An Optimized Fitness-Based Enhanced ant Colony Optimization for Optimal Road Selection in Intelligent Heterogeneous Vehicular ad-hoc Networks

Abstract

In Intelligent Heterogeneous Vehicular Ad Hoc Networks (IH-VANETs), long urban roads with a high density of vehicles and a maximum number of road signals increase unpredictable delays in terms of long travel times and heavy traffic congestion. These unpredictable delays are exacerbated by the rapid increase in vehicle density and irregular traffic flow on roads with high traffic signals. To address this gap, an optimized fitness-based enhanced ant colony optimization (OF-EACO) for IH-VANETs is proposed. OF-EACO aims to find optimal, uncongested short roads with low vehicle density and fewer traffic signals, thereby providing shorter travel times for vehicles without traffic congestion and unpredictable delays. To achieve this goal, the novel road fitness function of the proposed OF-EACO assigns a high road fitness score to roads according to short length, low vehicle density, and low signal count to support quick travel of vehicles between two ends without delay and traffic congestion. OF-EACO's roulette wheel takes the road fitness scores of available roads as input and outputs the optimal road. The optimal road is rich in all aspects and is intended to reduce travel time through short and un crowded roads. A network simulator is used to simulate the proposed OF-EACO, existing vehicular multi-hop routing algorithm with intelligent transportation system (VMR-ITS), and improved distance-based ant colony optimization routing (IDBACOR).Simulation results of the proposed OF-EACO indicated that, due to the use of optimal roads, it was able to achieve significant improvements in terms of vehicle travel cost, road establishment time, convergence speed, road traffic congestion overhead, routing overhead, Computational overhead, Computational Complexity, Actual Wall Time Analysis, and Energy Consumption compared to VMR-ITS and IDBACOR.

Keywords

IH-VANETs road length road fitness level road signal count and enhanced ant colony optimization

1. Introduction

IH-VANETs are highly dynamic networks formed by different types of moving vehicles. IH-VANETs facilitate vehicles to communicate with each other and with the Road traffic server(RTS). IH-VANETs integrate and support various vehicular networks, which vary from a vehicle-to-personal device (V2P), vehicle-to-vehicle (V2V), vehicle-to-network (V2N), and vehicle-to-infrastructure (V2I). According to Dubey et al. (2024), the idea of interconnecting V2P, V2V, V2N, and V2I is referred to as Intelligent Transport Systems (ITS). The concept of interconnecting vehicles and RTSs in IH-VANETs frames an instance network to allow vehicles in road networks to exchange information related to accident/road traffic, recreation, and emergency services. Entities of IH-VANETs can use this important information to enjoy different services like exchanging traffic/emergency information and selecting the optimal roads from available roads for travel without congestion and short travel time.

Figure 1 depicts the IH-VANET components and communication types. As depicted in Figure 1, all vehicles in IH-VANETs have an onboard unit (OBU) with hardware capabilities such as a global positioning system (GPS) and human-machine interface devices. GPS coordinates are used for localization. Modern networks are suitable for the adjustment and combination of adaptive features for precise location information. Furthermore, a function called Extended Intersection over Union can be used to improve convergence speed and localization accuracy (Tao et al., 2024). Wireless Access in Vehicular Environments (WAVE) is used for communication between vehicles and RTSs. RTSs are deployed at intersections. The RTS includes network components used for Dedicated Short-Range Communications (DSRC) based radio technology (IEEE 802.11p) (Ramamoorthy & Thangavelu, 2021).

Figure 1.

IH-VANETs components and communication types.

According to recent studies by the World Health Organization, urban road accidents (URTAs) account for 27.5 deaths per 100,000 population. Suggesting and proposing solutions to reduce URTAs is a difficult task because urban roads are more diverse than highway roads with more curves, junctions, and different vehicles with different speed factors (Siddiqi et al., 2022). Adding unpredictable delays and long travel times to journeys introduces higher vehicle travel costs and congestion delays. The idea of using IH-VANETs to reduce URTAs, travel time, road length, and traffic congestion has attracted industry and academia (El Hamdani et al., 2020). In road networks, a journey from one place to another place is not economical if it does not meet the value of travel time (VTT) (Flügel et al., 2022). VTT refers to the travel cost/time taken by vehicles to reach their destination without delay or traffic. Importantly, road length and traffic signals also increase VTT. Longer roads and more traffic signals introduce travel delays. Cost-benefit analyses (CBAs) for transport services require low VTT. An important aspect of CBA is the assessment of VTT to save fuel and enjoy other benefits.

Significantly, extensive research has focused on developing conventional models for Vehicular Ad Hoc Networks (VANETs) to provide shortest path or density–based metrics. But a research gap reveals that they still fail to consider factors like road length, road density, and traffic signal count in determining travel performance to reduce VTT in urban road environments. Road length directly increases VTT and overall delay. Variations in road density introduce unpredictable congestion patterns. Current frameworks have not yet addressed the issue of avoiding unpredictable congestion caused by density fluctuations. Additionally, assessing the fitness of roads should be based on road density, yet this is rarely incorporated into routing algorithms. The presence of multiple traffic signals causes irregular waiting times, leading to inaccurate delay and inefficient road segments. Furthermore, the research gap indicates that existing routing algorithms suffer from making routing decisions based on realistic road conditions such as road length, road density, and traffic signals.

The highlighted research gap motivates us to propose a unique, intelligent, and adaptive approach that systematically integrates road length, density-based fitness, traffic signal calculation, and road travel time variations for road selection. Existing models evaluate these factors separately. The proposed approach enables road selection with realistic and balanced traffic distribution. This integration is performed to select the optimal road under heterogeneous conditions to reduce VTT and traffic congestion. In addition to reducing VTT, it also increases road selection time and scalability, meeting the requirements of next-generation ITS. The research gap prompts us to consider variation in optimal road selection by considering variations in road lengths, density-based fitness, traffic signal count, and road travel time variations (Nithyanandam et al., 2024).

Problems related to optimal road selection led to the proposal of a novel OF-EACO to identify the optimal road for vehicles to travel between two endpoints. An optimal road refers to an efficient, uncongested short-haul road with low vehicle density and low signal count. OF-EACO uses nature-based EACO to suggest optimal roads for drivers to travel. EACO is integrated with modules like road length (RL), road fitness level (RFL), and road signal count (RSC) for optimal road selection.

EACO is a bio-behavioral approach that follows the actions of ants (Ramamoorthy, 2024). EACO follows the analogy of traditional ant colony optimization (ACO). An analogy describes how ants determine the route between a food source and a nest. During exploration, ants reach the required food location by deploying a chemical element such as a pheromone. Furthermore, ants available in the same group follow the same route by absorbing the pheromone to reach the food source. During exploration, ants deposit additional pheromones by estimating evaporation rates (Rokbani et al., 2021). Therefore, the optimal route between the food location and the nest is considered the starting and ending point for depositing high amounts of pheromone. In the proposed OF-EACO, the road identification problem is the main task, and the special behavior of EACO follows the analogy of traditional ACO to find optimal roads.

Concisely, the important contributions of this work are as follows:

Works related to Beacon-oriented approaches, Dual mode listening scheme (DMLS), RL, RFL, and RSC have been investigated and evaluated. The evaluation is done in terms of goals, methodology, scenarios, performance metrics, tools/simulators, and problems being addressed.

ACO-based protocols for optimal road selection for urban networks are studied in terms of design goals, methodologies, pheromone activation types, performance metrics, scenarios, and simulation.

Proposing an optimized fitness-based enhanced ant colony optimization for optimal road selection in IH-VANETs.

Presenting an enhanced ACO to identify optimal non-congested short-length roads according to RL, RFL, and RFS for assisting vehicle drivers to travel.

Investigating the performance of the proposed OF-EACO over IDBACOR and VMR-ITS in terms of vehicle travel cost, road establishment time, convergence speed, road traffic congestion overhead, routing overhead, Computational overhead, Computational Complexity, Actual Wall Time Analysis, and Energy Consumption.

The rest of the manuscript is structured as follows. Section 2 covers related work. Section 3 covers the proposed work. Simulation and results are covered in segment 4. Comprehensive performance analysis is discussed in Section 5. Section 6 covers circumstances affecting the Efficiency of the proposed OF-EACO. Conclusion and future enhancements are covered in segment 7. The final section covers references and required declarations.

2. Related Work

Optimal roads are essential for the timely movement of vehicles. The proposed work relies on beacons to collect information about road characteristics. Therefore, this section covers beacon-based works and other significant works related to road finding. In addition, works related to assessing road suitability by considering road length and road density in terms of vehicle and signal calculations are discussed. Furthermore, ACO-related works for optimal road finding are covered in the related work section.

According to Naderi et al. (2019), Roadside Units (RSUs) can play multiple roles to distribute data/control packets through geographic forwarding, and Arshad et al. (2019) stated that channel capacity, transmission power, and beacon generation rate affect the broadcast rate of beacons. Shakeel et al. (2015) demonstrated that using the concepts of sequence number and time-to-live can avoid spurious beacons and wide propagation of beacons (location, advertisement, and hello). Lin and Deng (2015) demonstrated that the road length can be found using Euclidean distance, and the length of segments can be found by exchanging location coordinates using advertising beacons (Chitra & Sathya, 2018). In Bilal et al. (2014), the road is divided into segments, and the vehicle in the middle of the segment is elected as the Group Leader (GL). The GL can broadcast Segment Density Packet (SDP) to other GLs on the same road to collect data from other segments (Nebbou et al., 2018).

VANET technology (Contreras & Gamess, 2020) relies on beacon propagation to count vehicles and traffic signals; this propagation helps vehicles to learn about each other and maintain a neighbor list. In Arzil et al. (2010), store-carry and forward approaches were followed to overcome local maxima problems, and Rahimi and Jabraeil Jamali (2019) focused on reducing the local maxima problem that occurs when choosing the next hop for packet routing. In Heidari et al. (2025), fuzzy logic based on multi-criteria decision making prioritizes vehicles to eliminate transmission storms and provides efficient data transmission. To enhance scalability and optimize shortest distance calculations, Liu and Meidani (2025a) proposed a framework that supports finding a single shortest path using graph neural networks (GNN). GNN captures spatial relationships and structural evidence in graphs to improve accuracy and efficiency. Furthermore, Liu and Meidani (2025b) use a variant of GNN to solve traffic allocation. Extensive experiments have been conducted to provide accurate estimates of link flows and utilization ratios. The similarities between works 10–19 are the use of beacon propagation, and the differences are the different design goals and the problems being addressed (listed in Table 1).

Table 1.
Summarization of Existing Works Related to Beacon-Based and Other Approaches.

Reference Design Goal Methodology Pheromone Activators Scenarios Parameters used for evaluation Tool/ simulator Problems being addressed by reference

Naderi et al. (2019) To create beacons according to VANET characteristics. Adaptive update strategy – V2V & V2I PDR, E2ED, and OR NS2 Balancing beacon overhead according to VANET characteristics.

Arshad et al. (2019) Malicious data detection during beacon propagation. Beacon trust management and fake data detection. – V2V OR and the Congestion rate ONE Detecting fake beacons.

Shakeel et al. (2015) Proposing Custom built approach to design adaptive beacons. Beacon adaptive scheme – V2V & V2I Beacon delivery delay Mobility Models To prove that the broadcasting rate can influence the beacon delivery rate.

Lin and Deng's (2015) To solve the two-lane placement problem Particle swarm optimization – V2V & V2I Road length and number of runs Mobility Models Minimizing the total cost required to install RSUs using Euclidean distance.

Chitra and Sathya (2018) To find the optimal location of vehicles. Ad hoc localization methods – V2V & V2I Localization Accuracy NS2 Solutions to reduce the impact of GPS errors through ad hoc localization methods.

Bilal et al. (2014) To estimate road density Road-Oriented Traffic Approach – V2V & V2I PDR, E2ED GLOMOSIM Reducing redundant beacons during road density estimation.

Nebbou et al. (2018) To measure road density Intersection-oriented geographical routing protocol – V2V & V2I PDR, E2ED, and OR OMNET++ Managing vehicle densities without causing network congestion.

Contreras and Gamess (2020) Aimed at counting vehicles and road traffic signals. Algorithm for propagating the Beacons – V2V & V2I – Veins Propagating beacons without broadcast storms.

