International Journal of Reconfigurable and Embedded Systems (IJRES)

Vol. 12, No. 3, November 2023, pp. 403~413

ISSN: 2089-4864, DOI: 10.11591/ijres.v12.i3pp403-413  403

Route optimization via improved ant colony algorithm with

graph network

Patil N. Siddalingappa1, Poornima Basavaraj2, Preethi Basavaraj3, Premasudha B. Gowramma4

1
Department of Information Science, Bapuji Institute of Engineering and Technology, Davangere, India
2
Department of Information Science and Engineering, Bapuji Institute of Engineering and Technology, Davangere, India
3
Department of Computer Science and Engineering, Bapuji Institute of Engineering and Technology, Davangere, India
4
Department of Master of Computer Applications, Siddaganga Institute of Technology, Tumkur, India

Article Info ABSTRACT

Article history: Route optimization problem using vehicle routing problem (VRP) and time
window constraint is explained as finding paths for a finite count of vehicles
Received Aug 11, 2022 to provide service to a huge number of customers and hence, optimizing the
Revised Dec 10, 2022 path in a given duration of the time window. The vehicles in the loop have
Accepted Mar 20, 2023 restricted intake of capacity. This path initiates from the depot, delivers the
goods, and stops at the depot. Each customer is to serve exactly once. If the
arrival of the vehicle is before the time window “opens” or when the time
Keywords: window “closes,” there will be waiting for cost and late cost. The challenge
involved over here is to scheduling visits to customers who are only
Ant colony optimization
available during specific time windows. Ant colony optimization (ACO)
Graph algorithm
algorithm is a meta-heuristic algorithm stimulated by the growing behaviour
Graph network
of real ants. In this paper, we combine the ACO algorithm with graph
Route optimization
network henceforth increasing the number of vehicles in a particular depot
Vehicle routing problem
for increasing the efficiency for timely delivery of the goods in a particular
time width. This problem is solved by, an efficient technique known as the
ACO+graph algorithm.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Patil N. Siddalingappa
Department of Information Science, Bapuji Institute of Engineering and Technology
Davangere, Karnataka, India
Email: patilns_12@[Link]

1. INTRODUCTION
Vehicle routing problem with time windows (VRPTW) has been a basic issue to transport goods
that are to be delivered within a time bandwidth is a trending topic for research in the upcoming years. This
task is elaborated as selecting routes for restricted number of vehicles to serve a group of customers in the
allotted time window. The vehicle has a specific capacity. It initially begins from the depot and ends at the
depot. All the customers are to be served exactly once. The main objective of the proposed system is to
reduce the total transport costs. This paper presents a transportation system for graph database to process the
required information. The databases used here demand information collected from all sources like vehicle
maps, and timetable, the path traversed from source to destination to incorporate the features to find an
optimal path, which will handle many users, to achieve efficiency to scale the system. Relationship between
the objects to integrate the primary constraints where a transport stops. To achieve efficiency to select a
particular algorithm in the form of a computer memory. Graph databases are chosen based on the concept of
structured query language (SQL). The possibility of storing a graph in a graph based system. To determine
the efficiency and scalability to use data integrity along with accessibility.

Journal homepage: [Link]

404  ISSN:2089-4864

As a result, an evolutionary algorithm (EA) is presented to handle optimization difficulties [1],

