar
Department of Computer Science and Engineering, G.B.Pant Institute of Engineering and
Technology, Ghurdauri, Pauri Garhwal Uttarakhand, India
ABSTRACT
Ad hoc networks frequently encounter congestion due to packet loss, link failures, and limited bandwidth.
These issues lead to significant energy and time expenditures for congestion recovery, ultimately
degrading network performance. Various techniques exist to mitigate the effects of congestion, and this
paper introduces a novel routing protocol named Load-Balanced Congestion-Adaptive Routing (LBCAR)
protocol, which incorporates a random route point model in Mobile Ad Hoc Networks. The paper promotes
an adaptive load-balanced routing approach combined with congestion control. A hybrid protocol is
proposed to address congestion control and achieve optimal performance. The algorithm integrates
metrics such as traffic density, routing path lifetime, and link failure detection to enhance network
performance. The results of the proposed approach are compared with those of other routing protocols to
assess its effectiveness.
KEYWORDS
AdHoc networks, Congestion control, data delivery, Load balancing, Mobile Networks
1. INTRODUCTION
Ad hoc wireless networks are very popular due to their reliability & capacity to balance traffic
and congestion control, QoS without much infrastructure. Such quality makes it very useful for
multiple applications since all network controls are with only nodes making it less energy
efficient. To save critical energy which is limited for any node we need a dynamic protocol that
can save energy and create a routing infra that can be established dynamically with self-
establishing and self-management on the needed times [1-2]. The possible utilization of ad hoc
networking incorporates business partners sharing data during a meeting, students utilizing
workstations, and laptops to take part in a lecture, emergency disaster relief personnel organizing
endeavors after a seismic tremor or storm, and soldiers handing off data for situational awareness
on the war zone [3]. During the simulation process in any random allocation of nodes, every
mobile node initiates from and forward to a random location [4]. Every node stays fixed for a
predetermined timeframe which is called delay time and subsequent moves in an actual line to a
few up-to-date arbitrarily picked areas at an arbitrarily picked acceleration to the predefined
maximum speed. When arriving at that new area, the node stays there for the delay time and
afterward picks another irregular location to continue the transmission process at some changed
arbitrarily picked speed, the node keeps on rehashing this conduct all through the execution time
[4-5]. It is found that this technique can create a lot of qualified node development and network
geography change [6-7]. Congestion is a major issue in MANET. It happens because of reasons
including the conduct of the multiple routers, hosts, the multiple transmission among routers, and
the media, and happens because of restricted assets at any phase of the route. Congestion happens

12
because users send a greater number of data than the network hosts can oblige, subsequently
making the buffer on such hosts top off and plausibility of flood prevails. The result of this can
be packet loss, and delay in delivery, and also leads to a decrement in the overall performance of
the network [8-9]. So, to diminish congestion, any protocol performing a routing process ought to
diminish the quantity of data packets in the network. In any case, essentially dropping flooded
packets will decrease information loyalty and increase energy dissemination. A lot of studies had
been proposed before, which tended to few of the performance features only. Different methods
have been created in an endeavor to limit congestion in ad hoc networks. Moreover, a simple
utilization of multipath routing plans could provocatively influence the lifetime network as the
collapse pace of energy will be higher [10-11]. This paper is structured into six main sections.
Section 1 offers an introduction to the research, summarizing the core concepts. Section 2 covers
related work, exploring existing studies that provide foundational insights and guide the creation
of a new framework. Section 3 discusses the proposed approach, explaining the methodology and
algorithm used in the research. Section 4 delves into the design of both the simulations and the
framework. Section 5 presents and analyzes the results, assessing optimization based on the
simulation factors outlined earlier. Lastly, Section 6 provides the conclusion and discusses
potential areas for future research.
2. RELATED WORK
One of the core challenges in these models is achieving an optimal combination for enhanced
quality of management in MANETs. However, some relevant approaches have been identified in
the literature. For instance, in [12-13], the authors proposed a flexible Genetic Algorithm aimed
at optimizing channel allocation in mesh wireless networks. A detailed Multi-Objective Cellular
Genetic Algorithm was proposed for tracking down an ideal transmission approach in MANETs
[14]. In [15] author reviewed the routing protocol named RPL under a heterogeneous traffic
design. In [16] authors proposed an improved algorithm dependent on Queue-workload-based
conditions (QWL-RPL). This proposed protocol accomplished a dependable way with
improvised overall results. The final product demonstrates that all things considered, there is a
12%–30% decrease in delay, a 25%–45% decrease in overheads, a 20%–40% decrease in jitter,
and a 5%–30% improvement in PRR. In [17-19] work proposed is about the RCER protocol that
utilizes the heterogenic nodes on the basis of their energy level. It consists of two processes; one,
to make the whole network extra energy-proficient, the network field is separated into
topographical groups, and another; to improve the next-hop selection, RCER endeavors ideal
routing in view of energy variance, value of Round-Trip Time and hop-count factors. In addition,
in light of processing the estimation of wireless connections and hubs status, this protocol re-
establishs routing ways and gives network dependability, and improved data delivery
performance. Several researchers proposed many algorithms [20-22] for efficiently handling the
limited resources. In [23-24] load distribution approach has been presented for energy efficient
MANET routing protocol. To implement the load reduction approach by adjusting the energy
utilization of all portable nodes by choosing a route with high energy nodes as opposed to the
shortest route [25]. Some MANET routing protocols depend on particular connection layer
properties, for example, CTS/RTS control succession utilized by prominent IEEE 802.11, and
MAC layers to keep away from impacts because of concealed and uncovered terminals. In
particular, before transmitting an information casing the source station sends a short control edge,
named RTS, to the getting station declaring the impending covering transmission [26]. Fuzzy
logic-based Load-Balanced Congestion Control improves MANET performance by
dynamically managing load, reducing packet loss, and enhancing efficiency [27]. The
proposed CFRS-CP method estimates route congestion probability using MAC overhead,
link quality, neighbor density, and vehicle velocity, leveraging these parameters to
optimize routing decisions [28]. The proposed LBCAR algorithm outperforms the other two
protocols, especially for applications that require high data transmission rates, fast response