Arzil et al. (2010) To find the path in a sparse network to increase the delivery chance. Adaptive road-based routing – V2V & V2I PDR and EED ONE Inefficiency of classical Routing Mechanisms with Urban Scenarios

Rahimi and Jabraeil Jamali (2019) Aimed at reducing the local maximum problems and reducing the rate of selection of absurd nodes for routing. Hybrid geographic-DTN routing protocol – PDR, EED, and RO NS2 Absence of Intelligent Decision-Making in Proactive Protocols

Reference	Design Goal	Methodology	Pheromone Activators	Scenarios	Parameters used for evaluation	Tool/ simulator	Problems being addressed by reference
Naderi et al. (2019)	To create beacons according to VANET characteristics.	Adaptive update strategy	–	V2V & V2I	PDR, E2ED, and OR	NS2	Balancing beacon overhead according to VANET characteristics.
Arshad et al. (2019)	Malicious data detection during beacon propagation.	Beacon trust management and fake data detection.	–	V2V	OR and the Congestion rate	ONE	Detecting fake beacons.
Shakeel et al. (2015)	Proposing Custom built approach to design adaptive beacons.	Beacon adaptive scheme	–	V2V & V2I	Beacon delivery delay	Mobility Models	To prove that the broadcasting rate can influence the beacon delivery rate.
Lin and Deng's (2015)	To solve the two-lane placement problem	Particle swarm optimization	–	V2V & V2I	Road length and number of runs	Mobility Models	Minimizing the total cost required to install RSUs using Euclidean distance.
Chitra and Sathya (2018)	To find the optimal location of vehicles.	Ad hoc localization methods	–	V2V & V2I	Localization Accuracy	NS2	Solutions to reduce the impact of GPS errors through ad hoc localization methods.
Bilal et al. (2014)	To estimate road density	Road-Oriented Traffic Approach	–	V2V & V2I	PDR, E2ED	GLOMOSIM	Reducing redundant beacons during road density estimation.
Nebbou et al. (2018)	To measure road density	Intersection-oriented geographical routing protocol	–	V2V & V2I	PDR, E2ED, and OR	OMNET++	Managing vehicle densities without causing network congestion.
Contreras and Gamess (2020)	Aimed at counting vehicles and road traffic signals.	Algorithm for propagating the Beacons	–	V2V & V2I	–	Veins	Propagating beacons without broadcast storms.
Arzil et al. (2010)	To find the path in a sparse network to increase the delivery chance.	Adaptive road-based routing	–	V2V & V2I	PDR and EED	ONE	Inefficiency of classical Routing Mechanisms with Urban Scenarios
Rahimi and Jabraeil Jamali (2019)	Aimed at reducing the local maximum problems and reducing the rate of selection of absurd nodes for routing.	Hybrid geographic-DTN routing protocol	–		PDR, EED, and RO	NS2	Absence of Intelligent Decision-Making in Proactive Protocols

As a biological entity, the hunting process of ACO allows ants to navigate towards a food source by laying down pheromones along their path. Similarly, the ants can return to the nest via the same route by adjusting the pheromone levels based on the rate of evaporation. This highly pheromone-laden path represents the optimal route in terms of short distance. This analogy is followed in all sources (Bijalwan et al., 2023; Calabrò et al., 2020; Cheng, 2023; Choudhary & Pahuja, 2023; Fatemidokht & Kuchaki Rafsanjani, 2018; Ibrahim et al., 2022; Lee et al., 2024; Lekshmi & Pramila, 2024; Li et al., 2019; Majumdar et al., 2016; Nandagopal et al., 2023; Nemade & Pujeri, 2023; Oonsrikaw & Thammano, 2018; Phuc & Thao, 2021; Ramamoorthy et al., 2024; Ramamoorthy & Thangavelu, 2022a, 2022b, 2022c; Sbayti & Housni, 2024; Suganyadevi & Nandalal, 2024; Tariuge & Matthew, 2024; Wang et al., 2019; Wang et al., 2022; Wu et al., 2023) to present optimal solutions concerning routing and other factors.

In Majumdar et al. (2016), the probability function of ACO depends on vehicle density and road length for route expansion. ACO quickly selects the optimal routes. In Sbayti and Housni (2024), Optimized Link State Routing (OLSR) and ACO find routes using an intelligent route selection method. Optimal metrics such as distance and mobility are considered for route selection. Bijalwan et al. (2023) proposed an improved ACO to reach the optimal solution with minimum iterations. The pheromone update rule and the evaporation factor improve ACO convergence speed to overcome local and global optima problems. Phuc and Thao (2021) added practical constraints with ACO to solve vehicle routing problems. The route constructed by the ants is recorded in the solution set. From the solution set, the optimal route is selected based on the pheromone intensity. The optimal route is time-specific and short in length. In Ibrahim et al. (2022), improved ant colony optimization (IACO) relies on local search and mutation operations to solve the vehicle routing problem. Local update rules are followed to deposit the strongest intensity of pheromones on the optimal roads. A global update rule is implemented each time to maintain the strongest intensity on the optimal routes.

In Oonsrikaw and Thammano (2018), the pheromone is updated according to the ratio between the length of the path and the value of the constant. A weight is assigned to each path based on the feedback from the strongest ants. The ACO-based dynamic path optimization algorithm (Cheng, 2023) calculates a weighting factor for each road based on the road length and estimated travel time. Based on the weighting factor, the optimal road is selected. The dynamic optimization makes the routing mechanism loop-free with an optimal convergence speed. IACO (Li et al., 2019) relies on innovative schemes to update pheromones on optimal solutions to deliver better results for reducing vehicle cost and time in vehicle routing problems. Furthermore, IACO (Wang et al., 2022) is improved by adding the Floyd algorithm. The Floyd algorithm applies a fallback strategy to improve the performance of the heuristic function. Also, an optimization-based multi-objective function is proposed, and energy consumption, length, and safety are considered for route formation.

Calabrò et al. (2020) applied ACO to identify the optimal road with the help of an objective function by considering travel distance and hop count. Similarly, a hybrid ant colony (HACO) (Wu et al., 2023) with a mutation operation can find the optimal road. This work adds three inventions to ACO, which include a new pheromone rule, adaptive metrics, and a mutation operator. In Ramamoorthy and Thangavelu (2022a), ACO is modified to add a source-based routing concept to guide ants in finding a road with optimal distance and hops. The highest amount of pheromone in terms of summation is considered to select the path. In Lekshmi and Pramila (2024), the same source-based ACO is enhanced with a priority scheduling concept for message dissemination. Authors classify messages to assign priority values to avoid the stagnation problem.

In Nemade and Pujeri (2023), ACO helps ambulance drivers to choose the route with minimum congestion according to the fitness value. The fitness value is calculated according to the network congestion and vehicle travel speed. In Fatemidokht and Kuchaki Rafsanjani (2018), ACO applies fuzzy rules to select the route according to the signal strength, congestion level, and speed. The pheromone intensity for optimal channels depends on fitness value. A hybrid (ACO and artificial bee colony optimization (ABC)) algorithm (Nandagopal et al., 2023) computes fitness degree according to link stability and residual strength. To provide quick communication, the network is divided into zones to apply proactive and reactive approaches. Different fitness degrees are made for proactive and reactive schemes. However, the hybrid optimization algorithm (Choudhary & Pahuja, 2023) is improved with an aware routing scheme to provide reliable routing for VANETs. In addition to link stability and residual strength, vehicle mobility is considered for route formation. ACO is configured with an ant colony operator to avoid native traps.

In hybrid ACO with Spider Monkey Optimization (SPO) (Suganyadevi & Nandalal, 2024), distance and traffic-based principles are used as metrics in the path-finding operations. Vehicle direction is considered to measure distance, and road inflow/outflow is considered to estimate traffic. Similarly, in the hybrid bio-inspired zone routing scheme (Ramamoorthy et al., 2023), path-finding and route-carrying ants are used in path-finding operations. Route Strength Factor (RSF) determines the suitability of a route according to the pheromone density. In Lee et al. (2024), ACO-based traffic-aware routing introduces a distributed mechanism for traffic detection and path-finding operations. A two-hop path finding mechanism is used to estimate traffic and road suitability.

According to Wang et al. (2019), a hybrid ant colony-artificial fish optimization algorithm improves the convergence speed by iterating quickly to avoid local optima. A strong positive feedback system improves the pheromone rule to produce optimal results. Similarly, ACO (Ramamoorthy & Thangavelu, 2022b) is configured to calculate the path strength factor to improve the algorithm speed with fewer iterations. In congestion-aware ACO (Ramamoorthy & Thangavelu, 2022c), node-level congestion and distance are used as metrics to find non-congested optimal roads with minimum iterations. Pheromone patterns continuously control pheromones to prevent inefficient nodes from participating in routing. Furthermore, ACO (Tariuge & Matthew, 2024) has been improved and is used to guide navigation systems. This work aims to control traffic congestion by informing drivers about traffic conditions (Ramamoorthy et al., 2024).

Table 1 summarizes the existing works related to beacon-based and other approaches. Table 2 summarizes the existing works related to ACO-based approaches. The similarities between works 23–46 are the use of ACO approaches, and the differences are the different design goals and the problems being addressed (listed in Table 2).

Table 2.

Summarization of Existing Works Related to the ACO-Based Approaches.

Reference	Design Goal	Methodology	Pheromone Activators	Scenarios	Parameters used for evaluation	Tool/ simulator	Problems being addressed by reference
Majumdar et al. (2016)	To establish an efficient path.	Probabilistic ACO	Artificial ants	V2V	Throughput and PLR	MATLAB	Adaptability of ACO to deal with dynamic networks.
Sbayti and Housni (2024)	To optimize the route selection process.	ACO	Artificial ants	V2V & V2I	Throughput, PDR, E2ED, and OR	NS3	By using ACO, enhancing path stability and compatibility
Bijalwan et al. (2023)	To establish a scalable and stable VANET topology with optimal convergence speed.	Enhanced ACO framework	Traversing ants	V2V & V2I	Throughput, PDR, E2ED, and OR	NS2	Local and global optima problems are addressed.
Phuc and Thao (2021)	To design optimal routes to reduce total costs like travel time.	ACO	Artificial ants	V2V & V2I	Total distance and Total time	Python	Pheromone influence factors are addressed.
Ibrahim et al. (2022)	To minimize total vehicle mileage in terms of distance.	IACO	Local search ants	V2V	Count of iterations to number of routes and total distance	MATLAB	Optimizing delivery routes while adhering to customer time requirements.
Oonsrikaw and Thammano (2018)	To minimize local optima problem	ACO	Strong ants and Weak ants	V2V	Optimal solutions	MATLAB	Using a Local search approach to avoid sub-optimal solutions
Cheng (2023)	To find the optimal road	Dynamic ACO	Forward and Backward Ants	V2V & V2I	Road length and Travel time	MATLAB	Dynamic route optimization through refining the ACO
Li et al. (2019)	To save vehicle travel with the Shortest path	IACO with Ant-Weight Strategy	nAnts	V2V & V2I	Cost, time, and emission	Qualitek-4	Multi-depot vehicle routing issue
Wang et al. (2022)	To improve the convergence time	IACO	Configured ants	V2V & V2I	Convergence speed	MATLAB	Adaptability and robustness of ACO
Calabrò et al. (2020)	To find optimal routes	agent-based ACO	Artificial Ants	V2V	Travel distance, Load factor, and vehicle density	NetLogo	Optimization of Vehicle Paths
Wu et al. (2023)	To find the optimal solution within a time window	Hybrid ACO	Artificial Ants	V2V	Convergence speed	TSP datasets	Parameter sensitivity to fine-tune pheromone models
Ramamoorthy and Thangavelu (2022a)	To find the optimal route with minimum hops.	Source-based ACO	Forward ants and Backward ants	V2V & V2I	Throughput, PLR, PDR, E2ED, and OR	NS2	The inefficiency is associated with using long routes with maximum hops.
Lekshmi and Pramila (2024)	To find the optimal route with minimum hops without congestion	Source-based ACO	Forward ants and Backward ants	V2V & V2I	Throughput, PLR, PDR, E2ED, and OR	NS2	The inefficiency is associated with source-based routing.
Nemade and Pujeri (2023)	To find the path with low congestion and delay	ACO	Forward ants and Backward ants	V2V & V2I	Throughput, PDR, E2ED, and OR	NS2	Optimizing Routing Paths through a fitness function
Fatemidokht and Kuchaki Rafsanjani (2018)	To find a short and less congested route	ACO with Fuzzy logic	Forward and Backward ants	V2V & V2I	Throughput, PLR, PDR, E2ED, and OR	NS2	Fuzzy-based route optimization.
Nandagopal et al. (2023)	To find optimal & Stable routes	ACO and ABC	Configured ants	V2V & V2I	Throughput, Delay time, PDR, and OR	NS2	The compatibility of reactive approaches for route selection has been addressed.
Choudhary and Pahuja (2023)	To find optimal & Stable routes	ACO with awareness routing	Configured ants	V2V & V2I	E2ED, OR, and PDR	NS2	Suggests solutions to avoid local traps.
Suganyadevi and Nandalal (2024)	To find the quickest path	ACO with SPO	Artificial ants	V2V & V2I	OR, Throughput, Packet failure	NS2	Pheromone influence factors are addressed.
Ramamoorthy et al. (2023)	To find the shortest path	Hybrid ACO	Route Finding and carry ants	V2V & V2I	Throughput, PLR, PDR, E2ED, and OR	NS2	The inefficiency is associated with using long routes.
Lee et al. (2024)	To find the shortest path	ACO with a distributed mechanism	Configured Ants	V2V & V2I	PDR,E2ED, and OR	NS3	Traffic awareness routing decisions addressed
Wang et al. (2019)	To improve convergence speed	Hybrid ACO	Artificial ants	V2V & V2I	Convergence speed	NetLogo	The suitability of ACO hybridization for making routing decisions has been addressed.
Ramamoorthy and Thangavelu (2022b)	To find a lifelong route	ACO	route request and Route Reply ants	V2V & V2I	PDR, Throughput, and Latency	NS2	– The inefficiency is associated with using long routes.
Ramamoorthy and Thangavelu (2022c)	To find an optimal path without congestion	Congestion-aware ACO	route request and Route Reply ants	V2V & V2I	Throughput, PLR, PDR, and E2ED	NS2	Inefficiency associated with using long, congested routes
Tariuge and Matthew (2024)	To address the vehicle routing problem	ACO	Artificial ants	V2V & V2I	–	–	Adaptability and robustness of ACO

In Tables 1 and 2, Packet delivery rate/ratio (PDR), end-to-end delay (E2ED), Routing overhead (RO), Packet loss rate (PLR), Delay Tolerant Network(DTN), Traveling Salesman Problem (TSP), Opportunistic network emulator (ONE), Global Mobile system Simulator (GLOMOSIM), Objective Modular Network Testbed (OMNET), Vehicle in Network Simulation (Veins), Matrix laboratory (MATLAB), and Network simulator (NS2/3).