hence enhancing efficiency. Particle swarm optimization (PSO) is one of these techniques that has proven to
be effective in solving a variety of challenging optimization problems. We describe an improved ant system
approach for the vehicle routing problem (VRP) with a single central depot and several similar vehicles in
this work. The goal is to solve combinatorial optimization issues by mimicking the behavior of ant colonies
based on observations of real ants hunting for food. Real ants have been discovered to use a fragrant essence
called pheromone to relay information about food sources. The behavioral mechanism described above can
be used to solve combinatorial optimization problems by simulation.
Wang et al. [2] offer an improved ant system solution for the VRP by increasing the total number of
vehicles for timely product delivery, and show that the ant pattern for food finding may be employed to
provide competitive outcomes. Cluster analysis is a type of mathematical analysis method encountered, as
well as a type of machine learning method without supervisor monitoring that belongs to NP difficult
problems [3], but the analysis method over here has been widely applied in image identification, information
retrieval, data mining, statistics, machine learning, spatial database [4], and bifurcation analysis. ACO has
improved the results of a series of combinatorial optimization solutions, and by applying ACO to practical
problems, Braekers et al. [5] researchers have proposed the artificial ant colony concept; artificial ant colony
and real ant colony have many similarities, but artificial ant colony has unique skills that real ant colony
lacks. In the section, based on introducing ACO thought, it designs the algorithm prototype system, provides
optimization, and improves the opinions for the system. The goal of VRPTW [6]-[8] is to choose the best
vehicle route from a set of possible options. The goal is to keep the whole cost as low as possible. The
dimension’s encoding approach is used for the VRPTW with demand points in this paper due to the unknown
number of cars. Two dimensions are involved for each demand point. The serial number of the vehicle that
serviced the demand point is the first point. The serial of the serviced demand point is the second point. Each
particle conveniently correlates to a matrix in which the first dimension corresponds to the serial number of
vehicles and the second dimension corresponds to the serial number of consumers.
They propose that a multiswarm cooperative particle swarm optimizer (MCPSO), which is a variant
of the PSO form [9]. It employs a multiswarm cooperative evolutionary method in which the master swarms
alter their particles based on their own knowledge and that of the slave swarms' particles, while the slave
swarms carry out PSO separately to obtain particle diversity. In the context of faster route generation by
implementing artificial intelligence (AI) technique onto it, implemented on an intelligent system to estimate
the consumption of energy and estimate the travel time by application of heuristics algorithms to solve
vehicle routing problems. An efficient meta-heuristic technique selected to be applied on this generalization
for vehicle routing problem to achieve better performance. A theoretical approach is carried out which states
that a generic approach is used to combine different techniques to solve this problem [10]. The technique is
based on re-optimizing or re-routing approach. This approach is tested with seven different techniques and
the results are obtained based on the heuristic approach. This also involves by minimizing the no of
customers for service. The feasibility of the solution approach decreases for path optimization by reducing
the number of vehicles. The infeasibility problem occurs while there will the customer no longer serviced in a
specific time window. A deep learning based automatic vehicle location system (AVLS) system-modelling
technique is proposed to develop vehicle routes in this system [11]-[14]. An optimization-based approach is
significantly using deep reinforcement learning strategy to estimate the parameters of the model. This
approach is sensitive to many parameters like time, which is quite time consuming. A neural network-based
optimization strategy is developed for vehicle routing in minimum computation time. Figure 1 shows the
sample representation of a graph network. Consider a road network below consists of vertices and arcs.
Vertices are depots, intersections, and customers, arcs are roads, which are directed or undirected [15].

Figure 1. Sample representation of a graph network

Int J Reconfigurable & Embedded Syst, Vol. 12, No. 3, November 2023: 403-413
Int J Reconfigurable & Embedded Syst ISSN:2089-4864  405

2. RELATED WORK
The multi-compartment electric vehicle routing problem (EVRP) with a soft time window and
multiple charging kinds is discussed here [16], [17] where electric multi-compartment cars are used that are
environmentally benign but require charging during the transportation process. The goal of this optimization
issue is to minimize the function composed of vehicle cost, distribution cost, time window penalty cost, and
billing service cost using a mathematical model. An estimation of the distribution algorithm based on Lévy
flight (EDA-LF) is presented to overcome this problem by doing a local search at each iteration to avoid the
algorithm from falling into a local optimum [18]. The EDA-LF method finds better solutions and is more
robust than the basic EDA algorithm, according to experimental results. A graph data model consists of data
structures [19], different constraints while implementing a task to perform connections in transport
environment. These databases refer to the category of NoSQL databases. They have high efficiency and
handle many users, while using a graph database to enhance the quality of the system to shorten the response
time and simply implement different functions. The vehicle routing problem along with time window
constraints along with road-network information [20]. At first, a customer-based graph is introduced to solve
the problem of specificity. The vehicle routes are defined for customer-based graphs, a dynamic approach is
considered for to estimate the travel-time with minimum cost and derive optimal solutions for the problem.
Vehicle routing problems are solved by customer graphs where many attributes are embedded in road
segments, alternate paths are considered on different customer graphs [21], [22]. Two methods are used to
handle customer-based graphs to represent different attributes defined in road segments.
To represent multigraph road networks where an arc is introduced for each path, an exact solution is
developed to determine the paths, which are not dominated by linking a set of points interested in a road
network. The proposed algorithm used here is multi-dimension multi-objective searching strategy. The
efficiency to solve a road network problem to reduce the cost. The optimal solution to the road network to
find instances, which are not feasible to use the shortest path, which is carried to in the form of simple
representation [23]. Significant research is carried out to solve problems related to VRP to accomplish the
objectives to reduce the total cost, total distance, the solutions for VRP are classified into three types exact
methods, heuristics and meta heuristics, the exact methods are further classified into branch and X, constant
programming and dynamic programming. VRP problems are said to be NP-hard, solving these problems is
time-consuming. The customer-based graph for time dependent VRPTW to build two algorithms with road
network for multigraphs. Several vehicle routing problems to compute customer-based graph network. The
main goal here is to emphasize on tackling large class of vehicle routing problem. The possible enhancement
for recent advancement in technology to obtain the real-time traffic condition scenario [24], [25]. A graph
database is implemented to find an optimal solution while connecting two stops in vehicle routing scenario.
Graph databases are categorized using segmentation known as non-relational database known as NoSQL
database. A graph model is constructed for transportation problem for vehicle routing in task orientation to
traverse an optimal search path. The path-shift from public transport system to classic modelling of the graph
technique has been exploited to transform between different relations of a database. A detailed representation
of public transportation system is carried out and a relational graph-based model is developed [26].
Edge computing, traffic management are studied in detail in the transportation of green vehicle
systems. There are different mechanisms for different state-of-art approaches to attain the necessary energy
efficiency for various green vehicle systems [27]. Offloading circumstances, in edge computation for
outsourcing the requirements by predicting the estimated arrival of different workloads, employing AI
technique to facilitate between switching the harvesting states from one state to another. In 6G era, numerous
vehicles are connected using the same infrastructure providing secure communication for vehicle routing for
different problems. The unauthorized users pose a security threat to different vehicle owners by intruding the
system while accessing the system resources. The similarities between VRP and JSPs problem are studied,
they have the same scenario in execution of their tasks, and their resources are constrained based on the
capacity to process their number of tasks. Here exists a time interval that depends on the consecutive
execution of tasks to minimize the various lengths of the windows to be executed while specifying a new
relationship altogether as far as the capacity constraints are considered [28]. An optimizing the criteria for
minimizing various definitions to compress the length of activities to minimize each transition time. A
similar technology is used for scheduling as well as vehicle routing problems.
The customer-based graphs representing a road network, to analyze different graph routing problems
to achieve alternate paths to handle different limits [28]. The efficiency to tackle the problem on the road
network is compared along the multigraph approach. They have developed a multigraph approach for branch-
and-price algorithm to improvise the solution of to conduct and analyze the characteristics of VRPTW, which
effect the performance with road network and multigraph settings. First step is to develop an algorithm to
estimate the associated shortest paths based on the road network to find the instances of expected departure
time. To adapt a road network which is based on time-dependent algorithm. The optimal solution found for the
road network for different instances, which are not feasible to use the shortest path for the representation.
Route optimization via improved ant colony algorithm with graph network (Patil N. Siddalingappa)
406  ISSN:2089-4864