13
times, and energy-efficient operations. Its load balancing and congestion awareness mechanisms
play a crucial role in achieving these improvements, making it suitable for use in large-scale and
energy-constrained networks such as wireless sensor networks or ad-hoc networks.
3. PROPOSED WORK
The default valuation of parameters utilized as a part of CALB provides direct QoS. In this
manner, considering the effect of valuation of parameters on the system execution attempts to
find an ideal valuation of parameters for LBCAR before sending. There are 9 parameters utilized
as a part of LBCAR where the values of possible combinations of estimation of parameters are
huge such as 1011 sets. This inspires to usage the meta-heuristic specifically ideal to handle the
combinatory exploration work [29-30]. Though for a weak node, the proficiency of a route
recovery system is made in this sort of implies that comparing routes are working to the robust
nodes. With the aid of the simulated results, the minimization of data loss and delay making use
of the proposed adaptive technology [31].
3.1. Proposed Model
The protocol proposed in this research work is named LBCAR- load-balanced-congestion-
adaptive-routing protocol. It is a combination of an energy-efficient and congestion adaptive
technique to upsurge the overall throughput value in the system [32-33]. In this protocol, a record
of the most recent traffic load assessments is maintained by every node with its nearby nodes in a
table called the local table. The route’s lifetime is decided by measurement of connection cost
and the congestion position of the route is decided by the measurement of traffic load. The route
with greatest life time and low traffic load is chosen for data transmission. This algorithm expects
that there are no unidirectional connections in the organization. This protocol essentially restricts
the overestimated highest amount of packets communicable over the route consuming most
fragile hub through high traffic load power and last lifetime [34-41].
To calculate the traffic load, let’s assume node(xi) as a sample for interface queue length N is the
amount of time count for sampling over total execution time qi(j) is jth
sample value, This equation
computes the traffic load for a node xi,
For the node xi the total length of interface queue is qmax(i) formerly for the node xi the traffic-
load-intensity-function is presented as following: The traffic load intensity is the ratio of the
node's actual traffic load to its maximum capacity.
This equation gives the link cost between two nodes a and b.For a link (a, b), link cost [12] has
two aspects: first ‘Ea,b’ is link-specific parameter and second is Pa,b node specific parameter or
Pa,b is the residual energy of the hub or node. Ea,b is the energy consumed for at least one

14
successful transmission over the link. The link cost is a ratio of the residual energy to the energy
required for a transmission.
where Ea,b is the energy consumed for at least one retransmission basic even with connection
error and Pa,b is the residual energy of the hub. Ea,b is estimated as
n is number of retransmission over link layer (hop by hop)
where pa,b is the packet error probability, Ea,b is a single packet transmission energy consumption
value. Let the quantity of adjoining hubs of xi is n, and all the functions of power are identified
for example (i) value of xi itself, just as the traffic load power. These n+1 value are arranged in
the increasing order. These are provided a succession value called as seq(i) and calculated as
The probability of forwarding rate of the information for the hub xi is assumed by the below
formula:
pi is identified with the current power of the traffic load of xi. It relies upon overall size of the
traffic load in local area of xi. In ad hoc networks, the density of nodes differs from areas and
nodes are dynamically distributed. The least the general traffic load is, the bigger the sending
possibility, all the nodes will link the route favorably and vice versa. The above equation
explained that in particular areas where the power is smaller and n is higher i.e., density is higher,
there the routing overhead and redundant forwarding of data both are condensed. Power is also
tied to link cost: as the probability of forwarding data decreases, so does the link cost, and the
reverse is true as well. In areas where node density is lower and power is higher, the likelihood of
establishing a route increases [19]. Along these lines, improvement in life time of the network
and load balance to enhance the overall network routinely. At the time of computation, route
computed can be ideal, but at the time of simulation the irregular traffic examples will
conceivably make the at present chosen ways ideal sooner or later. Thus, LBCAR is a protocol
utilized for route selection that incorporates systems for occasional and circulated route selection
and computation.
3.2. Network Architecture
Network design and presumptions, study a MANET with enormous amount of hubs conveying
by multi-hop paths with one another. The hubs might be laptops, PDAs, or cell phones which