3. Proposed Model

This work proposes OF-EACO to find the optimal road for vehicles to travel between two ends. The first step of OF-EACO covers the DMLS, which allows vehicles to register with the RTS. In DMLS, the RTS enables vehicles in its vicinity to register with it for an Advertisement Beacon (AB). Vehicles under RTS respond to AB with a Registration Beacon (RB) to become legitimate members of RTS. AB is designed by RTS for advertising to indicate its presence. A vehicle under the communication area of RTS receives AB and responds with RB to become a valid member of RTS. RB and AB play important roles in grouping vehicles under RTS. The status of the vehicle count under each RTS is maintained to provide an up-to-date vehicle count. For each time interval “t” (t = 10 s), the RTS and vehicle routing table are updated with the latest count of vehicles. The role of DMLS is crucial, such as bringing each vehicle under RTS and maintaining vehicle count and road signal count to assist EACO for optimal road selection.

The second step of OF-EACO is to find the road length (RL) from the network “ $N$ ”. A network N consists of several roads to allow vehicles $V_{X}$ to travel between two points, such as the starting point $S 1$ and endpoint $E 1$ . RL indicates road length between $S 1$ and $E 1$ . The location of the start point $S 1$ and end point $E 1$ can be obtained in the form of GPS coordinates. The third step starts by running the Road Fitness Level (RFL) model to find the fitness level for the available roads. For RFL, road density level (RDL) is required, and RDL is estimated for each road based on permissible vehicles for the maximum vehicles allowed at a time “ $t$ ” on road “ $R$ ”. The maximum number of vehicles allowed on the road is set to 200, and the number of available/presented vehicles is extracted from the counter field (CF_RTS) in the RTS routing table. RFL metric specifies whether the road is fit or not as “1” or “0”. A value of “1” indicates that the road is adequate and has sufficient capacity for vehicles to travel without delays and congestion. A value of “0” indicates that the road is inadequate and has insufficient capacity for vehicles to travel without congestion and delays. In the proposed model, non-delayed and non-congested roads are considered for travel.

The fourth step begins with the calculation of the road signal count on each road. RTS performs a count of road signals in its vicinity. Road signal (RS) advertises a hello beacon (HB) message for every time interval “ $t$ ” to register with RTS. “ $t$ ” is set to 200 s. Beacon messages of the RSs are used to detect road signal units. The RSC register in RTS maintains a count of road signals in its vicinity. Finally, the enhanced ant colony optimization (EACO) of optimized OF-EACO is used to find the optimal road based on RL, RFL, RSC, and road fitness score (RFS). EACO's pheromone deposit procedure credits pheromone to each road as per RL, RFL, and RSC. Similarly, EACO's pheromone update mechanism updates the pheromone for each road according to RL, RFL, RSC, and a dynamic parameter called evaporation rate to preserve the optimal amount of pheromone on the optimal roads. EACO calculates a road fitness score for each road according to the pheromone levels of the roads. The road with a short length, high fitness level, with low road signal count is assigned with optimal/high pheromone values via the pheromone update model. Furthermore, EACO's optimal road selection model selects the optimal road for vehicles to travel according to the road fitness score. The road with a high fitness score is rich in short length, low vehicle density, and low signal count for supporting quick travel of vehicles between the two ends without delay. The roulette wheel helps the road selection pattern for proper road selection.

Figure 2 presents the modules of OF-EACO. In OF-EACO, vehicles are randomly placed, and RTS are kept constant. The DMLS system allows vehicles to register with RTS. RL determines the length of each road. RFL finds the fitness level for available roads. The road signal calculation module calculates the road signal count for each road. Then, EACO starts operations with a pheromone deposit and updates procedures for crediting and updating pheromones on the “R” road based on RL, RFL, RSC, and evaporation rate. Based on the pheromone intensity, a road fitness score is calculated for the road to allow the optimal road selection model (ORSM) to select the optimal road “R” from the available roads. A roulette wheel helps ORSM to choose the road with the highest fitness score. A road with a high fitness score is shorter in length, with lower traffic/vehicle density and lower signal count.

Figure 2.

Architecture of OF-EACO.

3.1 Dual Mode Listening Scheme (DMLS)

In DMLS, the vehicle and RTS use beacons to communicate with each other. Each vehicle can register with RTS for AB to become a legitimate member under RTS. RTS considers the location of vehicles to agree vehicles as their legitimate members. For a vehicle to be recognized as a legitimate member of the RTS, it must be within the communication range of the RTS. Vehicles and RTS are equipped with GPS to transmit their locations in the form of X and Y coordinates.

In a road network, each RTS consists number of vehicles under its communication area. In DMLS, RTS can broadcast ABs within its communication area with a time interval of t. The RTS broadcasts AB, which includes the road traffic server ID ( $R T S_{I D}$ ), RTS communication range ( $R T S_{r a n g e}$ ), road traffic server location in terms of X and Y coordinates, and Advertisement Beacon time to live (AB_TTL). AB_TTL decides the lifetime of the AB. The AB is used to inform the presence of RTS to vehicles that are located under its communication area. The vehicles under RTS receive AB and reply with RB for each AB. In DMLS, vehicles share their details with its location with their RTS using RBs for registering with RTS and also to become legitimate members of RTS. The RB includes vehicle ID ( $V_{I D}$ ), vehicle one-time sequence number ( $V O T_{S N O}$ ), and vehicle current location ( $V_{L o c a t i o n (L o c a t i o n (X, Y C o o r d i n a t e s)}$ ) in terms of X and Y coordinates and Registration Beacon -time to live (RB_TTL). RB_TTL decides the lifetime of the RB. In this work, the time interval for AB_TTL and RB_TTL is set to 10 s (t = 10 s).

For every time interval t, the routing table of RTS and vehicles is updated with fresh vehicles and RTS. Significantly, to avoid double registration with more than one RTS, the vehicles discard ABs by comparing $R T S_{I D}$ of received AB with stored $R T S_{I D}$ in its routing table. Every vehicle stores $R T S_{I D}$ for first-time processing. Once a vehicle is registered with any RTS, then the vehicle will not accept any other AB from another RTS until it gets out of the communication range of its RTS. When a vehicle is out of the validity of RTS, the stored $R T S_{I D}$ is erased. The vehicle will get a new $R T S_{I D}$ when it enters a new RTS range. While receiving RB, the RTS only processes RB with new $V O T_{S N O}$ . Whenever RTS receives RB first time, the $V O T_{S N O}$ is extracted and checks its routing table for similarity with stored $V O T_{S N O}$ . In RTS, the $V O T_{S N O}$ of received RB is compared with the stored $V O T_{S N O}$ of its routing table for discarding already processed beacons.

Every RTS has its covered communication range (CCR), and CCR represents $R T S_{r a n g e}$ . $R T S_{r a n g e}$ is propagated via AB's. Once a vehicle is out of its registered RTS, then the same vehicle receives AB to get registered with another RTS. Routing tables of RTS and vehicles get refreshed at every time interval t to remove out-off identities of vehicles and RTS. The Counter field (CF_RTS) of the routing table of RTS maintains the total number of vehicles in its vicinity. Figure 3 shows the AB structure, and Figure 4 shows the structure of RB. Figure 5 shows the routing table of each vehicle, and Figure 6 shows the routing table of RTS.

Figure 3.

Advertisement beacon.

Figure 4.

Registration beacon.

Figure 5.

Vehicle routing table.

Figure 6.

RTS routing table.

In Figure 3, $R T S_{I D}$ represents the road traffic server ID, $R T S_{r a n g e}$ is the road traffic server range, $R T S_{L o c a t i o n (X, Y C o o r d i n a t e s)}$ is GPS coordinates of RTS, and AB_TTL is the lifetime of AB. In Figures 4 and 5, $V_{I D}$ indicate the vehicle's unique identity. $V O T_{S N O s (1 \dots n)}$ is the vehicle one one-time sequence numbers generated for each RB, $V_{L o c a t i o n (X, Y C o o r d i n a t e s)}$ is the current location of the vehicle, $R T S_{r a n g e}$ is the communication range of RTS, $R T S_{I D}$ is the identity of RTS. In Figure 6, $R T S_{I D}$ is the identity of the RTS, $R T S_{L o c a t i o n (X, Y C o o r d i n a t e s)}$ indicates the current location of RTS, $V_{I D (1 \dots n)}$ indicates the identities of vehicles under its communication range, $V O T_{S N O s (1 \dots n)}$ is the stored vehicle one-time sequence numbers, CF_RTS represents the total number of vehicles enrolled under its $R T S_{r a n g e}$ , and Road signal count (RSC) indicates the count of road signals. The developed DMLS algorithm is depicted in Algorithm 1. Algorithm 1 shows the step-by-step activities of DMLS.

3.2. Road Length (RL)

The RL indicates the total length of the road segment between the starting point and the end of the vehicle. The location of the start point and end point can be obtained in the form of GPS coordinates. In this work, RL can be calculated according to Equation 2. Equation 1 shows the road segment. $R_{S}^{T}$ indicates a road segment with the starting point and endpoint of a vehicle for travel.

\begin{aligned} R_{S}^{T} & = {{V e h i c l e}_{s t a r t i n g p o int}, {V e h i c l e}_{D e s t i n a t i o n P o int}} \end{aligned}

(1)

\begin{aligned} | R_{L} | & = \sqrt{R_{S}^{T} {(X_{e} - X_{s})}^{2} - R_{S}^{T} {(Y_{e} - Y_{s})}^{2}} \end{aligned}

(2)

where

| R_{L} |

is the Road length between the start and end points of the road L.

R_{S}^{T} (X_{e} - X_{s})^{2}

is the X coordinates of the end location and starting location.

R_{S}^{T} (Y_{e} - Y_{s})^{2}

is the Y coordinates of the end location and starting location.

3.3. Road Fitness Level (RFL)

Let R be the road that belongs to the network $N (R \in N)$ . R is efficient for allowing vehicles to travel when it is equipped with a minimum number of vehicles. Road density indicates the number of vehicles present on the road R.The RFL is determined based on RDL. The road R with the minimum vehicles imposes less traffic and less congestion. Roads with low congestion are optimal for travel without delay. The R with the maximum number of vehicles can cause congestion and introduce delays in reaching the destination. In this context, RDL is a significant factor in finding RFL (Subramaniam et al., 2022). To find RDL, equation 3 can be used.

\begin{aligned} {R D L}_{t}^{R} = \frac{N_{V}^{t} i n R o a d R}{{M N}_{V}^{t} i n R o a d R} \end{aligned}

(3)

${R D L}_{t}^{R}$ is the road density level for time t. $N_{V}^{t}$ is the total number of vehicles on the road R for time t. $M N_{V}^{t}$ is the maximum number of vehicles permitted on the road $R$ for time t. In this, $M N_{V}^{t}$ is set with 200 vehicles for a time t. $N_{V}^{t}$ is extracted from the Counter field (CF_RTS) of the RTS routing table. Initially, the CF_RTS is set to zero and updated for every new joining of the vehicle under RTS. Under DMLS, the CF_RTS of RTS is updated on the time scale of 10sec (t = 10 s). For the road R, the RFL is described as (Equation 4):

\begin{aligned} {R F L}_{t}^{R} = {\begin{cases} 1 & i f R D L_{t}^{R} < 1 \\ 0 & i f R D L_{t}^{R} \geq 1 \end{cases} \end{aligned}

(4)

here

{R F L}_{t}^{R}

is the road fitness level of

K^{t h}

road for a time

^{'} t^{'}

, if

R D L_{t}^{R}

is less than one return value is “1”, or if

R D L_{t}^{R}

is greater than or equal to “1” the return value is “0”.

R D L_{t}^{R}

Values are varied in terms of 0 or 1. The Boolean concept of Boolean logic finds 0 or 1. “1” represents the road “R” is equipped with a minimum vehicles and suitability for travel without delay and congestion (less traffic). “0” indicates that road “R” has the maximum number of vehicles and is not suitable or optimal for travel. The road “R” with congestion may introduce a travel delay, and vehicles take more time to reach their destination. The values “0” or “1” are assigned by using equations 3 and 4.

3.4. Road Signal Count (RSC)

The road signals are operated under RTS. The RSC represents the number of road signals available for road “R”. RSs are deployed at junctions of each road/pedestrian crossing place. RS propagates an HB for every time interval t to enroll with RTS. The t is set to 200 s. Figure 7 shows the structure of the hello beacon. RTS maintains an up-to-date count of RSC. The RSC register in RTS maintains a count of road signals in its vicinity.

Figure 7.

Hello beacon.