Friggstad [11] a VRPTW plays a key role in all business operations, recently this has allowed the operations to
be carried out in a real time scenario. Latest technologies also possess new problems to the business
management techniques, which affects the decision-making approach for dynamic VRPTW problems as well. A
generic approach is proposed here to solve this problem, this algorithm serves as a basis for other new
algorithms like the insertion approach and DVRPTW, as well as MACS, and these algorithms were compared
with many state-of-art-techniques, the best performing algorithm is selected in comparison with other literature
review techniques [29]. Meta–heuristics technique requires more CPU time, and are very complex for
implementation purposes. The real-life applications choose flexible approach, to cluster these practical
applications; few methods are more attractive and effective for tackling complex constraints. The solution
approach for local meta-heuristic techniques improves the complex constraints involved in it. The ACO
approach is implemented by considering the analogy of real ant colonies for food [30].

3. PROPOSED METHOD
The optimization according to the path by considering three parameters: i) a depot where many
customers are serviced, ii) many vehicles serviced starting from the starting point, and iii) to traverse the path
starting from the depot to the termination point. A customer is served by one particular vehicle, another
constraint is the vehicle should meet the time-window constraint; this should meet the requirements such as
customer satisfaction, minimizing the cost, to find the optimal path. A penalty cost is provided for early or
late arrivals that improvise customer fulfilment. The penalty obtained for late arrival is high in aspect of early
arrival. When delivery is concerned taking into account several aspects involved with optimal distribution of
path apart from goods requirement, refrigerated vehicle weight as well as time window is considered.
Consider a graph 𝐺 = (𝑉, 𝐴) a fully connected graph that depicts the network of the distribution of vehicles.
Specifically 𝑉 = {0, 1, 2, ….., n} be the set of nodes, where 0 represents the distribution center and 1, 2, 3,
…….., n depicts the customer points. 𝐴= {(𝑎, 𝑏): 𝑎, 𝑏∈𝑁, a ≠ b} that depicts the routes. Considering the
distribution center, customer location, and the demand known in prior. The distance is calculated between
two customers (𝑎, 𝑏) by Euclidean distance measure.

3.1. Cost evaluation

The cost of the vehicle consists of vehicle wear maintenance, the total expense, and the salary of the
driver. Let 𝑔ℎ depict the used vehicle ℎ. 𝑎𝑥𝑦𝑘
is a 0-1 decision variable where 𝑥𝑖𝑗𝑘 denotes that the vehicle 𝑘
𝑥
departs from the distribution center to the customer 𝑖 and 𝑗, else 𝑎𝑦𝑧 =0. The vehicle fixed cost 𝐹𝐶 is by given
(1). The customer’s demand is used to estimate the distribution cost of the vehicle. 𝑇𝐶 Represents the burden
per unit weight and 𝑑𝑎𝑏 the distance between the customers 𝑎 and 𝑏, 𝑚𝑎 denotes the cargo demand of the
customer 𝑎. The transportation cost is mentioned in (2). The products, which are perishable, that consist of
cargo damage from wear and tear the timestamp that at which the vehicle leaves the distributed center, the
products that are delivered within specific time requirement. The cargo damage occurs within a time window
as accepted by the customer. The association between the distance and damage coefficient grows
exponentially; 𝝉 is the declining co-efficient to detect the freshness of the product and 𝑡𝑖 is the actual
estimated time to reach the customer a and 𝑡𝑜 the time window required by the customer. The total damage
cost can be find by (3).