15
may transfer oftentimes. On-demand routing protocol like DSR or AODV are used by nodes for
starting multi-hop routing route. In this study we mainly explain the processes of queue in
transmission layer management, connection failure recognition and route repair in local region
method. As in MANETs, nodes are self-coordinating and autonomous from one another, they can
move arbitrarily and regularly. Thus, the link failure may happen because of profoundly
decreased received signal strength. Also, the geography of MANETs changes very recurrently,
that causes into failure of links between neighbouring nodes. Link failure causes to broken route
among source to destination that leads to overall reduction in throughput as well. Hence, this is
an independent exploration subject in wireless networks. discovery can be executed utilizing
either intermittent CONNECT token messages or connection layer criticism. These CONNECT
token messages are neighborhood promotions for the proceeded with incidence of the connection.
The nodes in the proposed instrument trade CONNECT token messages intermittently to
guarantee link connectivity. The resulting two constraints are related with a CONNECT token
message: The amount of highest time span amid two sequential CONNECT token message
transportation is named as CONNECT-INTERVAL and the highest number of loss of
CONNECT token messages that a hub can endure before it proclaims the link failure is named as
CONNECT-LOSS permitted. In the event that a node doesn't get any CONNECT token message
since its neighbor node inside CONNECT-LOSS permitted CONNECT-INTERVAL, at that
point the node indicates that the connection is no longer available for data transmission. Local
repairing of routes can decrease the impact of link failure on network performance. Link failure
reduces the execution of a network significantly. In the existing studies, we have lot of local route
repair techniques available. An old method to keep the historical backdrop is known as Query
localization method and it overflows the RREQ-route request message to some confined
constrained area through local query procedure.
3.3. The CALB Architecture
Our proposed strategy leverages the cross-layer storage capability of nodes, spanning both the
transmission and network layers. In this approach, when any intermediate node detects a link
failure, a route disconnection notification message is generated and directed toward the source
node. Upon receiving this notification, all intermediate nodes along the selected route cease
forwarding data packets to prevent further transmission.They stored the approaching information
parcels in their neighbourhood transmission layer lines When link failure happens to avoid the
data losses it collects packets into the transmission layer queue and initiate to find somewhat
other fractional temporary way to the receiver node. At the point after the main sender node gets
the route disconnection notification control message, it simply breaks the communication of
information packets and waits for any further information about new fractional temporary path to
the destination. If any fractional temporary path is found, intermediate nodes makes and leads
back to the main sender another warning message known as RSN-route-successful-notification.
At that point continues its transmission interaction from the transmission layer line. [20]

16
Fig. 1: Message exchange process in CALB
A distinct control message, called a 'Route Unsuccessful' (RUN) message, is sent by an
intermediate node back to the source node if no partial path is found. Upon receiving the RUN
message, the source node initiates a new route discovery process to find an alternative path.
Entirely midway nodes continue their communication from the neighborhood buffer in the wake
of accepting RSN message. At the point source hub continues the transmission cycle when it gets
this RSN message,. The source hub just retransmits the dropped packets during the connection
failure. So, as not like the conventional routing protocols, the source node doesn't have to
convey all the information packets. Thus, packet delivery ratio is maximized, the packet drop rate
is decreased and inclusive throughput value improved. A congestion-aware routing system is
used by CALB so as to provide easy identification of the level of congestion in the network by
the nodes and to make a suitable move. In a heavily loaded congested system, nodes stored the
packets coming to them even after link failure to diminish packet drop rate. Accordingly, CALB
controls the clog in MANET. For a busy network, in CALB the nodes need to send an ALERT
message so as not to upsurge the forwarding rate of data to their previous nodes. The buffer
packets are collected in their local queues so as to deliver consistent data delivery and to handle
congestion mechanisms. Each transitional node keeps a link in the transmission layer for different
end-point hubs with the assistance of cross-layer line. Fig. 1 describes the processing, as source
node B distributes data packets to A, S, and D as three destination nodes. As the links B to C and
C to D are failed which were the original shortest routes to destination D, then node B the storing
the information packets in its transmission layer it doesn't have to imprint any new succession
number for every packet during protecting. Moreover, B starts for new partial path and found the
route to D as B to E, E to S, S to F, F to G, G to H and finally H to D while it is a longer path as
compared to original route. In this way, the transmission layer line is just for temporarily storage
of the packets.
3.4. Proposed Algorithm
To handle link failure cases to manage the congestion, nodes can store packets in their limited
queue in the transmission layer in CALB processing. Accordingly, every versatile hub plays out
some particular capacities as opposed to ordinary portable hubs. We will portray the overall tasks
in detail as in the following process. In traditional routing protocols, whenever the source node
needed to send packets it begins route route-finding process. It creates a token message as RREQ
and broadcast it in the network. Then until any shortest path (route) to the destination is detected
source node wait for further activity. furthermore, upon receiving the token message, the receiver
node sends a route reply message back to the source node. Each time the sender node receives
this route reply message, it initiates the information transmission process.In the event that there
happens a connection failure because of node portability or some other reasons like restricted
transfer speed, absences of energy, and so on, and if the sender node receives a route failure
warning message, it triggers specific actions to address the failure, it further stops transmission of
information packets. So, after failure of the shortest path source node restarts another new route
discovery cycle to convey the information packets. Source node needed to retransmit the whole
set of packets, even if this is longer path than the original shortest path while it diminishes the
data delivery rate. In the proposed model, when a link failure occurs, the source node halts its
transmission and temporarily waits for a partial path to be established instead of immediately
initiating a new route discovery process to the receiver. Once a partial temporary path is found,
the source node is notified, allowing it to resume transmission. The decrement in data delivery
rate means decrement in the overall performance of the network. The below algorithm shows that
the set of all source nodes is S and s is source node, where.