In Figure 7, $R S_{I D}$ is the unique identity of the road signal. $R S O T_{S N O}$ is the road signal one-time sequence numbers generated for each HB, $R S_{L o c a t i o n (X, Y C o o r d i n a t e s)}$ is the current location of the road signal, and HB_TTL is the hello beacon time to live field, which decides the lifetime of the hello beacon. Here, TTL is set to 10 s. Initially, the RSC value is set to zero for every t. Based on registration with the RTS value of the RSC is incremented/ decremented. Table 8 shows the algorithm for road signal registration with RTS. Algorithm 2 shows the step-by-step operations used by RS to register in RTS.

Figure 8 illustrates the flowchart of the proposed OF-EACO. This flowchart details the step-by-step operations carried out by the proposed OF-EACO to identify an optimal road. To facilitate operations and initiate the proposed model, essential parameters are needed. These parameters include three types: network topology inputs, vehicle dynamics, and EACO hyper parameters. Network topology inputs and vehicle dynamics are essential for performing DMLS and calculating RL, RFL, and RSC. EACO hyper parameters are necessary to identify the optimal roads. The process begins with network topology inputs. For this, OpenStreetMap data was downloaded to include roads, vehicles, speed limits, RTS, and RSC. Next, vehicle mobility traces were generated through an urban mobility simulation with traffic density profiles, including the number of vehicles and speed distribution. Finally, EACO hyper parameters—such as colony size, initial pheromone values, evaporation rate, pheromone patterns, and fitness function—require a call to ORSM to determine the optimal roads.

Figure 8.

Flowchart of OF-EACO.

3.5. Enhanced ant Colony Optimization (EACO)

EACO is a successful strand of swarm intelligence (SI). SI aims to build intelligent systems by simulating the cooperative behavior of insects such as ants and bees (Jindal & Bedi, 2018; Thilak & Amuthan, 2018). EACO is an upgraded bio-inspired system activated by the natural activity of ants. Ant foraging activity demonstrates that ants can discover the optimal lane between origin and target in the absence of visual cues. EACO's configured ants can invent the shortest path between two points with less complexity. EACO deposits the pheromone in the direction the ants travel. In EACO, search ants use pheromones as a communication medium for indirect communication without visual cues. The search ants navigate to food sources using the maximum pheromone concentration set by preceding ants. Similarly, the same search ants navigate from food source to nest by following the same route, depositing specific amounts of pheromones according to evaporation rate. Ants are often attracted to high quantities of pheromones, which allow them to choose the shortest path with a short time delay. EACO’s dynamic qualities, such as a basic structure, route selection capabilities, self-organization, and distributed administration control, make it adaptable to solving VANET routing challenges.

In the upgraded EACO, a set of programmable software agents, such as search ants, find an optimal solution to a given optimization problem. To implement EACO, the discovered problem is turned into a weighted network to identify a more efficient solution. Search ants solve problems by traveling forward in a weighted network. The solution construction activity is stochastic and is impacted by factors such as RL, RFL, RSC, and the evaporation rate associated with roads whose values are regularly adjusted by search ants at runtime. In the proposed framework, the roulette wheel selects the high-fitness score road with the maximum pheromone density for quick travel without delay. The pheromone of the road is computed based on RL, RFL, and RSC. Search ants in EACO travel toward the target in a forward way based on road parameters such as RL, RFL, and RSC. For optimal road selection, EACO includes an objective function. Equation 5 denotes the OF-EACO's objective function. An important aim of the objective function of OF-EACO is to find a high-fitness score road between the origin and the destination (Ramamoorthy et al., 2024). A road with a high fitness score value is an optimal road with a short distance, low density, and low signal counts.

\begin{aligned} P_{R_{1}, R_{2}, R_{3} \dots, R_{n}}^{Z} = {\begin{cases} \frac{{R F S}_{R}^{t}}{\sum_{N = 1}^{N} {R F S}_{R}^{t}} & i f R \in N \\ 0 & o t h e r w i s e \end{cases} \end{aligned}

(5)

In equation 5, $P_{R_{1}, R_{2}, R_{3} \dots, R_{n}}^{Z}$ indicates the $Z^{t h}$ ant probability rate for picking the road R among the available roads $R_{1}, R_{2}, R_{3} \dots, R_{n}$ . ${R F S}_{R}^{t}$ is the fitness of the individual road R and $\sum_{N = 1}^{N} {R F S}_{R}^{t}$ is the summation of ${R F S}_{R}^{t}$ among $R_{1}, R_{2}, R_{3} \dots, R_{n}$ . The ${R F S}_{R}^{t}$ value is calculated according to equation 8. N is the network of roads, and $Z^{t h}$ ant can select a road according to the fitness function. Algorithm 3 illustrates the sequential tasks executed by the proposed OF-EACO.

3.5.1 Pheromone Deposit Model

Pheromone concentration determines the suitability and optimality of a route. Consider there is a road R with a start point $S 1$ and an endpoint $E 1$ .While considering $S 1$ and $E 1$ of R, search ants deposit an assured volume of pheromone on R. The pheromone deposit model credits a certain quantity of pheromone on the road R according to its $R L, R F L,$ and $R S C$ . This model assists search ants to maintain pheromone on R based on $R L, R F L,$ and $R S C$ . The pheromone deposit model can also be considered a positive feedback model that motivates search ants to quickly and efficiently identify optimal solutions. A high amount of pheromone is deposited on short $R L$ , low $R S C$ , and high $R F L$ . Equation 6 denotes the pheromone deposit model.

\begin{aligned} {Δ τ}_{R}^{N} (t) = (\frac{1}{{R L}_{R}^{t}} + {R F L}_{R}^{t} + (\frac{{R S C}_{R}^{t}}{100})) \div 10 \end{aligned}

(6)

In equation 6, R is the road in network N. ${Δ τ}_{R}^{N} (t)$ is the pheromone deposited by the $M^{t h}$ ant on the road R of network N at time $" t "$ . ${R L}_{R}^{t}$ is the length of the road R at time $" t "$ . ${R F L}_{R}^{t}$ is the fitness level of the road R at time $" t "$ . ${R S C}_{R}^{t}$ is the road signal count at time $" t "$ for the road R.

3.5.2 Pheromone Update Model

The objective of the pheromone update model is to update the pheromone on the roads. The evaporation process reduces the pheromone concentration on the roads over time. Remarkably, this model updates the pheromone on the roads for every time interval $" t "$ based on the evaporation rate $(ρ)$ , $R L$ , $R F L$ , and $R S C$ . This model helps EACO to avoid local optima problems and also improves the convergence speed of the algorithm. The rate of $ρ$ varies according to the time $" t "$ . Equation 7 depicts the pheromone update model.

\begin{aligned} τ_{R}^{N} (t) = (\frac{1}{{R L}_{R}^{t}} + (O_{t} - ρ_{t}) + {R F L}_{R}^{t} + (\frac{{R S C}_{R}^{t}}{100})) \div 10 \end{aligned}

(7)

where

τ_{R}^{N} (t)

is the updated pheromone on the road R at the time

" t "

for the network N.

{R L}_{R}^{t}

is the length of the road R at time

" t "

O_{t}

represents old pheromone value at time

" t "

for road R.

ρ_{t}

is the pheromone evaporation rate at time

" t "

for road R .

{R F L}_{R}^{t}

is the fitness level of road R at time

" t "

{R S C}_{R}^{t}

is the road signal count at time

" t "

for road R. In the proposed model, the rate of

^{'} ρ^{'}

is set to be dynamic. The dynamic values vary from 0.01 to 0.05. Three types of scenarios are considered to demonstrate the proposed model. Scenario 1 finds the optimal road with no evaporation rate (

ρ

= 0). Scenario 2 finds the optimal path with a constant evaporation rate of 0.02 (

ρ

= 0.02). Scenario 3 has different evaporation rate variants such as 0.02, 0.03, 0.04, and 0.05 (

ρ

= 0.02, 0.03, 0.04, and 0.05). The computed values for pheromones are listed in Tables 3, 4, and 5. Tables 3, 4, and 5 depict the pheromones credited, updated, and optimal path selection using the roulette wheel for all scenarios.

Table 3.

Distribution of Pheromone Values ( $ρ$ = 0), Road Fitness Scores, and the Roulette Wheel for Scenario 1.

Table 4.

Distribution of Pheromone Values ( $ρ$ = 0.02), Road Fitness Scores, and the Roulette Wheel for Scenario 2.

Table 5.

Distribution of Pheromone Values (ρ = 0.04,0.03,0.05,0.01,and 0.02), Road Fitness Scores, and the Roulette Wheel for Scenario 3.

Figure 9 presents the pheromone intensity levels and fitness scores assigned to each road according to the pheromone models. Equations 6 and 7 represent the formulas for pheromone models used to assign pheromones. Figure 9 shows the pheromone values and road fitness scores assigned to each road according to the starting and ending points of the vehicles. Tables 3, 4, and 5 show the detailed distribution of pheromone values, road fitness scores, and the roulette wheel for all roads for 3 scenarios according to RL, RFL, RSC, and evaporation rate. To demonstrate the proposed OF-EACO working style, 5 road segments from road 1 to 5 are depicted. Figure 9 helps to closely follow the concept of OF-EACO.

Figure 9.

Pheromone intensity levels and fitness scores for roads.

3.5.3 Road Fitness Score (RFS)

The fitness function computes RFS for each road according to $τ_{R}^{N} (t)$ .The road with the short length, high fitness level, with low road signal count was assigned with optimal/high pheromone values via the pheromone update model. High RFS is assigned for the optimal road. The road with high RFS is an optimal shortest path with low vehicle density and signals. Equation 8 gives a fitness function.

\begin{aligned} {R F S}_{R}^{t} = \frac{τ_{R}^{N} (t)}{\sum_{n = 1}^{N} τ_{R}^{N} (t)} \end{aligned}

(8)

where

τ_{R}^{N} (t)

is the pheromone value of the road R at time

" t "

\sum_{n = 1}^{N} τ_{R}^{N} (t)

is the summation of “n” pheromone values associated with the roads of network N.

{R F S}_{R}^{t}

is the road fitness score of individual road

R

3.5.4 Optimal Road Selection Model (ORSM)

The roulette wheel concept is deployed for optimal road selection. An optimal road implies a short-length road with low vehicle density and low signal count. According to Pencheva et al. (2009), fitness scores are used as input to a roulette wheel to make optimal decisions. From the availability of roads, the optimal road for travel is selected. The roulette wheel applies a fitness-proportional mode of selection, which gives high priority to selecting high fitness scores. In this work, ${R F S}_{R}^{t}$ represents the fitness score of each road. A high fitness score road is equipped with a short length, low vehicle density, and low signal count. According to the roulette wheel, a higher fitness score has a higher probability of being selected for road travel. Also, the road with the lowest fitness score has the lowest probability rate of selection. Each road's fitness score is assigned to a roulette wheel for selection. Equation 9 indicates the roulette wheel and the probability ratio of picking a road.

\begin{aligned} P_{R}^{t} = \frac{{R F S}_{R}^{t}}{\sum_{N = 1}^{N} {R F S}_{R}^{t}} \end{aligned}

(9)

where

P_{R}^{t}

indicates the probability level of selection of the road R.

{R F S}_{R}^{t}

is the fitness score of the specific road R. N represents the total available fitness scores.

\sum_{N = 1}^{N} {R F S}_{R}^{t}

indicates the summation of N. A higher fitness score means a higher pheromone level on the road, and the same road is considered for travel. A high fitness score road is optimal in terms of short road length, low vehicle density, and low signal count. Algorithm 4 describes the sequential tasks implemented by the ORSM of the proposed OF-EACO.

Figure 10 depicts the optimal path selected for scenario 1 ( $ρ$ = 0). Figure 11 shows the optimal path selected for scenario 2 ( $ρ$ = 0.02). Figure 12 shows the optimal path selected for scenario 3 ( $ρ$ = 0.02, 0.03, 0.04, and 0.05).

Figure 10.

Optimal road selected for scenario 1 ( $ρ$ = 0).

Figure 11.

Optimal road selected for scenario 2 ( $ρ$ = 0.02).

Figure 12.

Optimal road selected for scenario 3 ( $ρ$ = 0.02, 0.03, 0.04, and 0.05).

4. Simulation and Results

4.1 Environment for Simulation

This partition provides the simulation background. The proposed OF-EACO and existing works like IDBACOR and VMR-ITS are simulated in Network Simulator (NS 2.35) using C++ and interpreter- OTcl (Object Tool Command Language) (Nawej et al., 2021; Ramamoorthy et al., 2022). In communication networks, NS 2.35 is widely used to simulate routing protocols/schemes. The open-source and event-centric nature of NS 2.35 provides extensive support for protocols to simulate bio-inspired behavior. A mobility model, such as MOVE (Mobility Model Generator for Vehicular Networks), generates the road scene for an IH-VANET.MOVE is built on the Java programming language. MOVE is based on SUMO 1.21.0 (Simulation of Urban Mobility), which simulates urban transportation. SUMO supports an open-source mode for the simulation of road traffic. NS2.35, in collaboration with SUMO, OpenStreetMap, and MOVE, enables the modeling of an IH-VANET network for proposed OF-EACO, IDBACOR, and VMR-ITS.