𝐹𝐶 = ∑𝑋𝑥=1 ∑𝑛𝑦=1 ∑𝑛𝑧=1 𝑎𝑦𝑧

𝑥
𝑔ℎ (1)

𝑇𝐶 = ∑𝑋𝑥=1 ∑𝑛𝑦=1 ∑𝑛𝑧=1 𝑎𝑦𝑧

𝑥
𝑑𝑎𝑏 𝑚𝑎 (2)

𝑇𝐷𝐶 = ∑𝑋𝑥=1 ∑𝑛𝑦=1 ∑𝑛𝑧=1 𝑎𝑦𝑧

𝑥
𝑑𝑎𝑏 𝑚𝑎 (1 − 𝑓 −τ1 ( 𝑡𝑖 − 𝑡𝑜 )) (3)

3.2. Distribution cost

The measure of time the distribution center that responds to the requirements of the customer known
as satisfaction. The expected time of the customer to deliver the goods, the satisfaction is higher, the expected
time period of the customer in the interval is given by [ 𝑓𝑎 , 𝑙𝑎 ]. The arrival time is given by 𝐴𝑇𝑎 for the
number of customers is given as n. the total satisfaction is considered as 100 for the delivery, the penalty
denoted as 𝝳, for early delivery penalty factor is denoted as δ1 and for late delivery δ2 , here δ1 < δ2 ,the
satisfaction of a single customer is termed as C𝑖 is defined as (4), (5).

Int J Reconfigurable & Embedded Syst, Vol. 12, No. 3, November 2023: 403-413
Int J Reconfigurable & Embedded Syst ISSN:2089-4864  407
100 𝑓𝑎 −𝐴𝑇𝑎
C𝑖 = { − δ1 , 𝑖𝑓 𝑓𝑎 > 𝐴𝑇𝑎
𝑛 𝑓𝑎
100
{ , , 𝑖𝑓 𝑓𝑎 < 𝐴𝑇𝑎 < 𝑙𝑎 (4)
𝑛
100 𝐴𝑇𝑎 −𝑙𝑎
{ − δ2 , 𝑖𝑓 𝑙𝑎 > 𝐴𝑇𝑎
𝑛 𝑙𝑎

𝑓𝑎 −𝐴𝑇𝑎 𝐴𝑇𝑎 −𝑆𝑎

𝐷𝐶= 5δ1 + 10δ2 (5)
𝑓𝑎 𝑆𝑎

3.3. Optimization of the ACO+graph network model

Various factors such as vehicle cost, distribution cost, and customer satisfaction cost is determined
by (6), (7), (8). In (9) and (10) determines that each customer is serviced by one vehicle service. Equation
(11) the delivery vehicle leave the distribution center for delivering the goods and then return back to the
distribution center, in (12), the total weight of the goods indeed required by the customer could not exceed
the load of the vehicle. In (13) we can say that the no of vehicles in the distribution center is limited. In (14)
the relationship of customers from y to y+1. In (15) once the y-th customer is serviced the weight of the
vehicle is estimated by 𝑑 𝑦 , the vehicle leaving the distribution enter is the same vehicle as the one returning
the distribution center, the value taken by the decision variable should not be negative in the range 0-1, the
time window wherein the delivery vehicle arrives late or early.

𝑀𝑖𝑛𝑡𝑜𝑡𝑎𝑙 = 𝐹𝐶 + 𝑇𝐶 + 𝑇𝐷𝐶 (6)

𝑥 𝑢𝑎𝑏 𝑃𝑡
𝑀𝑖𝑛𝑡 = 𝐴𝑎𝑏 (𝛾 (𝜇 𝐵𝑡 + + 𝑃𝑐 ) /𝛼)ℎ (7)
𝑣𝑎𝑏 𝑛

100 𝑏𝑎 − 𝐷𝑏 100 100 𝐷𝑎 − 𝑀𝑎

𝑀𝑎𝑥𝑡 = [ − 𝜇1 ]+ +[ − 𝜇2 ] (8)
𝑁 𝑏𝑎 𝑁 𝑁 𝑀𝑎

∑𝑛𝑥=1,𝑦=1 ∑𝑋𝑥=1 𝑎𝑦𝑧

𝑥 = 1,
(9)

𝑥 − ∑𝑛 ∑𝑋 𝑎 𝑥 = 0, 𝑏 ∈ n,
∑𝑛𝑦 =1 ∑𝑋𝑥=1 𝑎𝑞𝑧 𝑧=1 𝑥=1 𝑏𝑧 (10)

∑𝑛𝑦 =1 ∑𝑋𝑥=1 𝑎𝑦𝑜

𝑥 = ∑𝑛 𝑋 𝑥
𝑦=1 ∑𝑥=1 𝑎𝑜𝑧 = 𝑋, ∀ y, z ∈ G, x ∈ X (11)