17
The design choices made in the LBCAR algorithm focus on improving network performance,
particularly under the conditions of link failures and congestion, by offering a more dynamic and
adaptive approach than traditional routing protocols. One of the primary reasons this approach
was chosen over others is its ability to handle link failures and congestion efficiently without
necessitating a complete restart of the route discovery process, which is a common drawback in
traditional MANET routing protocols. Here’s a comprehensive explanation of how these design
choices contribute to the success of the protocol:
 Handling Link Failures with Partial Path Recovery: Traditional routing protocols
typically rely on the discovery of the shortest path between a source and a destination,
which is efficient under ideal conditions but problematic when link failures occur due to
node mobility, bandwidth limitations, or energy depletion. In these situations, the source
node often needs to restart the entire route discovery process, resulting in significant
delays, packet retransmissions, and performance degradation. In contrast, LBCAR takes
a more proactive approach. When a link failure occurs, rather than initiating an entirely
new route discovery process, the algorithm temporarily halts packet transmission and
waits for a partial path to be re-established. This reduces the overhead and delays
associated with frequent route rediscovery, allowing the source node to quickly resume
transmission once an alternative path is found. This design decision contributes to higher
data delivery rates and reduces packet loss, which is particularly important in networks
with high mobility or frequent disconnections.
 Congestion Management with Adaptive Queuing: In LBCAR, congestion is managed
effectively by utilizing the limited queue available at each node in the transmission layer.
Unlike traditional routing protocols that may struggle with managing high traffic loads or
congestion, LBCAR introduces congestion awareness by allowing nodes to temporarily
store packets in their queue while searching for an alternate route during congestion. This
adaptive queuing mechanism helps manage temporary traffic spikes without
overwhelming the network, preventing congestion from spreading and affecting overall
performance. This approach ensures that the network remains balanced, even during
periods of high traffic, by distributing the load dynamically. As a result, queue
management at the node level contributes to smoother data flow, reducing delays and
improving overall network efficiency.
 Avoidance of Immediate Route Discovery Restarts: Traditional protocols require a
full restart of the route discovery process whenever a connection failure occurs, even if
the shortest path is no longer available. This leads to significant overhead, as the source
node must re-broadcast route requests and wait for replies, which in turn increases delays
and degrades performance, especially in scenarios with frequent disconnections or node
mobility. LBCAR mitigates this by pausing packet transmission and waiting for an
alternate or partial route to be found before attempting a full route discovery. This design
choice reduces unnecessary route discovery cycles, ensuring that data transmission can
resume more quickly after a failure. Additionally, by utilizing temporary paths, LBCAR
minimizes the retransmission of packets, which conserves bandwidth and energy,
ultimately improving the network lifetime.
 Load Balancing for Improved Resource Utilization: Another critical advantage of
LBCAR is its load-balancing mechanism, which distributes traffic across multiple nodes
to prevent any single node from becoming a bottleneck. Traditional routing protocols,
especially those that prioritize shortest-path routes, often overburden certain nodes,
leading to congestion and faster energy depletion in those nodes. By taking both
congestion and node energy levels into account when making routing decisions, LBCAR
balances the load more effectively. This prevents individual nodes from becoming
overwhelmed and helps prolong the lifetime of the network by ensuring that energy
resources are used more evenly across the entire network.

18
 Energy Efficiency and Network Longevity: In mobile ad-hoc networks, energy
efficiency is a major concern, particularly for portable nodes with limited battery life.
LBCAR addresses this by incorporating energy-aware routing decisions. Instead of
focusing solely on the shortest path, LBCAR selects routes based on the energy
availability of nodes, ensuring that low-energy nodes are not overburdened and can
continue functioning longer. This helps extend the operational lifetime of the entire
network.
 Contribution to Protocol Success: The combination of these design choices—handling
link failures through partial path recovery, adaptive queuing for congestion management,
and load balancing for efficient resource utilization—makes LBCAR more resilient,
responsive, and efficient in high-traffic, high-mobility environments. By reducing the
need for frequent route discovery and managing congestion adaptively, LBCAR ensures
higher data throughput, lower latency, and more balanced energy consumption across the
network. These features make LBCAR particularly suited for large-scale, energy-
constrained networks such as wireless sensor networks and mobile ad-hoc networks
(MANETs), where traditional approaches often fall short due to their reliance on fixed
route discovery processes and their inability to adapt to dynamic network conditions
effectively.
Algorithm (CALB):
any source node s ∈ S
Broadcast message m ∈M
Initiate
s broadcasts RREQ m
s stays on-hold until any m is acknowledged
if (m=RREP) at that time
transporting of data packets initiates by s
else if (m=RDN) at that time
transmission is stopped by s and it pauses for a partial temporary path
else if (m=RUN) at that time
a fresh route discovery process is initiated by s
else if (m=RSN) at that time
transmission process is resumed by s
else
s will again paused for any new message
end
End
The CALB algorithm manages message broadcasts and route discovery in a network. A source
node s begins by broadcasting a Route Request (RREQ) and waits for an acknowledgment. If it
receives a Route Reply (RREP), it starts transmitting data; if it gets a Route Discovery Negative
(RDN), it stops and pauses for an alternative path. If a Route Update Needed (RUN) message is
received, the node re-initiates the route discovery process, while a Route Service Normal (RSN)
message signals that the route is valid, allowing the node to resume transmission. If none of these
messages arrive, the node stays on hold awaiting further instructions.
This paper investigated the packet-sending issue of route estimation of the LBCAR protocol is
utilized and it does not demonstrate the ideal outcome. So we are changing the default estimation
of protocol by utilizing the enhancement procedure and discovering the ideal consequence of
sending packet.
Problem statement:

19
 Simulation of basic LBCAR protocol with standard value.
 Optimization system to consequently tune the LBCAR design.
 Performance assessment of streamlined LBCAR Protocol
4. SIMULATION FRAMEWORK
In the proposed scenario, the experimental network setup utilizes omnidirectional antennas for
each mobile node, employing a Two-Ray ground radio propagation model. The simulation
framework for this study is developed based on the configuration of a dynamic network with a set
of three parameters for performance measurement using network simulation software Network
Simulator Version 2. The experimental setup is shown in Table I and simulation parameters are
given in Table II. The Simulation design with 50 nodes is shown in fig. 2. The communication
among the network is shown through the transmission of 100 to 1,000 data packets with a
minimum speed of 4 packets each second to 25-100 packets each second.
Table I Experimental Setup
Parameters Values
Channel Type Wireless Channel
MAC type 802_11
Max packet in queue 50
Model of Antenna Omni Antenna
Network interface type Wireless Physical
Number of mobile nodes 85
Routing protocol LBCAR
Type of Interface queue Drop Tail
Link layer type LL
Frame dimensions 3000 * 3000
Simulation time 10
Range of nodes 40
4.1. Experimental Design
Based on the described technique, we will incorporate the waypoint technique along with load
balancing to optimize and enhance the derived parameters. A parallel event-driven testing
framework, NS2, was used in conjunction with VMware to obtain the expected results of the
protocols. The simulation tests were continuously conducted on a computer running NS2 within a
VMware virtual machine, allowing us to assess the effects of simulation speed and framework
configuration on the experimental outcomes. The implementation follows a systematic design,
with the modules outlined below:
Module 1
 Base files will be created for MANET design and transfer of packets.
Module-2

20
 Topology of MANET with a particular number of nodes deployed in any dynamic
network
 Transmission of packets among the nodes by using existing two protocols AOMDV-ER
AND AOMDV.
 Values of parameters such as delay time, throughput, and energy variance are calculated
according to given optimization parameters for two existing protocols.
Module-3
 Topology of MANET with a particular number of nodes deployed in any dynamic
network
 Transmission of packets among the nodes by using the proposed LBCAR protocol (which
is developed as a hybrid technique cum protocol using C++ in the NS2 package).
 Values of parameters such as delay time, throughput and energy variance are calculated
according to given optimization parameters (using the proposed routing protocol).
Simulation design: With 50 nodes
Fig. 2: Simulation design with 50 nodes
Table II Simulation Parameters
Sr. No. Parameters Values
1 Node placement Random
2 Propagation Two-way ground
3 Size 1500 X 1500
4 Nodes 150
5 Transmission range 250 m
6 Bandwidth 1 Mbps
7 Traffic CBR
8 Packet Size 512
9 Mobility Random way point
4.2. Performance Metrics
The performance metrics in the proposed work such as delay time, throughput and energy
variance have been estimated for our proposed LBCAR algorithm and compared with AOMDV

21
and AOMDV-ER. The experimental parameters are to be calculated to increase network life and
network effectiveness.
 Delay time: It is a measurement of time consumed by every packet transmission from
the sender to receiver. The one of the indications of network congestion is higher end-to-
end delay.
(7)
Arrival Time: When the packet reaches the receiver.
Sent Time: When the packet leaves the sender.
N: Total number of packets sent successfully
 Throughput: Number of bytes received of data × 8 / Simulation time × 1,000 kbp (8)
Total Bytes Received: Total data received (in bytes).
Simulation Time: Total time of the simulation (in seconds).
 Energy Variance: It is the measurement of total energy utilization by the nodes reduced
by change in the energy after a fixed interval of time.
(9)
Initial Energy: Energy of the node before the process starts.
Remaining Energy: Energy left in the node after a certain time.
N: Total number of nodes in the network.
Table III Parameters for Node Processing
Sr. No. Parameters
1 Route selection time
2 Route link time
3 Amount of energy
4 Total time
5 Fixed wait time
6 Threshold energy
7 Delay time
8 Node number
To cover a more extensive region for message gathering, some neighboring vehicles can serve as
potential forwarders, and each forwarder needs to sit tight for a specific timeframe (i.e., dispute
time) before sending the message.
5. RESULTS AND ANALYSIS
Simulation results have demonstrated the efficiency of the proposed LBCAR protocol for
MANET. We measured the data delivery performance under three major parameters by varying
the number of nodes for three different protocols. The results shown are as below
Input nodes= 100

22
Table IV Performance results of proposed protocol LBCAR
No. of nodes. Throughput
Delay Energy variance (no.
of nodes x50)
20 198 0.10713 0.35
40 206 0.14291 1.38
60 218 0.19962 2.5
80 229 0.23265 4.3
100 234 0.27112 7.6
Table IV shows the results of the proposed protocol LBCAR for parameters delay, throughput,
and energy variance. It described that as the number of nodes upsurges the values of all
parameters improvised but in a low pace.
Comparison Results:
The comparison results of two protocols as AOMDV-ER AND AOMDV with our proposed
protocol LBCAR for three parameters as delay, throughput and energy variance are presented
below graphs as well as in tabular form as follows.
Fig. 3: Comparison graph of simulation for throughput among AOMDV, AOMDV-ER AND LBCAR
routing protocols
In fig.3 It is the amount of data successfully received at the destination over a given period
(measured in kbps). Higher throughput indicates better data transmission performance. As the
number of nodes increases, throughput improves for all three protocols. However, LBCAR
consistently achieves higher throughput compared to AOMDV-ER and AOMDV across all
scenarios. LBCAR’s load-balancing and congestion-aware mechanisms help avoid bottlenecks in
the network. By distributing the load evenly among multiple nodes, it ensures smoother data
transmission and reduces packet loss, resulting in higher throughput. AOMDV-ER, with its
energy-aware enhancements, performs slightly better than AOMDV. However, it still lags behind