In this view from OpenStreetMap, a significant portion of the Hosur city map is downloaded, which is shown in Figure 13. The downloaded map is exported as Extensible Markup Language (XML) formatted files (map.osm) to create real-time road segments. Netconvert takes the roadmap placed on the map.Osm documents and produces the SUMO infrastructure using the map.Osm for use in map.net.xml. The corresponding scenario in SUMO is shown in Figure 14. Traffic Modeler uses map.net.xml to generate the traffic network and road.xml. The traffic modeler is a graphical user interface program that simulates vehicle traffic. Corresponding Vehicular traffic on the street graph is shown in Figures 15 and 16. TraNSLite converts traffic scenarios for NS 2.35 into executable packages in terms of trace files as per SUMO. The activity of collecting data from NS2 is associated with the simulation execution. During execution, NS2 generates a trace file (.tr). The .tr file records all the actions that take place in the simulator in chronological order as event types. Event type files include send(s), receive(r), drop(d), forward(f), time stamp, beacon ID, packet ID, vehicle ID, beacon size, and packet size, etc. These event type files serve as data sources for the data extraction process. AWK scripts are used in the data extraction process to parse the .tr files to estimate performance metrics.

Figure 13.

OpenStreetMap of Hosur urban.

Figure 14.

Same Hosur urban map as a street graph in SUMO..

Figure 15.

Vehicle traffic in SUMO.

Figure 16.

A close-up view of vehicle traffic in SUMO.

Figure 17 depicts the simulation steps used for simulating the proposed OF-EACO and existing works. For simulation, urban road scenarios with different length variations up to 9 km and vehicles with minimum and maximum speed limits of 10–100 km/h are considered. One-lane urban roads with an RTS count of 5, a maximum of 200 vehicles, and different variations of road signals ranging from 17 to 22 are considered for simulations. The transmission range of RTS is fixed at 500 m and 300 m for vehicles and road signals. The beacons are propagated with a size of 1024 bytes. For the simulation, a pheromone is deposited, and significant evaporation rates (0.04, 0.03, 0.05, 0.01, and 0.02) are followed to evaluate the proposed work. To characterize the realistic nature of an IH-VANET environment, variations in vehicle count, road length, and signal count are considered. Different variations in all terms help the proposed OF-EACO to prove its suitability in all respects. The proposed OF-EACO and existing works are simulated in NS2 as per Table 6.

Figure 17.

Simulation steps followed in the simulation of the proposed OF-EACO and existing works.

Table 6.

Simulation Parameters.

Simulation parameters	Values
Simulation Time	Maximum 400s
No of Roads & Maximum Road length	5 & 9 Km
Traffic model	CBR (Constant Bit Rate)
Transmission range	500 and 300 m
Data packet size	1024 byte
Mobility model	SUMO
MAC protocol	IEEE 802.11p
Link bandwidth	1 Mbps
Number of vehicles	200
Vehicles Average speed	10–100 km/h
No of RTS	5
Road Signal Count	17 to 22
Evaporation rate	$ρ$ = 0.04,0.03,0.05,0.01,and 0.02
Initial energy	100J

4.2 Results

The result section presents the performance of the proposed OF-EACO, IDBACOR, and VMR-ITS. Vehicle Travel Cost (VTC), Road Establishment Time (RET), Convergence speed (CS), Road Traffic Congestion Overhead (RTCO), Routing Overhead (RO), Computational overhead (CO), Computational Complexity (CC), Actual Wall Time Analysis (A-WTA), and Energy Consumption (EC) are the performance metrics used to evaluate the performance of all works. Table 7 shows the comparison of OF-EACO with existing schemes.

Table 7.
Comparison of OF-EACO with Existing Schemes.

Pheromone Activation

Approaches Methodology Design Aim Deposit Update Evaporation rate Conceptual Decision Making Scalability Simulator

IDBACOR Modified ACO with the distance-oriented method To invent the shortest route Distance Distance, ancient pheromone quantity, and evaporation rate Variable-rate Optimizes paths using distance Probabilistic rule-based decision making Convergence slows down in large networks NS2

VMR-ITS Hybrid multi-hop routing algorithm Route planning to reduce travel time and delay – – – Optimizes paths according to vehicle density Decision according to road density Convergence slows down in large networks NS2

Proposed OF-EACO Optimal fitness-based EACO with DMLS To invent a short, optimal road to reduce vehicle travel time, delay, and Overhead RL, RFL, and RSC Old pheromone value, RL, RFL, RSC, and evaporation rate Variable-rate Optimizes roads using RL, RFL & RSC Roulette wheel-based selection via fitness scores Supports fast convergence in large-scale networks NS2

			Pheromone Activation
IDBACOR	Modified ACO with the distance-oriented method	To invent the shortest route	Distance	Distance, ancient pheromone quantity, and evaporation rate	Variable-rate	Optimizes paths using distance	Probabilistic rule-based decision making	Convergence slows down in large networks	NS2
VMR-ITS	Hybrid multi-hop routing algorithm	Route planning to reduce travel time and delay	–	–	–	Optimizes paths according to vehicle density	Decision according to road density	Convergence slows down in large networks	NS2
Proposed OF-EACO	Optimal fitness-based EACO with DMLS	To invent a short, optimal road to reduce vehicle travel time, delay, and Overhead	RL, RFL, and RSC	Old pheromone value, RL, RFL, RSC, and evaporation rate	Variable-rate	Optimizes roads using RL, RFL & RSC	Roulette wheel-based selection via fitness scores	Supports fast convergence in large-scale networks	NS2

The metrics are chosen to align with the proposed OF-EACO goals. VTC directly indicates the optimal road capacity based on travel time and distance, which is essential for IH-VANETs. RET and CS are used to estimate how efficiently OF-EACO picks the best roads under different mobility conditions. To evaluate scalability and communication reliability, RTCO and RO are included. CO and CC are used to assess the computational feasibility of OF-EACO. A-WTA estimates the wall time of the approaches. Additionally, EC is selected to measure the precise energy consumption rate of the approaches. Collectively, all these metrics offer a thorough and justified basis for performance evaluation.

4.2.1 Vehicle Travel Cost (Seconds)

Vehicle Travel Cost (VTC) indicates the total time taken by the vehicle to complete the journey between the start and endpoints. VTC is measured in seconds. Mathematically, VTC can be defined as

\begin{aligned} V T C_{N}^{T} = \frac{D_{N}^{T} (T)}{S_{N}^{T} (T)} \end{aligned}

(10)

$V T C_{N}^{T}$ is the travel cost of the N^th vehicle for the trip “T”. $D_{N}^{T} (T)$ is the distance traveled by the N^th vehicle for a trip “T”, and $S_{N}^{T} (T)$ is the speed of the N^th vehicle for a trip “T”. Figures 18, 19, and 20 show the VTC obtained for all works for 10 to 50 km/h, 30–50 km/h, and 50–100 km/h speed scales. The proposed OF-EACO gives a lower VTC compared to IDBACOR and VMR-ITS. IDBACOR only focuses on choosing the optimal shortest route, and VMR-ITS is better at

Figure 18.

VTC for 10–50 km/h speed scale.

Figure 19.

VTC for 30–50 km/h speed scale.

Figure 20.

VTC for 50–100 km/h speed scale.

choosing the optimal route without congestion, which shows better results in VTC for VMR-ITS.IDBACOR applies a distance-based method that finds the distance of the path and assigns a pheromone according to the distance. IDBACOR is not configured to find the route according to the vehicle density and other important parameters, which takes more time for vehicles to reach their destination compared to VMR-ITS and the proposed OF-EACO. Using a vehicle traffic server in VMR-ITS guides the vehicle to choose the optimal route according to vehicle density, but it ignores the road signals installed on the roads for the optimal road selection, which makes the vehicles take a longer time to reach their destination. Furthermore, OF-EACO is enriched in selecting optimal roads through the ORSM module. ORSM takes important parameters like road length, RTS, and RSC count for road selection. ORSM applies the optimal pheromone to prepare RFS according to RL, RFL, RSC, and evaporation rate. The optimal road is assigned with the highest RFS score. An optimal road refers to a road of short length with low vehicle density and low signal count. The use of RL, RFL, and RSC in the proposed OF-EECO contributed to the reduction of VTC. For the speed scale of 10–50 km/h, OF-EACO reduces VTC by 3% to 4% over VMR-ITS and 7% to 9% over IDBACOR. For the speed scale of 30–50 km/h, OF-EACO gives 5% to 7% better than VMR-ITS and 8% to 10% better than IDBACOR. Similarly, OF-EACO gives 7% to 10% better than VMR-ITS and 11% to 13% better than IDBACOR for a speed scale of 50 to 100 km/h.

4.2.2 Road Establishment Time (Seconds)

Road Establishment Time (RET) refers to the total time taken to establish/find a road from the starting point to the destination. RET is measured in seconds. Mathematically, RET can be defined as

\begin{aligned} R E T_{R}^{T} = \sum_{m = 1}^{t} (T_{t} + P D_{t}) \end{aligned}

(11)

Where $R E T_{R}^{T}$ indicates the total time taken to construct the road. $T_{t}$ is the transmission time. $P D_{t}$ is the processing delay. $T_{t}$ is estimated according to $B e a c o n S i z e s / T r a n s m i s s i o n r a t e$ of the channel. To find beacon sizes, beacons initiated by vehicles to register with RTS are considered. A summation is calculated in terms of sizes for all beacons launched by vehicles, and a summary value is assigned to the metric “beacon sizes”. $P D_{t}$ is the time taken by the vehicle to process the optimal road selection mechanism. Figures 21, 22, and 23 show the RET cost for all tasks for 10–50 km/h, 30 to 50 km/h, and 50–100 km/h speed scales.

Figure 21.

RET for 10–50 km/h speed scale.

Figure 22.

RET for 30–50 km/h speed scale.

Figure 23.

RET for 50–100 km/h speed scale.

The proposed OF-EACO outperforms IDBACOR and VMR-ITS in terms of RET. Pre-implementation of modules like DMLS, RL, RFL, and RSC helps the ORSM module to select the optimal road quickly. The use of the roulette wheel in ORSM facilitates quick road selection according to RFS with less time compared to IDBACOR and VMR-ITS. The roulette wheel concept has also been adopted in IDBACOR, but the EACO in IDBACOR faces significant delays as it requires maximum iterations to complete the task of assigning fitness scores to roads. This delay introduces more time for road selection. Minimum iterations are imposed to update routing information for route selection by using ORSM in the EACO of the proposed work. Similarly, VMR-ITS takes more time to find the road compared to the proposed OF-EACO.VMR-ITS uses the concept of forward and backward communication for route selection, and the route selection process waits to complete a cycle of forward and backward communication, which imposes more time for route selection.

Relatively, the proposed OF-EACO uses ORSM for road selection, which is free from forward and backward communication. Instead of relying on forward and backward communication, the proposed work uses a Roulette cycle to accelerate RET over VMR-ITS. For the speed scale of 10–50 km/h, OF-EACO gives 4% to 7% better than VMR-ITS and 7% to 9% better than IDBACOR. For the speed scale of 30–50 km/h, OF-EACO reduces RET by 3% to 5% over VMR-ITS and 6% to 8% over IDBACOR. Similarly, OF-EACO gives 2% to 3% better than VMR-ITS and 4% to 5% better than IDBACOR for a speed scale of 50 to 100 km/h.

4.2.3 Convergence Speed (Number of Iterations)

Convergence speed (CS) indicates how fast the algorithm reaches the optimal solution. Important parameters such as repetitions, evaporation rate, pheromone update rule, pheromone stability, and ant movements affect the CS. Mathematically, CS can be defined as

\begin{aligned} {C S}_{P}^{O} = (\frac{O p t i m a l S o u l t i o n V a l u e}{i}) + P h e r o m o n e s t a b l i t y \end{aligned}

(12)

${C S}_{P}^{O}$ is the convergence speed for a problem “P” to produce the optimal solution “O”. The optimal solution value represents the identified optimal road for iteration “i”. Pheromone stability refers to the available pheromone for road “R” according to the pheromone update rule (Equation 7). Figures 24, 25, and 26 clearly show that the CS of the proposed OF-EACO is much faster compared to IDBACOR for the speed scale of 10–50 km/h (Figure 24), 30–50 km/h (Figure 25), and 50–100 km/h (Figure 26). For 10–50 km/h, the proposed OF-EACO converges to the best value after at most 20 iterations, while IDBACOR can find the optimal solution after at most 31 iterations. The optimal solution represents the best path in terms of distance.

Figure 24.

CS for 10–50 km/h speed scale.

Figure 25.

CS for 30–50 km/h speed scale.

Figure 26.

CS for 50–100 km/h speed scale.

For 30–50 km/h, the proposed OF-EACO reaches the optimal solution in 27 iterations, and IDBACOR takes 40 iterations to generate the optimal solution. Similarly, for 50–100 km/h, the proposed OF-EACO reaches the optimal solution in 31 iterations, and IDBACOR takes 47iterations to generate the optimal solution. IDBACOR and the proposed OF-EACO are rich in finding optimal paths, but the convergence speed varies. Using RFS reduces the total number of iterations required by OF-EACO compared to IDBACOR. However, IDBACOR uses a dynamic pheromone update rule, which is subject to exploration problems. The exploration problem raises slow convergence. Furthermore, the proposed OF-EACO uses modified dynamic pheromone update rules to control pheromones on optional roads according to different evaporation rates and RFS. This adaptation makes OF-EACO superior to IDBACOR in terms of convergence speed.