∑𝑛𝑦=1, 𝑎𝑦𝑥𝑒 𝐷 𝑦 ≤ 𝐷𝑥𝑒 , ∀ 𝑥 ∈ X, e∈ 𝐸𝑋 (12)

∑𝑋𝑥=1 ∑𝑛𝑧=1 𝑎𝑧𝑥 ≤ X, ∀ x ∈ X (13)

∑𝑛𝑧=0( 𝐷𝑧𝑦
𝑃 − 𝐷 𝑃 ) = 𝑑 𝑦 , ∀(y, z) ∈ G
𝑦𝑧 (14)

∑𝑋𝑥=1 ∑𝑛𝑧=0 𝑎𝑦𝑧

𝑥
= ∑𝑋𝑥=1 ∑𝑛𝑦,𝑧=0 𝑎𝑧𝑦,
𝑥
∀ 𝑥, 𝑦 ∈ 𝐺, 𝑥 ∈ 𝑋,
d𝑦 ≥ 0, ∀𝑦 ∈ 𝐺
𝑥 𝑥
𝑎𝑦𝑧 (1 − 𝑎𝑦𝑧 ) = 0, ∀ 𝑦, 𝑧 ∈ G, x ∈ X,
(15)
𝑓𝑦 ≤ 𝐴𝑇𝑦 ≤ 𝑙𝑦 , y ∈ n
𝑓𝑦 − 𝐴𝑇𝑦 > 0, ∀ y ∈ G
𝑙𝑦 − 𝐴𝑇𝑦 < 0∀ y ∈ G

Ant colony optimization technique is employed here to solve the optimization problems to
understand the pattern of ants searching for food they find a short route from food to their particular nest they
follow each other by releasing a chemical known as pheromone, the replication of their above procedure of
the ants following the trail makes the trail more accurate. The algorithm proposed here to develop each
vehicle for the respected duration of time window. This model can be applied to a multiple start point with
multiple delivery points who source us to have a pickup and delivery source from source to destination while
in the process of selecting the best node to the delivery point. We have chosen an algorithm here for selecting
the appropriate route if it satisfies the particular criteria or it can select the next node until it fits into the
constraints. Algorithm 1 displays the improvised ACO+graph network.

Route optimization via improved ant colony algorithm with graph network (Patil N. Siddalingappa)
408  ISSN:2089-4864

Algorithm 1. Improvised ACO+graph network

Step 1: initialize the parameters, consisting of number of ants max, the number of levels
is given as 𝐿𝑚𝑎𝑥 , the pheromone repetitions is given by 𝑃𝐿 , weights associated with
pheromone level is given as µ and ρ. The max and min values of pheromone are given as 𝛽𝑚𝑎𝑥
and𝛽𝑚𝑖𝑛 .
Step 2: To place the m ants randomly, initialization 𝑃ℎ𝑒𝑟𝑜𝑚𝑜𝑛𝑒𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 , considering the
adjacent node based on given node.
Step 3: Updation and restriction of 𝑃ℎ𝑒𝑟𝑜𝑚𝑜𝑛𝑒𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 , value between min and max of
pheromone, transition probability of each node is obtained. Thereby visiting each node and
completely visit all nodes.
Step 4: Estimate the total cost evaluation, calculate the optimized path for minimal cost
and customer satisfaction by using the Pareto optimal principle, when N reaches the max
value, the 𝑃ℎ𝑒𝑟𝑜𝑚𝑜𝑛𝑒𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 is initialized again.
Step 5: The value of N is less than the maximum, go to step 2 else finish the iteration,
output the Pareto optimal principle and optimize the distribution path.

3.4. Data and parameter selection

The dataset considered is Solomon standard considering 0 to 100 customers, considering the
parameters like customer satisfaction, the time in which the goods are serviced in time window constraints.
The ant colony optimization algorithm, the number of levels considered in 𝑃ℎ𝑒𝑟𝑜𝑚𝑜𝑛𝑒𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 . The
number of ants considered has a direct influence on the optimization of the path for effective transmission.
The cost minimization needed to be considered for different levels in the ant colony optimization algorithm.
VRP integrated along graph network, of multigraph-based model known as ACO+graph network model.

4. RESULTS AND ANALYSIS

4.1. Dataset details
The data is selected based on analysis of the Pareto optimum solution, which is the most effective
solution to the multiple aims programming scenario. The best version of it suppresses the sub-objectives to
achieve the main objective quickly. As a result, there is very little potential for conflict between objectives
and sub-objectives, and a solution is obtained after noise removal. The proposed ACO+graph network
method outperforms the competition in numerous test cases and 100 customers in three different
circumstances. This information is organized in a Pareto optimal solution format. Because no one of the
resolutions in the Pareto front has been subjugated through solutions which are outside, (or by solutions
inside the Pareto front curve), the non- subjugated explanations have the few objective disruptions. This
feature gives the decision-maker more options to choose from. Multi-objective optimization of cost, carbon
emissions, and customer happiness is explored in this study, with the result determined as a Pareto optimal
solution.