23
LBCAR, indicating that while energy-awareness helps improve throughput, load balancing has a
greater impact on transmission efficiency.
Table V Comparison results of simulation for Throughput among AOMDV, AOMDV-ER, LBCAR
routing protocols
No. of
nodes.
LBCAR
AOMDV-
ER
AOMDV
20 198 179 170
40 206 192 190
60 218 209 186
80 229 216 208
100 234 224 217
Fig. 4: Comparison graph of simulation for End-to-end delay among AOMDV, AOMDV-ER AND
LBCAR routing protocols
In fig 4. Delay increases as the number of nodes grows, which is typical in networks due to
higher communication overhead and congestion. However, LBCAR maintains lower end-to-end
delays than both AOMDV-ER and AOMDV at all node counts. LBCAR’s congestion-aware
routing reduces the probability of packets being queued or dropped, which helps maintain a
smooth flow of traffic even as the network scales. Additionally, efficient load balancing reduces
delays by preventing overburdening specific nodes. AOMDV-ER performs better than AOMDV,
likely because its energy-aware mechanisms indirectly reduce congestion by selecting more
energy-efficient paths, which avoids overloaded routes.
Table VI Comparison results of simulation for End-to-end Delay among AOMDV, AOMDV-ER, LBCAR
routing protocols
No. of
nodes.
LBCAR
AOMDV-
ER
AOMDV
20 0.10713 0.11623 0.12265
40 0.14291 0.15331 0.17285

24
No. of
nodes.
LBCAR
AOMDV-
ER
AOMDV
60 0.19962 0.20223 0.22241
80 0.23265 0.23431 0.27261
100 0.27112 0.26325 0.31122
Fig. 5: Comparison graph of simulation for energy variance among AOMDV, AOMDV-ER AND LBCAR
routing protocols
In fig 5. As the number of nodes increases, energy variance rises for all protocols, but LBCAR
consistently has the lowest variance. AOMDV shows the highest energy variance, indicating
inefficient energy utilization. LBCAR’s energy-aware routing mechanism ensures that energy
consumption is evenly distributed among nodes, preventing certain nodes from being overused.
This extends the overall network lifetime and reduces the likelihood of node failures. AOMDV-
ER improves upon AOMDV by incorporating some energy-efficient strategies, but it still shows
a higher variance than LBCAR. This suggests that LBCAR’s combination of energy-awareness
with load balancing is more effective at distributing energy consumption.
Table VII Comparison results of simulation for Energy variance among AOMDV, AOMDV-ER, LBCAR
routing protocols
No. of
nodes.
LBCAR
AOMDV-
ER
AOMDV
1000 0.35 0.39 0.58
2000 1.38 2.18 3.95
4000 2.5 5.4 6.9
6000 4.3 8.9 10.82
8000 7.6 12.7 17.62
From the results shown in fig. 3, fig.4 and fig.5 we see that the proposed protocols show better
results in comparison to other two existing protocols in the aspect of all three simulation
parameters. From table V, table VI and table VII following results are drawn from the proposed
work:
The comparison of AOMDV, AOMDV-ER, and LBCAR based on throughput, delay, and energy
variance reveals the following key insights: LBCAR achieves the highest throughput, showing
better data transmission performance. The load-balancing strategy in LBCAR ensures efficient
data routing, reducing packet loss and congestion. LBCAR shows the lowest end-to-end delay,

25
making it ideal for real-time communication scenarios. Its congestion-aware routing ensures fast
delivery times, even as the network scales. LBCAR demonstrates the lowest energy variance,
meaning energy consumption is evenly distributed across nodes. This helps prevent node failures
and extends the overall network lifetime, which is crucial for energy-constrained networks.
6. CONCLUSIONS
In the LBCAR proposed algorithm, the idea of congestion adaptiveness and load balancing is
combined effectively. Through this technique, one can adaptively change the probability of
sending the messages as directed by the dispersion and in route discovery time load status of
nodes. results drawn from the simulation process are compared with AOMDV-ER and original
AOMDV. The proposed work is comparatively significantly enhanced the throughput and
reduced the delay and energy variance with increasing the network lifetime and balancing the
load in the network. Future research on the LBCAR algorithm should focus on optimizing its
performance for larger and heterogeneous networks, exploring the integration of machine
learning for real-time traffic prediction, and evaluating its effectiveness in dynamic and mobile
environments like MANETs and VANETs. Additionally, assessing its scalability in IoT and
smart city networks is essential to ensure its broader applicability. Security and robustness should
also be addressed to protect against malicious nodes and attacks. However, the current study has
some limitations: it is based on simulations that may not reflect real-world complexities,
scalability issues may arise in larger networks, and there has been limited evaluation in highly
mobile environments. The assumption of uniform node characteristics may not hold in
heterogeneous networks, and security vulnerabilities, such as denial-of-service (DoS) attacks,
have not been considered.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
REFERENCES
[1] Patra, S. P. Mohanty, and R. K. Biswal, "Load-Balanced Routing for Congestion Mitigation in IoT
Networks Using Fuzzy Logic," IEEE Internet Things J., vol. 7, no. 5, pp. 4258-4268, May 2020,
doi: 10.1109/JIOT.2020.2974820.
[2] Sharma, P. Patil, and R. Joshi, "Enhanced Load Balancing Routing Protocol for Congestion Control
in MANET," IEEE Trans. Mobile Comput., vol. 19, no. 2, pp. 215-226, Feb. 2020, doi:
10.1109/TMC.2019.2954823.
[3] Alba, E., et al.: A Cellular MOGA for Optimal Broadcasting Strategy in Metropolitan MANETs.
Computer Communications 30(4), 685–697 (2007)
[4] P. Rimal, D. P. Van, and M. Maier, "Load-Balanced Congestion Control in Fog-Cloud Computing
Networks," IEEE Trans. Network Service Manag., vol. 16, no. 1, pp. 219-230, Mar. 2019, doi:
10.1109/TNSM.2018.2877884.
[5] Binitha S, S Siva Sathya. A Survey of Bio inspired Optimization Algorithms. International Journal
of Soft Computing and Engineering (IJSCE), May 2012, ISSN: 2231-2307, Volume-2, I (2).
[6] B. Johnson and S. PalChaudhuri, “Power mode scheduling for Ad Hoc networks,” in Proc.
International Conference on Network Protocols, 2008, vol. 19, no. 5, pp. 192-193.
[7] Koutsonikolas, C.-C. Wang, and Y. C. Hu, “CCACK: Efficient Network Coding Based
Opportunistic Routing through Cumulative Coded Acknowledgments,” in Proceedings of the 29th
IEEE International Conference on Computer Communication (INFOCOM). Piscataway, NJ, USA:
IEEE Press, 2010, pp. 2919–2927.