4.2.4 Road Traffic Congestion Overhead (%)

Road Traffic Congestion Overhead (RTCO) refers to the ratio of total congestion delay to total travel time. Total travel time denotes the duration spent on the road by the vehicle to reach its destination, while total congestion delay signifies the extra time consumed on the road due to congestion/traffic at a given time “t”. Mathematically, RTCO can be expressed as

\begin{aligned} R T C O_{R}^{t} = \frac{T o t a l C o n g e s t i o n D e l a y_{R}^{t}}{T o t a l T r a v e l T i m e_{R}^{t}} \times 100 \end{aligned}

(13)

$R T C O_{R}^{t}$ is the RTCO on road “R” at a time $" t "$ . $T o t a l C o n g e s t i o n D e l a y_{R}^{t}$ is the supplementary time the vehicle spends on the road at a time $" t "$ . Mathematically, $T o t a l C o n g e s t i o n D e l a y_{R}^{t}$ can be expressed a s

\begin{aligned} T o t a l C o n g e s t i o n D e l a y_{R}^{t} = \sum_{z = 1}^{N} (T o t a l T r a v e l T i m e_{R}^{t} - T i m e c o n s u m e d a t R S_{R}^{t}) \end{aligned}

(14)

here,

T i m e c o n s u m e d a t R S_{R}^{t}

indicates the time spent by the vehicle at road signals.

Figures 27, 28, and 29 clearly show that the RTCO of the proposed OF-EACO is very low compared to VMR-ITS and IDBACOR for the speed scale of 10–50 km/h (Figure 27), 30–50 km/h (Figure 28), and 50–100 km/h (Figure 29). In the proposed OF-EACO, the optimal road is selected from the available “n” roads as per RFS. RFS was prepared for each road according to its pheromone values. The road with high RFS is an optimal shortest path with low vehicle density and signals. EACO's ORSM selects the optimal road according to the RFS to support the lowest RTCO.

Figure 27.

RTCO for 10–50 km/h speed scale.

Figure 28.

RTCO for 30–50 km/h speed scale.

Figure 29.

RTCO for 50–100 km/h speed scale.

Similarly, IDBACOR chooses the optimal road, which is the shortest-length road for travel, but fails to consider the level of road density and traffic signals. This observation in IDBACOR increases the total travel time and total congestion delay. IDBACOR focuses only on selecting the shortest route in terms of length without considering vehicle density and other information. However, VMR-ITS supports vehicle traffic servers to manage traffic information to guide path-planning algorithms to provide low traffic overhead. At best, VMR-ITS focuses on vehicle count and signal units regardless of route length. Compared to the proposed OF-EACO, VMR-ITS introduces a little RTCO. For the speed scale of 50–100 km/h, OF-EACO records 2%-4% RTCO, while VMR-ITS records 3%-7% and IDBACOR records 4%- 8%. For the speed scale of 30–50 km/h, OF-EACO records 3%-6% RTCO, while VMR-ITS records 5%-10% and IDBACOR records 5%- 9%. Similarly, OF-EACO registers 4% to 7% RTCO and VMR-ITS, and IDBACOR registers 8% to 13% and 10% to 14% at speeds of 10 to 50 km/h.

4.2.5 Routing Overhead (%)

Routing Overhead (RO) is the ratio of total routing packets to the maximum allowable control and road explorer ants. Routing packets represent Advertisement Beacon (AB), Registration Beacon (RB), Hello Beacon (HB), Pheromone Credit/Update messages ( $P h_{U}^{C}$ ) and Road Explorer Ants (RE-ants). Mathematically, RO can be defined as

\begin{aligned} R O_{R}^{t} = \frac{A B + R B + H B + P h_{U}^{C} + R E A n t s}{M a x a l l o w a b l e c o n t r o l a n d R E A n t s} \times 10 \end{aligned}

(15)

$R O_{R}^{t}$ is the RO on road “R” at a time $" t "$ . In the proposed OF-EACO, under DMLS, vehicle routing tables and RTS are updated in time to limit the use of control packets. Under the RSC module, hello beacons are used in a time-limited manner. Furthermore, the EACO of the proposed work requires a minimum number of iterations to obtain optimal solutions compared to VMR-ITS and IDBACOR.IDBACOR implements maximum control ants to reach the optimal solution in terms of path finding with higher rounds of iterations compared to the proposed work. Compared to IDBACOR, VMR-ITS demands a minimum control packets to perform its route planning algorithm .

Using fewer control packets to discover routes registers a lower RO for VMR-ITS compared to IDBACOR. For the speed scale of 50–100 km/h, OF-EACO registers RO of 2%–3% lower than VMR-ITS and 3%–6% lower than IDBACOR. For the speed scale of 30–50 km/h, OF-EACO registers RO of 2%–4% lower than VMR-ITS and 5%–9% lower than IDBACOR. Similarly, OF-EACO registers RO of 3% to 7% lower than VMR-ITS and 5% to 12% lower than IDBACOR at speeds from 10 to 50 km/h. Figure 30 depicts the RO achieved by all works for a 10–50 km/h speed scale. Figure 31 depicts the RO achieved by all works for a 30–50 km/h speed scale. Figure 32 depicts the RO achieved by all works for a 50–100 km/h speed scale.

Figure 30.

RO for 10–50 km/h speed scale.

Figure 31.

RO for 30–50 km/h speed scale.

Figure 32.

RO for 50–100 km/h speed scale.

4.2.6 Computational Overhead (%)

In IH-VANETs, Computational overhead (CO) refers to the percentage of processing time spent on auxiliary tasks (DMLS, RL, RFL, RSC, pheromone credit/update messages, and RE-ants) instead of actual useful work, such as ORSM. Mathematically, CO can be represented as

\begin{aligned} C O_{R}^{t} = \frac{C_{A u x i l i a r y T a s k s}^{t} - C_{A c t u a l T a s k}^{t}}{C_{A u x i l i a r y T a s k s}^{t}} \end{aligned}

(16)

$C O_{R}^{t}$ is the percentage of CO that occurred while identifying road “R” at a time “t”. $C_{A u x i l i a r y T a s k s}^{t}$ is the additional work performed while identifying road “R” at a time “t”. $C_{A c t u a l T a s k}^{t}$ is the actual useful work performed to find the correct road “R” at the time “t”. The proposed OF-EACO registers a computational complexity of $O (n \log n)$ compared to the existing VMR-ITS and IDBACOR. Due to its efficient computational complexity, CO is reduced compared to VMR-ITS and IDBACOR.OF-EACO's actual work ORSM uses the roulette wheel concept to quickly select the optimal road according to the RFSs. ORSM operates efficiently and carries out its work effectively. Furthermore, both VMR-ITS and IDBACOR record a computational complexity of $O (n^{2})$ .To reach a solution, the actual work of VMR-ITS is to wait for the entire cycle of forward and backward communication to complete. This dependency produces a higher CO compared to OF-EACO. However, IDBACOR's reliance on more iterations to reach the optimal solution results in a significant percentage of CO. Figures 33, 34, and 35 clearly show that the CO of the proposed OF-EACO is very low compared to VMR-ITS and IDBACOR for the speed scale of 10–50 km/h, 30–50 km/h, and 50–100 km/h. For the speed scale of 10 to 50 km/h, OF-EACO registers CO of 9% to 13% lower than VMR-ITS and 10% to 14% lower than IDBACOR. For the speed scale of 30–50 km/h, OF-EACO registers CO of 4%–5% lower than VMR-ITS and 6%–12% lower than IDBACOR. Similarly, OF-EACO registers CO 3%–4% lower than VMR-ITS and 6%–10% lower than IDBACOR at speeds from 50 to 100 km/h.

Figure 33.

CO for 10–50 km/h speed scale.

Figure 34.

CO for 30–50 km/h speed scale.

Figure 35.

CO for 50–100 km/h speed scale.

4.2.7 Computational Complexity (CC)

Table 8 presents CC. The proposed OF-EACO records a CC of $O (n \log n)$ , which is relatively efficient because it relies on road fitness scores to select the optimal roads quickly. Fast processing with good convergence speed ensures timely decisions for routing in vast networks. Notably, with optimal convergence speed and low overhead, the proposed OF-EACO has a CC of $O (n \log n)$ . With a CC of $O (n^{2})$ , VMR-ITS is ideal for dynamic networks because it focuses only on finding the appropriate road. However, the route selection process of VMR-ITS always needs to complete the cycle of forward and backward communication to produce the optimal solution. Waiting to complete the entire cycle takes more time for route selection. This increases the time complexity of VMR-ITS. However, using ORSM in the proposed OF-EACO reduces the time complexity. IDBACOR has a high CC of $O (n^{2})$ because it requires more iterations to assign fitness scores to find optimal roads. The time complexity of IDBACOR increases significantly due to the slow convergence. However, the proposed OF-EACO has a high convergence speed, which facilitates the use of fewer iterations to assign fitness scores to roads. The OF-EACO proposal outperforms existing approaches in terms of computational usage for dynamic IH-VANETs.

4.2.8 ActualWall Time Analysis (A-WTA) (Seconds)

In CA-WT, wall time (WT) refers to the actual elapsed execution time engaged by OF-EACO from start to finish. The start-to-finish time includes all the necessary computation and memory operations intended to complete the road selection. Mathematically, WT can be represented as

\begin{aligned} W T = E n d T i m e - S t a r t t i m e \end{aligned}

(17)

In the implementation, the start and end times of OF-EACO, VMR-ITS, and IDBACOR are logged using a clock, and the execution is repeated for each scenario. The scenarios are scaled with vehicle variations including 20, 40, 60, 80,100,150, and 200. Each scenario is implemented for speed scales of 10–50 km/h, 30–50 km/h, and 50–100 km/h. The measured values also indicate that the proposed OF-EACO is efficient in terms of taking less WT compared to the existing VMR-ITS and IDBACOR. This superiority is achieved due to the optimal convergence speed of OF-EACO. Furthermore, OF-EACO operates with less computational overhead than VMR-ITS and IDBACOR, which also helps it consume less WT. Figures 36, 37, and 38 clearly show that the CA-WT of the proposed OF-EACO is very low compared to VMR-ITS and IDBACOR for the speed scales of 10–50 km/h, 30–50 km/h, and 50–100 km/h. For the speed scale of 10 to 50 km/h, OF-EACO registers CA-WT of 7% to 10% lower than VMR-ITS and 9% to 12% lower than IDBACOR. For the speed scale of 30–50 km/h, OF-EACO registers CA-WT of 6%–10% lower than VMR-ITS and 8%–11% lower than IDBACOR. Similarly, OF-EACO registers CA-WT 5%–9% lower than VMR-ITS and 7%–10% lower than IDBACOR at speeds from 50 to 100 km/h.

Figure 36.

WT for 10–50 km/h speed scale.

Figure 37.

WT for 30–50 km/h speed scale.

Figure 38.

WT for 50–100 km/h speed scale.

In addition to the asymptotic complexity as CC, it is essential to estimate the actual WT of the proposed OF-EACO, as it directly reflects the actual implementation cost under real-world scenarios. As shown in graphs (36, 37, and 38), the gradual increase in WT demonstrates the scalability of OF-EACO in the real-time frontier. The WT analysis confirms the real-world feasibility of OF-EACO in realistic IH-VANETs. Furthermore, the discussion on CC suggests that the OF-EACO grows exponentially as the number of vehicles grows. The measured WT suggests that the OF-EACO is practical for urban-scale networks. Although the implementations are with 20–200 vehicles per road, this range effectively captures dissimilar traffic densities from low to moderately solid urban circumstances. Our measured WT is less than 12, 14, and 16 s, even for 200 vehicles, for seed scales of 10–50 km/h, 30–50 km/h, and 50–100 km/h. This observation is very close to linear empirical scaling. Real-world deployments may have more than 200 vehicles, but the results suggest that OF-EACO scales linearly with the empirical scaling. This near-linear empirical scaling recommends that OF-EACO can be extrapolated to larger networks beyond 200 vehicles, preserving realistic finishing times appropriate for real-time routing in real-world IH-VANET environments.

4.2.9 Energy Consumption (%)

Energy Consumption (EC) represents the total energy consumed by participating vehicles during road finding. Equation 18 calculates the energy used by all vehicles involved in the finding process. EC is mathematically calculated as

\begin{aligned} E C_{R}^{t} = \sum_{i = 1}^{N} E_{T o t a l}^{(z)} \end{aligned}

(18)

$E C_{R}^{t}$ is the energy consumed by a group of vehicles while finding the road “R” at a time “t”. $E C_{R}^{t}$ is represented in joules. “N” represents the number of vehicles, which varies from 20 to 200. $E_{T o t a l}^{(z)}$ is the total energy consumed by vehicle “Z” while participating in road formation. $E_{T o t a l}^{(z)}$ is calculated for each vehicle according to the mathematical function given below.

\begin{aligned} E_{T o t a l}^{(z)} = E_{T x}^{(z)} (\frac{n_{T x}^{z} . B_{S}^{z} . P_{T x}^{z}}{B_{S}^{Z}}) + E_{R x}^{(z)} (\frac{n_{R x}^{z} . B_{S}^{z} . P_{R x}^{z}}{B_{S}^{Z}}) + E_{Pr}^{(z)} (N_{O p s}^{(z)} . E_{O P}^{(z)}) \end{aligned}

(19)

$E_{T x}^{(z)}$ is the total energy (joules) consumed by vehicle “Z” to transmit beacons. $n_{T x}^{z}$ refers to the number of beacons transmitted. $B_{S}^{z}$ refers to the beacon size. $P_{T x}^{z}$ refers to the power (watts) used for transmission. $E_{T x}^{(z)}$ is the total energy (joules) consumed by vehicle “Z” to transmit beacons. $n_{R x}^{z} . B_{S}^{z} . P_{R x}^{z}$ refers to the number of beacons received, the beacon size, and the power (watts) used to receive the beacons. $E_{Pr}^{(z)}$ is the energy (joules) consumed to process beacons. $N_{O p s}^{(z)}$ is the number of processing operations. $E_{O P}^{(z)}$ is the energy (watts) used for each operation. The proposed OF-EACO registers a lower EC compared to VMR-ITS and IDBACOR. In VMR-ITS, the concept of forward and backward communication introduces processing delays.