4.2. Result analysis

The analysis of result is carried out using the Pareto optimal solution, which concludes that project
benefit of 80% work comes from 20% of work carried out. This is optimized further to state that the sub-
objectives are curbed in effectively achieving the objective. Due to this issue, there is less scope to achieve
the objectives and to obtain a solution without noise. The results are tested with test cases of 25, 50, 75, and
100 customers using the proposed approach to provide a better solution. The detailed description of each
result is shown below. The abbreviations considered in graph are: cost in CNY (C(CNY), carbon emissions
(CE), customer satisfaction (CS), and number of vehicles used (NY).

4.3. Test case for 25 customers

The path is optimized to achieve efficient performance for the C101 (25) to obtain the results using
Pareto optimal principle. We can find from the graph that the algorithm ACO+graph theory gives better
results in comparison with that of other state-of-art-techniques. Figure 2 shows the comparison of
ACO+graph network, ACOMO, ACO for c101_25 customers. The proposed algorithm ACO+graph network
which sums up the total cost to 3,139, the vehicle used to get an optimal path is (0, 5, 3, 7, 8, 10, 11, 9, 6, 4,
2, 1, and 0), (0, 13, 17, 18, 19, 15, 16, 14, 12, 0) and (0, 20, 24, 25, 23, 22, 21, 0). Figure 3 displays the
solution for c101 with 25 customers.

4.4. Test case for 50 customers

The analysis here is carried out on the test set for 50 customers with c101 (50), 5 vehicles have been
deployed with the paths considered. The total cost estimated here sums up to 5943. Figure 4 displays
comparison of ACO+graph network, ACOMO, ACO for c101_50 customers. Figure 5 displays the solution

Int J Reconfigurable & Embedded Syst, Vol. 12, No. 3, November 2023: 403-413
Int J Reconfigurable & Embedded Syst ISSN:2089-4864  409

for c101 with 50 customers. Optimal path is (0, 43, 42, 41, 40, 44, 46, 45, 48, 50, 49, 47,0), (0, 5, 3, 7, 8, 10,
11, 9, 6, 4, 2, 1,0), (0, 20, 24, 25, 27, 29, 30, 28, 26, 23, 22, 21,0), (0, 32, 33, 31, 35, 37, 38, 39, 36, 34,0), (0,
13, 17, 18, 19, 15, 16, 14, 12, 0).

C101_25
4000
3482 3394
COMPARISION

3500 3139
3000
2500
2000
1500
1000
500 19,6 100 19,8 98,5 20,5 98,5
0
C(CNY) CE CS C(CNY) CE CS C(CNY) CE CS
ACO+graph network ACOMO ACO
METHODOLOGY

Figure 2. Comparison of ACO+graph network, ACOMO, ACO for c101_25 customers

Figure 3. Solution for c101 with 25 customers

C101_50
12000
9914
9584
10000
COST

8000
5943
6000

4000

2000
33,5 100 53,8 99,3 54,7 98,5
0
C (CNY) CE CS C (CNY) CE CS C (CNY) CE CS
ACO+Graph Network ACOMO ACO
METHODOLOGY

Figure 4. Comparison of ACO+graph network, ACOMO, ACO for c101_50 customers

Route optimization via improved ant colony algorithm with graph network (Patil N. Siddalingappa)
410  ISSN:2089-4864

Figure 5. Solution for c101 with 50 customers

4.5. Test case for 75 customers

The test dataset used here is c101 (75) has been used for analysis. To optimize an efficient path 8
number of vehicles are used for the analysis of results. Figure 6 displays the comparison of ACO+graph
network, ACOMO, ACO for c101_75 customers. Figure 7 shows the solution for c101 with 75 customers.
Optimal path is (0, 43, 42, 41, 40, 44, 46, 45, 48, 51, 50, 52, 49, 47, 0), (0, 5, 3, 7, 8, 10, 11, 9, 6, 4, 2, 1, 75,
0), (0, 32, 33, 31, 35, 37, 38, 39, 36, 34, 0), (0, 67, 65, 63, 62, 74, 72, 61, 64, 68, 66, 69, 0), (0, 20, 24, 25, 27,
29, 30, 28, 26, 23, 22, 21, 0), (0, 57, 55, 54, 53, 56, 58, 60, 59, 0), (0, 13, 17, 18, 19, 15, 16, 14, 12, 0) and (0,
71, 70, 73, 0)

C101_75
25000
21352
20196
20000
COST

15000
10640
10000

5000
55 100 94,9 99,5 112
0
C (CNY) CE CS C (CNY) CE CS C (CNY) CE CS
ACO+Graph Network ACOMO ACO
METHODOLOGY ]

Figure 6. Comparison of ACO+graph network, ACOMO, ACO for c101_75 customers

Figure 7. Solution for c101 with 75 customers

Int J Reconfigurable & Embedded Syst, Vol. 12, No. 3, November 2023: 403-413
Int J Reconfigurable & Embedded Syst ISSN:2089-4864  411