26
[8] Yuan, X. Liu, and Y. Chen, "Energy-Efficient Load-Balanced Routing for Congestion Control in
MANETs," IEEE Trans. Veh. Technol., vol. 69, no. 8, pp. 8204-8215, Aug. 2020, doi:
10.1109/TVT.2020.3000121.
[9] Falko Dressler, Ozgur B. Akan. Bio-Inspired Networking: From Theory to Practice. IEEE
Communications Magazine, November 2010.
[10] Cervera, M. Barbeau, J. Garcia-Alfaro, and E. Kranakis, “A multipath routing strategy to prevent
flooding disruption attacks in link state routing protocols for manets,” Journal of Network and
Computer Applications, vol. 36, no. 2, pp. 744–755, 2013.
[11] Du, W. Chen, and Y. Zhang, "Adaptive Load-Balanced Routing for Congestion Control in
VANETs," IEEE Trans. Intell. Transp. Syst., vol. 21, no. 4, pp. 1558-1568, Apr. 2020, doi:
10.1109/TITS.2019.2912733.
[12] T. Tran, D. L. Nguyen, and M. S. Kim, "Congestion Control and Load Balancing in Smart Grid
Networks Using SDN," IEEE Access, vol. 8, pp. 152233-152245, Aug. 2020, doi:
10.1109/ACCESS.2020.3018422.
[13] Haseeb K, Abbas N, Saleem MQ, Sheta OE, Awan K, et al. (2019) Correction: RCER: Reliable
Cluster-based Energy-aware Routing protocol for heterogeneous Wireless Sensor Networks. PLOS
ONE 14(10): e0224319.
[14] Lin, X. Wang, and Q. Zhang, "Load-Aware Routing and Congestion Control in Wireless Mesh
Networks," IEEE Trans. Mobile Comput., vol. 19, no. 9, pp. 2142-2153, Sep. 2020, doi:
10.1109/TMC.2020.2969647.
[15] Zhang, G. Tan, and W. Fan, "Load-Balanced Routing Algorithm for Congestion Mitigation in IoT
Networks," IEEE Internet Things J., vol. 7, no. 4, pp. 3271-3282, Apr. 2020, doi:
10.1109/JIOT.2020.2964820.
[16] Jadoon R, Zhou W, Jadoon W, Ahmed Khan. RARZ: Ring-Zone Based Routing Protocol for
Wireless Sensor Networks. Applied Sciences 8: 1023, 2018.
[17] Ji X, Wang A, Li C, Ma C, Peng Y. ANCR—An Adaptive Network Coding Routing Scheme for
WSNs with Different-Success-Rate Links. Applied Sciences 7: 809, 2017.
[18] Singh, R. Kumar, and S. Jain, "Dynamic Load-Balanced Routing Protocol for Congestion Control in
MANETs," IEEE Syst. J., vol. 14, no. 1, pp. 294-305, Mar. 2020, doi:
10.1109/JSYST.2019.2920459.
[19] Miao, F. Ren, C. Lin, and A. Luo. A-ADHOC: An adaptive real time distributed MAC protocol for
vehicular ad hoc networks. Proc. 4th China COM, Xi’an, China. Aug. 2009, pp. 1–6.
[20] Reddeppa Reddy and S. V. Raghavan, “Smort: scalable multipath on-demand routing for mobile ad
hoc networks,” Ad Hoc Networks, vol. 5, no. 2, pp. 162–188, 2007.
[21] L. Zhu, Z. Gao, and X. Li, "Load-Aware Congestion Control in Data Center Networks Using
Machine Learning," IEEE Trans. Cloud Comput., vol. 9, no. 2, pp. 763-774, Apr. 2021, doi:
10.1109/TCC.2020.2970851.
[22] Gawas and M. M. Gawas, “Efficient multi objective cross layer approach for 802.11e over
MANETs,” in Proceeings of the 2018 14th International Wireless Communications Mobile
Computing Conference (IWCMC), pp. 582–587, Tangier, Morocco, June 2018.
[23] M. A. Gawas, L. J. Gudino, and K. R. Anupama, “Cross layer congestion aware multi rate multi
path routing protocol for ad hoc network,” in Proceedings of the 2015 International Conference on
Signal Processing and Communication (ICSC), pp. 88–93, Noida, Uttar Pradesh, March 2015.
[24] M. K. Khan and I. Ahmed, "A Novel Load-Aware Routing Protocol for Congestion Control in
Wireless Mesh Networks," IEEE Access, vol. 8, pp. 159173-159184, Aug. 2020, doi:
10.1109/ACCESS.2020.3018674.
[25] Musaddiq, A., Zikria, Y.B., Zulqarnain et al. Routing protocol for Low-Power and Lossy Networks
for heterogeneous traffic network. J Wireless Com Network 2020, 21,2020.
[26] Liu, Y. Zhang, and G. Cao, "Deep Learning-Based Load-Balanced Routing and Congestion Control
for SDN," IEEE Trans. Network Service Manag., vol. 18, no. 2, pp. 1234-1245, Jun. 2021, doi:
10.1109/TNSM.2021.3065034.
[27] R. Reddy, B. K. Mohanty, and S. K. Tripathy, "Load-Balanced Congestion Control in Mobile Ad
Hoc Networks Using Fuzzy Logic," IEEE Access, vol. 7, pp. 118942-118951, Aug. 2019, doi:
10.1109/ACCESS.2019.2934245.
[28] Rashmi Patil and Rekha Patil,” Cross Layer Based Congestion Free Route Selection in Vehicular
Ad Hoc Networks”, International Journal of Computer Networks & Communications (IJCNC)
Vol.14, No.4, pp 81-98, July 2022, DOI: 10.5121/ijcnc.2022.14405