To output the optimal road, VMR-ITS must complete a cycle of forward and backward communication. This requirement delays the process of finding the optimal road, and the routing approach consumes a significant amount of energy compared to the proposed OF-EACO. Furthermore, IDBACOR relies heavily on INFO beacons to produce an optimal solution. The use of more INFO beacons for managing the network, route finding, and route maintenance results in higher energy consumption than the proposed OF-EACO. The proposed OF-EACO requires a minimum number of iterations to reach the optimal solution compared to VMR-ITS, which saves the vehicle's energy. Also, the lower use of beacons in the proposed OF-EACO compared to IDBACOR can avoid high energy consumption. Figures 39, 40, and 41 clearly show that the EC of the proposed OF-EACO is much lower compared to VMR-ITS and IDBACOR for the speed scales of 10–50 km/h, 30–50 km/h, and 50–100 km/h. For the speed scale of 10 to 50 km/h, OF-EACO records 3% to 5% lower $E C_{R}^{t}$ than VMR-ITS and 6% to 9% lower $E C_{R}^{t}$ than IDBACOR. For the 30–50 km/h speed scale, OF-EACO registers 2%–5% less $E C_{R}^{t}$ than VMR-ITS and 5%–10% less than IDBACOR. Similarly, OF-EACO registers 2%–4% lower $E C_{R}^{t}$ than VMR-ITS and 4%–8% lower than IDBACOR at speeds from 50 to 100 km/h.

Figure 39.

EC for 10–50 km/h speed scale.

Figure 40.

EC for 30–50 km/h speed scale.

Figure 41.

EC for 50–100 km/h speed scale.

5. Comprehensive Performance Analysis

The proposed OF-EACO and existing approaches are tested for all performance metrics at speed scales of 10 to 50 km/h, 30 to 50 km/h, and 50 to 100 km/h. Adopting optimal roads with low density and fast optimal road selection improves the VTC and RET of OF-EACO over VMR-ITS and IDBACOR. Using fewer iterations to reach optimal solutions makes OF-EACO better than existing approaches in terms of CS. OF-EACO's reliance on optimal short roads with low vehicle density and signal produces lower RTCO for all conditions. The use of very few beacons and iterations improves the RO of the proposed OF-EACO compared to existing approaches. The minimal time spent on auxiliary tasks reduces the CO of OF-EACO compared to existing approaches. Compared to existing methods, OF-EACO has a lower A-WTA due to its faster convergence speed. The minimum number of iterations and reliance on beacons in OF-EACO saves energy for vehicles. Table 9 depicts the Quantitative comparison of the proposed OF-EACO over VMR-ITS and IDBACOR.

Table 8.
Computational Complexity Analysis.

Approach CC

VMR-ITS $O (n^{2})$

IDBACOR $O (n^{2})$

Proposed OF- EACO $O (n \log n)$

Approach	CC
VMR-ITS	$O (n^{2})$
IDBACOR	$O (n^{2})$
Proposed OF- EACO	$O (n \log n)$

Table 9.

Quantitative Comparison of Proposed OF-EACO vs VMR-ITS and IDBACOR.

Performance metrics	IDBACOR (Average for all scenarios)	VMR-ITS (Average for all scenarios)	Proposed OF-EACO (Average for all scenarios)	Average Improvement (%) of OF-EACO over IDBACOR	Average Improvement (%) of OF-EACO over VMR-ITS
VTC (seconds)	22.0	17.6	10.3	11%	7%
RET (seconds)	17.1	14.7	9.8	7%	5%
CS (Iterations)	39	–	26	5%	–
RTCO (%)	20.9	19.8	14.1	5%	4%
RO (%)	29.2	25.6	17.4	7%	5%
CO (%)	19.4	14.8	8.8	12%	7%
CA-WT (Seconds)	15.8	15.0	7.5	11%	10%
EC (%)	20.6	15.9	10.7	9%	5%

The performance assessment metrics VTC, RET, CS, RTCO, RO, CO, and EC determine the practical viability of OF-EACO. The assessment of the metrics for SUMO traffic scenarios also demonstrates the practical viability of the proposed OF-EACO. Similarly, CC and A-WTA demonstrate the computational viability of OF-EACO. The $O (n \log n)$ CC of OF-EACO ensures the fitness evaluation and optimal road selection. Also, estimating the pheromone update scale according to vehicle density, traffic signals, and road lengths. A-WTA confirms near-linear scaling of 20–200 without spikes. This evaluation confirms the computational efficiency of OF-EACO for real-time networks.

6. Circumstances Affecting the Efficiency of the Proposed OF-EACO

OF-EACO offers an optimal, uncongested short road with low vehicle density and fewer traffic signals, resulting in shorter travel times for vehicles. The proposed OF-EACO demonstrates significant success in selecting optimal roads for IH-VANETs. Results indicate that the proposed OF-EACO outperforms IDBACOR and VMR-ITS in all aspects.

Since the proposed OF-EACO is optimal, it may encounter certain situations under different circumstances that could lead to a halt or impact the road selection process. These include poor pheromone trails and aggressive evaporation rates. Both factors can affect the road selection process and cause the search to turn into random exploration. For example, an extreme evaporation rate ( $ρ$ = 1) can prematurely disrupt the learning process and prevent convergence. Additionally, while roulette wheel-based selection enhances probabilistic choice, it still faces challenges balancing exploration and exploitation. In some cases, a high-fitness road might dominate others, skewing the search process toward exploitation.

7. Conclusion

The proposed work of this article aims to identify the optimal road by considering road type, RL, RFS, and RSC. The optimal road is characterized as an efficient, uncongested short-distance route with low vehicle density and minimal signal count. In this article, an optimized fitness-based enhanced ACO is presented to help vehicles travel quickly on optimal short-length roads without delay and congestion for urban scenarios. To provide optimal roads, the proposed OF-EACO initially applies DMLS to register vehicles under RTS, and other modules such as RL, RFL, and RSC are implemented to find road length, road fitness level, and signal count. Next, EACO's ORSM dynamically executes a roulette wheel according to the higher RFS to find optimal roads among the available roads. In OF-EACO, only a high fitness score has short RL, optimal RFL, low RSC, and high pheromone values based on the dynamic evaporation rate. On optimal roads, the vehicle experiences less travel time, less traffic or congestion, and less delay.

In the proposed OF-EACO, the use of optimal roads without congestion and travel delay reduces the VTC for speed scales of 10–50 km/h, 30–50 km/h, and 50–100 km/h under different mobility conditions. For all three speed ranges, the proposed OF-EACO shows an average of 7% lower VTC than VMR-ITS and 11% lower than IDBACOR. Pre-implementing all modules and using a roulette wheel allows the ORSM in the proposed system to quickly select the best road, saving RET. For speeds of 10–50 km/h, 30–50 km/h, and 50–100 km/h with vehicle variations, OF-EACO reduces the average RET by 5% compared to VMR-ITS and 7% compared to IDBACOR. Modified pheromone rules decrease the number of iterations and enable OF-EACO to rapidly find optimal solutions despite different vehicle variations. Across these speed ranges, OF-EACO averages 26 iterations, whereas IDBACOR requires 39. OF-EACO's reliance on optimal, low-density roads with minimal signals results in lower RTCO for all speed scales. The proposed OF-EACO achieves an average RTCO that is 4% lower than VMR-ITS and 5% lower than IDBACOR. Its smaller number of iterations and fewer beacons reduce the RO significantly, recording a 5% decrease compared to VMR-ITS and 9% compared to IDBACOR across all speeds. Minimizing auxiliary tasks lowers the CO of OF-EACO; it exhibits an average CO that is 7% lower than VMR-ITS and 12% lower than IDBACOR for all speed ranges. With optimal convergence speed and low overhead, OF-EACO has a CC of $O (n \log n)$ indicating high efficiency, as VMR-ITS and IDBACOR also register a CC of $O (n^{2})$ .Faster convergence reduces the WT, with OF-EACO showing 10% less WT than VMR-ITS and 11% less than IDBACOR across all speeds. The minimal number of iterations and beacon reliance help conserve vehicle energy. Consequently, OF-EACO consumes about 5% less EC than VMR-ITS and 9% less than IDBACOR on average.

The limitations of the proposed OF-EACO include the reliance on GPS-based precise location information to make routing decisions. The localization dependency affects the efficiency of the proposed system. The future scope and improvements of the current study are twofold. First, updating the current OF-EACO with ad hoc localization methods for different traffic patterns with different practical scenarios for highway road networks. Furthermore, RTS in this work can introduce overhead for highway scenarios. To avoid this, we plan to integrate edge computing technology with machine learning techniques into RTS for the efficient processing of heuristic data near vehicles while satisfying quality of service (QoS) parameters such as privacy, integrity, and security. Second, the current OF-EACO is focused on the IEEE 802.11p standard for communication. Extending OF-EACO to IH-VANET communication technologies is interesting. Combining OF-EACO with Long-Term Evolution-based Vehicle-to-Everything (LTE-V2X) and 5G New Radio-based V2X (5G NR-V2X) offers high data rates, optimal connectivity, low latency, and minimal overhead. This integration facilitates effective operations in dense networks with varying QoS requirements.

Footnotes

ORCID iD

Raghu Ramamoorthy

Author Contributions

All authors contributed equally.

Ethical Standard

This work does not require ethics approval.

Consent to Participate

This work does not require consent to participate because it does not involve human subjects.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability

Authors confirm that the data supporting the findings of this study are available within the article.

References

Arshad

Ullah

Khalid

Ahmad

Khalid

Shahwar

Cao

(2019). Beacon trust management system and fake data detection in vehicular ad-hoc networks. IET Intelligent Transport Systems, 13(5), 780–788. https://doi.org/10.1049/iet-its.2018.5117

Arzil

S. A.

Aghdam

M. H.

Jamali

M. A. J.

(2010, October). Adaptive routing protocol for VANETs in city environments using real-time traffic information. 2010 International Conference on Information, Networking and Automation (ICINA) (Vol. 2, pp. V2–132). IEEE. https://doi.org/10.1109/ICINA.2010.5636966

Bijalwan

Hussain

Purohit

K. C.

Kumar

M. A.

(2023). Enhanced ant colony optimization for vehicular ad hoc networks using fittest node clustering. Sustainability, 15(22), 15903. https://doi.org/10.3390/su152215903

Bilal

S. M.

Khan

A. U. R.

Khan

S. U.

Madani

S. A.

Nazir

Othman

(2014). Road-oriented traffic information system for vehicular ad hoc networks. Wireless Personal Communications, 77(4), 2497–2515. https://doi.org/10.1007/s11277-014-1651-0

Calabrò

Torrisi

Inturri

Ignaccolo

(2020). Improving inbound logistic planning for large-scale real-world routing problems: A novel ant-colony simulation-based optimization. European Transport Research Review, 12(1), 21. https://doi.org/10.1186/s12544-020-00409-7

Cheng

(2023). Dynamic path optimization based on improved ant colony algorithm. Journal of Advanced Transportation, 2023(1), 7651100. https://doi.org/10.1155/2023/7651100

Chitra

Sathya

S. S.

(2018). Acknowledgment-based localization method (ALM) to improve the positioning in vehicular Ad hoc networks. International Journal of Computer Network and Information Security, 10(11), 50. https://doi.org/10.5815/ijcnis.2018.11.06

Choudhary

Pahuja

(2023). Awareness routing algorithm in vehicular ad-hoc networks (VANETs). Journal of Big Data, 10(1), 122. https://doi.org/10.1186/s40537-023-00742-3

Contreras

Gamess

(2020). An algorithm based on VANET technology to count vehicles stopped at a traffic light. International Journal of Intelligent Transportation Systems Research, 18(1), 122–139. https://doi.org/10.1007/s13177-019-00184-3

10.

Dubey

Tiwari

Yadav

P. K.

Pandey

(2024). Reliable cooperative communication in cognitive vehicular networks for intelligent transportation systems. International Journal of Communication Systems, 37(6), e5694, 1–25. https://doi.org/10.1002/dac.5694

11.

El Hamdani

Benamar

Younis

(2020). Pedestrian support in intelligent transportation systems: Challenges, solutions, and open issues. Transportation Research Part C: Emerging Technologies, 121, 102856. https://doi.org/10.1016/j.trc.2020.102856

12.