4.6. Test case for 100 customers

The test set used here is c101 (100) dataset. The total cost achieved is 1362 by deploying 10
vehicles, to choose the most optimal path. Below mentioned Figure 8 shows comparison of ACO+graph
network, ACOMO, ACO for c101_100 customers. Figure 9 shows the solution for c101 with 100 customers.
Optimal path is (0, 5, 3, 7, 8, 10, 11, 9, 6, 4, 2, 1, 75, 0), (0, 43, 42, 41, 40, 44, 46, 45, 48, 51, 50, 52, 49, 47,
0), (0, 20, 24, 25, 27, 29, 30, 28, 26, 23, 22, 21, 0), (0, 67, 65, 63, 62, 74, 72, 61, 64, 68, 66, 69, 0), (0, 90, 87,
86, 83, 82, 84, 85, 88, 89, 91, 0), (0, 57, 55, 54, 53, 56, 58, 60, 59, 0), (0, 98, 96, 95, 94, 92, 93, 97, 100, 99,
0), (0, 32, 33, 31, 35, 37, 38, 39, 36, 34, 0), (0, 13, 17, 18, 19, 15, 16, 14, 12, 0), (0, 81, 78, 76, 71, 70, 73, 77,
79, 80, 0).

C101_100
35000 32103,4
31203
30000

25000
COST

20000
13562
15000

10000

5000
70 100 130 99,7 135,5 98,6
0
C (CNY) CE CS C (CNY) CE CS C (CNY) CE CS
ACO+Graph Network ACOMO ACO
METHODOLOGY

Figure 8. Comparison of ACO+graph network

Figure 9. Solution for c101 with 100 customers

By comparison of this results with other state-of-art-techniques we can conclude that our proposed
ACO+graph network performs better with other existing methods, by employing a smaller number of
vehicles for to achieve effective performance by minimizing the cost value and the proposed system shows
the effective performance to be countered with real-world problems. Our proposed system shows the total in
reducing the cost management purposes, for modified ACO+graph network algorithm. This technique
outperforms the ACO algorithm.

Route optimization via improved ant colony algorithm with graph network (Patil N. Siddalingappa)
412  ISSN:2089-4864

5. CONCLUSION
This paper here involves combining the features of graph theory and ACO to solve the vehicle
routing problem with time window constraints. Implementing the task solution for optimal path using the
graph-based network, here we try to integrate the VRP problem with time window constraints by merging it
with the graph network. Our proposed model works effectively in comparison with other state of art
techniques. The VRP problem while considering other measures such as minimizing cost, and optimal path to
achieve better performance. A multi-objective optimization model is achieved in a real-time scenario, and an
improved ACO algorithm integrated along with a graph network is considered here for satisfying the
customers by delivering the goods in a specific time window. We have used Pareto optimal principle over
here for multi-objective optimization. Future work to be carried out here suggests that we consider other
parameters like temperature for perishable goods; this also establishes the amount of energy consumed by the
vehicle during transportation of goods from source to destination.