27
[29] Gupta, N. Suri, and D. S. Kim, "Load Balancing and Congestion Control Using Deep
Reinforcement Learning in SDN," IEEE J. Select. Areas Commun., vol. 38, no. 7, pp. 1455-1465,
Jul. 2020, doi: 10.1109/JSAC.2020.2998502.
[30] Tafazolli and L. Hanzo, “A survey of qos routing solutions for mobile ad hoc networks,” IEEE
Communications Surveys & Tutorials, vol. 9, no. 2, pp. 50–70, 2007.
[31] Arzil, M. H. Aghdam, and M. A. J. Jamali. Adaptive routing protocol for vanets in city
environments uses real-time traffic information. Proc. ICINA, Oct. 2010, vol. 2, pp. 132–136.
[32] G.S.Tomar, L. Shrivastava and S.S.Bhadauria," Load Balanced Congestion Adaptive Routing for
Randomly Distributed Mobile Adhoc Networks," Wireless Pers Commun 2014, DOI
10.1007/s11277-014-1663-9
[33] K. Das, S. Roy, and P. Dutta, "Congestion-Control Aware Load-Balanced Routing in Wireless
Sensor Networks," IEEE Sens. Lett., vol. 3, no. 8, pp. 1-4, Aug. 2019, doi:
10.1109/LSENS.2019.2927834.
[34] S. Lee, K. Kim, and J. Lee, "Hybrid Congestion Control Mechanism in SDN-Enabled Networks,"
IEEE Commun. Lett., vol. 21, no. 5, pp. 1133-1136, May 2017, doi:
10.1109/LCOMM.2017.2655523.
[35] S. Taghizadeh, H. Bobarshad and H. Elbiaze, “CLRPL: Context-Aware and Load Balancing RPL
for Iot Networks Under Heavy and Highly Dynamic Load,” IEEE Access, vol. 6, pp. 23277–23291,
Apr. 2018.
[36] Valarmathi, A., & Chandrasekaran, R. M, (2010, October). Congestion aware and adaptive dynamic
source routing algorithm with load-balancing in MANETs. International Journal of Computer
Applications (0975–8887), 8(5),1–4
[37] X. Chen, H. M. Jones, and A. D. S. Jayalath, “Congestion-aware routing protocol for mobile ad hoc
networks,” in Proceedings of the IEEE 66th Vehicular Technology Conference, pp. 21–25,
Baltimore, MD, USA, October 2007.
[38] X. Li, Y. Lin, and Q. Wang, "Load-Balanced and Congestion-Aware Routing in Hybrid Wireless
Networks," IEEE Trans. Commun., vol. 68, no. 11, pp. 6758-6769, Nov. 2020, doi:
10.1109/TCOMM.2020.3008921.
[39] X. Wu, H. Wang, and S. Li, "Load-Balanced Multipath Routing for Congestion Control in Wireless
Ad Hoc Networks," IEEE Trans. Wireless Commun., vol. 18, no. 12, pp. 5723-5733, Dec. 2019,
doi: 10.1109/TWC.2019.2937593.
[40] Y. He, J. Wang, and L. Ma, "Congestion Control Algorithm for Industrial Wireless Networks Based
on Load Balancing," IEEE Trans. Ind. Informat., vol. 14, no. 9, pp. 4100-4109, Sep. 2018, doi:
10.1109/TII.2018.2808927.
[41] Y. Zhou, X. Yang, and H. Yu, "Adaptive Congestion Control Protocol for Wireless Sensor
Networks," IEEE Sensors J., vol. 18, no. 6, pp. 2539-2548, Mar. 2018, doi:
10.1109/JSEN.2018.2805752.
AUTHOR
Dr.Jogendra Kumar is working as Assistant Professor, Faculty of Computer
Science and Engineering Department, G.B.Pant Institute of Engineering and
Technology Pauri Garhwal Uttarakhand-246194. He has fifteen years of teaching
experience in Engineering, UG and PG level. Her research interest includes Wireless
Networks, IoT, Block Chain Technology, Big Data Analytics, Machine Learning and
WSN. Two Ph.D scholars were pursuing their research under his guidance. He is
also a International Scientific Committee member for Researchers in various
universities. He has received two awards. He has published many research papers, books, book chapters in
SCI, WoS, IEEE, SCOPUS journals. He also published many patents in IPR. He serves as Editor in Book
Chapters, Editorial Board Member and Reviewer in various International Journals. He is an active member
in Professional Bodies like ISTE, IAENG (USA) and IACSIT.

Enhancing Congestion Control using a Load-Balanced Routing Algorithm for Distributed Networks

Recommended

More Related Content

Similar to Enhancing Congestion Control using a Load-Balanced Routing Algorithm for Distributed Networks (20)

More from IJCNCJournal (20)

Recently uploaded (20)

Enhancing Congestion Control using a Load-Balanced Routing Algorithm for Distributed Networks