Fatemidokht

Kuchaki Rafsanjani

(2018). F-ant: An effective routing protocol for ant colony optimization based on fuzzy logic in vehicular ad hoc networks. Neural Computing and Applications, 29(11), 1127–1137. https://doi.org/10.1007/s00521-016-2631-y

13.

Flügel

Halse

A. H.

Hartveit

K. J.

Ukkonen

(2022). Value of travel time by road type. European Transport Research Review, 14(1), 1–35. https://doi.org/10.1186/s12544-022-00554-1

14.

Heidari

Jamali

M. A. J.

Navimipour

N. J.

(2025). Fuzzy logic multicriteria decision-making for broadcast storm resolution in vehicular Ad Hoc networks. International Journal of Communication Systems, 38(5), e6034. https://doi.org/10.1002/dac.6034

15.

Ibrahim

M. F.

Mustofa

M. I.

Meilanitasari

Wijaya

S. U.

(2022). An improved ant colony optimization algorithm for vehicle routing problem with time windows. JurnalTeknikIndustri, 23(2), 105–120. https://doi.org/10.22219/JTIUMM.Vol23.No2.105-120

16.

Jindal

Bedi

(2018). An improved hybrid ant particle optimization (IHAPO) algorithm for reducing travel time in VANETs. Applied Soft Computing, 64, 526–535. https://doi.org/10.1016/j.asoc.2017.12.038

17.

Lee

Heo

Kim

Choi

(2024). Multiple-junction-based traffic-aware routing protocol using ACO algorithm in urban vehicular networks. Sensors, 24(9), 2913. https://doi.org/10.3390/s24092913

18.

Lekshmi

Pramila

R. S.

(2024). Prioritization and scheduling in the vehicle-to-infrastructure (V2I) scheme for vanets using enhanced congestion control source-based ant colony optimization. Automatika, 65(3), 778–792. https://doi.org/10.1080/00051144.2024.2307232

19.

Soleimani

Zohal

(2019). An improved ant colony optimization algorithm for the multi-depot green vehicle routing problem with multiple objectives. Journal of Cleaner Production, 227, 1161–1172. https://doi.org/10.1016/j.jclepro.2019.03.185

20.

Lin

C. C.

Deng

D. J.

(2015). Optimal two-lane placement for hybrid VANET-sensor networks. IEEE Transactions on Industrial Electronics, 62(12), 7883–7891. https://doi.org/10.1109/TIE.2015.2418314

21.

Liu

Meidani

(2025a). Graph neural networks for travel distance estimation and route recommendation under probabilistic hazards. International Journal of Transportation Science and Technology, 1–12. https://doi.org/10.1016/j.ijtst.2025.02.006 [9]

22.

Liu

Meidani

(2025b). Multi-class traffic assignment using multi-view heterogeneous graph attention networks. Expert Systems with Applications, 286, 128072. https://doi.org/10.1016/j.eswa.2025.128072

23.

Majumdar

Prasad

P. R.

Kumar

S. S.

Kumar

K. S.

(2016, May). An efficient routing algorithm based on ant colony optimization for VANETs. 2016 IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT) (pp. 436–440). IEEE. https://doi.org/10.1109/RTEICT.2016.7807858

24.

Naderi

Zargari

Ghanbari

(2019). Adaptive beacon broadcast in opportunistic routing for VANETs. Ad Hoc Networks, 86, 119–130. https://doi.org/10.1016/j.adhoc.2018.11.011

25.

Nandagopal

Kumar

P. S.

Rajalakshmi

Anandamurugan

(2023). Mobility aware zone-based routing in vehicle ad hoc networks using hybrid metaheuristic algorithm. Intelligent Automation & Soft Computing, 36(1), 113–126. https://doi.org/10.32604/iasc.2023.031103

26.

Nawej

Owolawi

Walingo

(2021, April). Design and simulation of vanets testbed using openstreetmap, sumo, and ns-2. 2021 IEEE 6th International Conference on Computer and Communication Systems (ICCCS) (pp. 582–587). IEEE. https://doi.org/10.1109/ICCCS52626.2021.9449197

27.

Nebbou

Fouchal

Lehsaini

(2018, May). A distributed measurement of road density. 2018 IEEE International Conference on Communications (ICC) (pp. 1–5). IEEE. https://doi.org/10.1109/ICC.2018.8422758

28.

Nemade

C. H.

Pujeri

(2023). Emergency automobile data transmission with ant colony optimization (ACO). Journal of Advances in Information Technology, 14(5), 1003–1011. https://doi.org/10.12720/jait.14.5.1003-1011

29.

Nithyanandam

Ambayiram

Natarajan

(2024). Beetle colony optimization algorithm-based node clustering scheme for efficient data dissemination in vehicular ad hoc networks. International Journal of Communication Systems, 37(4), e5680. https://doi.org/10.1002/dac.5680

30.

Oonsrikaw

Thammano

(2018, June). Enhanced ant colony optimization with local search. 2018 IEEE/ACIS 17th International Conference on Computer and Information Science (ICIS) (pp. 291–296). IEEE. https://doi.org/10.1109/ICIS.2018.8466388

31.

Pencheva

Atanassov

Shannon

(2009). Modelling of a roulette wheel selection operator in genetic algorithms using generalized nets. International Journal Bioautomation, 13(4), 257–264.

32.

Phuc

P. N. K.

Thao

N. L. P.

(2021). Ant colony optimization for multiple pickup and multiple delivery vehicle routing problem with time window and heterogeneous fleets. Logistics, 5(2), 1–13. https://doi.org/10.3390/logistics5020028

33.

Rahimi

Jabraeil Jamali

M. A.

(2019). A hybrid geographic-DTN routing protocol based on fuzzy logic in vehicular ad hoc networks. Peer-to-Peer Networking and Applications, 12(1), 88–101. https://doi.org/10.1007/s12083-018-0642-4

34.

Ramamoorthy

(2024). An enhanced location-aided ant colony routing for secure communication in vehicular ad hoc networks. Human-Centric Intelligent Systems, 4(1), 25–52. https://doi.org/10.1007/s44230-023-00059-7

35.

Ramamoorthy

Kumar

Naidu

R. C. A.

(2022, December). Hybrid multihop routing mechanism with intelligent transportation system architecture for efficient routing in VANETs. 2022 international conference on disruptive technologies for multi-disciplinary research and applications (CENTCON) (Vol. 2, pp. 69–74). IEEE. https://doi.org/10.1109/CENTCON56610.2022.10051361

36.

Ramamoorthy

Naidu

R. C. A.

Kumar

Ashwin

(2023, November). Performance investigation of hybrid bio-inspired zone routing protocol over proactive and reactive routing protocols. 2023 International Conference on Sustainable Communication Networks and Application (ICSCNA) (pp. 1–6). IEEE. https://doi.org/10.1109/ICSCNA58489.2023.10370667

37.

Ramamoorthy

Thangavelu

(2021). Group-based dual-mode key management scheme for secure communication in vehicular ad hoc networks. Wireless Personal Communications, 120(2), 949–973. https://doi.org/10.1007/s11277-021-08498-w

38.

Ramamoorthy

Thangavelu

(2022a). An enhanced hybrid ant colony optimization routing protocol for vehicular ad-hoc networks. Journal of Ambient Intelligence and Humanized Computing, 13(8), 3837–3868. https://doi.org/10.1007/s12652-021-03176-y

39.

Ramamoorthy

Thangavelu

(2022b). An enhanced bio-inspired routing algorithm for vehicular ad hoc networks. Trends in Sciences, 19(10), 4188–4188. https://doi.org/10.48048/tis.2022.4188

40.

Ramamoorthy

Thangavelu

(2022c). An enhanced distance and residual energy-based congestion aware ant colony optimization routing for vehicular ad hoc networks. International Journal of Communication Systems, 35(11), e5179. https://doi.org/10.1002/dac.5179

41.

Ramamoorthy

Velu

Naidu

R. C. A.

Kumari

Kumar

(2024, December). Impact of ant colony optimization on optimal routing for vehicular ad hoc networks. 2024 5th International Conference on Communication, Computing & Industry 6.0 (C2I6) (pp. 1–6). IEEE. https://doi.org/10.1109/C2I663243.2024.10894774

42.

Rokbani

Kumar

Abraham

Alimi

A. M.

Long

H. V.

Priyadarshini

Son

L. H.

(2021). Bi-heuristic ant colony optimization-based approaches for traveling salesman problem. Soft Computing, 25, 3775–3794. https://doi.org/10.1007/s00500-020-05406-5

43.

Sbayti

Housni

(2024). A new routing method based on ant colony optimization in vehicular ad-hoc network. Statistics, Optimization & Information Computing, 12(1), 167–181. https://doi.org/10.19139/soic-2310-5070-1766

44.

Shakeel

S. M.

Ould-Khaoua

Rehman

O. M. H.

Al Maashri

Bourdoucen

(2015). Experimental evaluation of safety beacon dissemination in VANETs. Procedia Computer Science, 56, 618–623. https://doi.org/10.1016/j.procs.2015.07.262

45.

Siddiqi

M. H.

Draz

Ali

Iqbal

Alruwaili

Alhwaiti

Alanazi

(2022). FANET: Smart city mobility off to a flying start with self-organized drone-based networks. IET Communications, 16(10), 1209–1217. https://doi.org/10.1049/cmu2.12291

46.

Subramaniam

Rambabu

Chandrasekaran

Kumar

N. S.

(2022). A traffic density-based congestion control method for VANETs. Wireless Communications and Mobile Computing, 2022(1), 7551535, 14 pages. https://doi.org/10.1155/2022/7551535

47.

Suganyadevi

Nandalal

(2024). Swarm intelligence-inspired meta-heuristics hybrid optimization for multi-constraint routing in vehicular Adhoc networks. IETE Journal of Research, 70(1), 95–115. https://doi.org/10.1080/03772063.2023.2253763

48.

Tao

Zheng

Wang

Qiu

Stojanovic

(2024). Enhanced feature extraction YOLO industrial small object detection algorithm based on receptive-field attention and multi-scale features. Measurement Science and Technology, 35(10), 105023. DOI https://doi.org/10.1088/1361-6501/ad633d

49.

Tariuge

Matthew

(2024). An effective optimization algorithm for ant colony vehicular congestion management. Journal of Computer and Communications, 12(9), 119–130. https://doi.org/10.4236/jcc.2024.129007

50.

Thilak

K. D.

Amuthan

A. J. F. G. C. S.

(2018). Cellular automata-based improved ant colony-based optimization algorithm for mitigating DDoS attacks in vanets. Future Generation Computer Systems, 82, 304–314. https://doi.org/10.1016/j.future.2017.11.043

51.

Wang

Chen

(2019, July). Evacuation entropy path planning model based on hybrid ant colony-artificial fish swarm algorithms. IOP Conference Series: Materials Science and Engineering (Vol. 563, No. 5, p. 052025). IOP Publishing. https://doi.org/10.1088/1757-899X/563/5/052025

52.

Wang

Yang

Gao

Cui

Wang

(2022). Research on smooth path planning method based on improved ant colony algorithm optimized by Floyd algorithm. Frontiers in Neurorobotics, 16(955179), 1–18. https://doi.org/10.3389/fnbot.2022.955179

53.

Gao

Wang

Zhang

(2023). A hybrid ant colony algorithm based on multiple strategies for the vehicle routing problem with time windows. Complex & Intelligent Systems, 9(3), 2491–2508. https://doi.org/10.1007/s40747-021-00401-1

			Pheromone Activation
Approaches	Methodology	Design Aim	Deposit	Update	Evaporation rate	Conceptual	Decision Making	Scalability	Simulator
IDBACOR	Modified ACO with the distance-oriented method	To invent the shortest route	Distance	Distance, ancient pheromone quantity, and evaporation rate	Variable-rate	Optimizes paths using distance	Probabilistic rule-based decision making	Convergence slows down in large networks	NS2
VMR-ITS	Hybrid multi-hop routing algorithm	Route planning to reduce travel time and delay	–	–	–	Optimizes paths according to vehicle density	Decision according to road density	Convergence slows down in large networks	NS2
Proposed OF-EACO	Optimal fitness-based EACO with DMLS	To invent a short, optimal road to reduce vehicle travel time, delay, and Overhead	RL, RFL, and RSC	Old pheromone value, RL, RFL, RSC, and evaporation rate	Variable-rate	Optimizes roads using RL, RFL & RSC	Roulette wheel-based selection via fitness scores	Supports fast convergence in large-scale networks	NS2

An Optimized Fitness-Based Enhanced ant Colony Optimization for Optimal Road Selection in Intelligent Heterogeneous Vehicular ad-hoc Networks

Abstract

Keywords

1. Introduction

4.1 Environment for Simulation

4.2.8 ActualWall Time Analysis (A-WTA) (Seconds)

Table 8. Computational Complexity Analysis. Approach CC VMR-ITS O ( n 2 ) IDBACOR O ( n 2 ) Proposed OF- EACO O ( n log n )

7. Conclusion

Footnotes

ORCID iD

Author Contributions

Ethical Standard

Consent to Participate

Funding

Declaration of Conflicting Interests

Data Availability

References

Table 8.
Computational Complexity Analysis.

Approach CC

VMR-ITS $O (n^{2})$

IDBACOR $O (n^{2})$

Proposed OF- EACO $O (n \log n)$