REFERENCES
[1] A. L. Kok, E. W. Hans, and J. M. J. Schutten, “Vehicle routing under time-dependent travel times: the impact of congestion
avoidance,” Computers and Operations Research, vol. 39, no. 5, pp. 910–918, May 2012, doi: 10.1016/[Link].2011.05.027.
[2] Z. Wang, Z. Hu, and X. Yang, “Multi-agent and ant colony optimization for ship integrated power system network
reconfiguration,” Journal of Systems Engineering and Electronics, vol. 33, no. 2, pp. 489–496, Apr. 2022, doi:
10.23919/JSEE.2022.000048.
[3] J. Müller, “Approximative solutions to the bicriterion vehicle routing problem with time windows,” European Journal of
Operational Research, vol. 202, no. 1, pp. 223–231, Apr. 2010, doi: 10.1016/[Link].2009.04.029.
[4] S. N. Kumar and R. Panneerselvam, “A survey on the vehicle routing problem and its variants,” Intelligent Information
Management, vol. 04, no. 03, pp. 66–74, 2012, doi: 10.4236/iim.2012.43010.
[5] K. Braekers, K. Ramaekers, and I. V. Nieuwenhuyse, “The vehicle routing problem: state of the art classification and review,”
Computers and Industrial Engineering, vol. 99, pp. 300–313, Sep. 2016, doi: 10.1016/[Link].2015.12.007.
[6] B. Eksioglu, A. V. Vural, and A. Reisman, “The vehicle routing problem: a taxonomic review,” Computers and Industrial
Engineering, vol. 57, no. 4, pp. 1472–1483, Nov. 2009, doi: 10.1016/[Link].2009.05.009.
[7] R. Elshaer and H. Awad, “A taxonomic review of metaheuristic algorithms for solving the vehicle routing problem and its
variants,” Computers and Industrial Engineering, vol. 140, p. 106242, Feb. 2020, doi: 10.1016/[Link].2019.106242.
[8] G. D. Konstantakopoulos, S. P. Gayialis, and E. P. Kechagias, “Vehicle routing problem and related algorithms for logistics
distribution: a literature review and classification,” Operational Research, vol. 22, no. 3, pp. 2033–2062, Jul. 2022, doi:
10.1007/s12351-020-00600-7.
[9] S. Dobzinski, N. Nisan, and M. Schapira, “Approximation algorithms for combinatorial auctions with complement-free bidders,”
in Proceedings of the Annual ACM Symposium on Theory of Computing, New York, NY, USA: ACM, May 2005, pp. 610–618.
doi: 10.1145/1060590.1060681.
[10] U. Feige, “On maximizing welfare when utility functions are subadditive,” in Proceedings of the Annual ACM Symposium on
Theory of Computing, New York, NY, USA: ACM, May 2006, pp. 41–50. doi: 10.1145/1132516.1132523.
[11] Z. Friggstad, “Multiple traveling salesmen in asymmetric metrics,” in Lecture Notes in Computer Science (including subseries
Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2013, pp. 173–188. doi: 10.1007/978-3-642-40328-
6_13.
[12] Z. Friggstad, M. R. Salavatipour, and Z. Svitkina, “Asymmetric traveling salesman path and directed latency problems,” SIAM
Journal on Computing, vol. 42, no. 4, pp. 1596–1619, Jan. 2013, doi: 10.1137/100797357.
[13] T. Qasim et al., “An ant colony optimization based approach for minimum cost coverage on 3-D Grid in wireless sensor
networks,” IEEE Communications Letters, vol. 22, no. 6, pp. 1140–1143, Jun. 2018, doi: 10.1109/LCOMM.2018.2819643.
[14] J. Lv, X. Wang, and M. Huang, “Ant colony optimization-inspired ICN routing with content concentration and similarity relation,”
IEEE Communications Letters, vol. 21, no. 6, pp. 1313–1316, Jun. 2017, doi: 10.1109/LCOMM.2016.2631515.
[15] Y. Sun, W. Dong, and Y. Chen, “An improved routing algorithm based on ant colony optimization in wireless sensor networks,”
IEEE Communications Letters, vol. 21, no. 6, pp. 1317–1320, Jun. 2017, doi: 10.1109/LCOMM.2017.2672959.
[16] Y. Li, Z. Gao, and J. Li, “Vehicle routing problem in dynamic urban traffic network,” in 8th International Conference on Service
Systems and Service Management - Proceedings of ICSSSM’11, IEEE, Jun. 2011, pp. 1–6. doi: 10.1109/ICSSSM.2011.5959534.
[17] M. López-Ibáñez and T. Stützle, “An analysis of algorithmic components for multiobjective ant colony optimization: A case study
on the biobjective TSP,” in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and
Lecture Notes in Bioinformatics), 2010, pp. 134–145. doi: 10.1007/978-3-642-14156-0_12.
[18] F. Dahan, K. El Hindi, A. Ghoneim, and H. Alsalman, “An enhanced ant colony optimization based algorithm to solve QoS-aware
web service composition,” IEEE Access, vol. 9, pp. 34098–34111, 2021, doi: 10.1109/ACCESS.2021.3061738.
[19] D. Zhang, X. You, S. Liu, and K. Yang, “Multi-colony ant colony optimization based on generalized jaccard similarity
recommendation strategy,” IEEE Access, vol. 7, pp. 157303–157317, 2019, doi: 10.1109/ACCESS.2019.2949860.
[20] M. Xu, X. You, and S. Liu, “A novel heuristic communication Heterogeneous dual population ant colony optimization algorithm,”
IEEE Access, vol. 5, pp. 18506–18515, 2017, doi: 10.1109/ACCESS.2017.2746569.
[21] W. Liu, “Route optimization for last-mile distribution of rural e-commerce logistics based on ant colony optimization,” IEEE
Access, vol. 8, pp. 12179–12187, 2020, doi: 10.1109/ACCESS.2020.2964328.
[22] H. Zhu, X. You, and S. Liu, “Multiple ant colony optimization based on pearson correlation coefficient,” IEEE Access, vol. 7, pp.
61628–61638, 2019, doi: 10.1109/ACCESS.2019.2915673.
[23] M. Liu, X. You, X. Yu, and S. Liu, “KL Divergence-based pheromone fusion for heterogeneous multi-colony ant optimization,”
IEEE Access, vol. 7, pp. 152646–152657, 2019, doi: 10.1109/ACCESS.2019.2948395.
[24] Y. H. Jia, Y. Mei, and M. Zhang, “Confidence-based ant colony optimization for capacitated electric vehicle routing problem with
comparison of different encoding schemes,” IEEE Transactions on Evolutionary Computation, vol. 26, no. 6, pp. 1394–1408, Dec.
2022, doi: 10.1109/TEVC.2022.3144142.

Int J Reconfigurable & Embedded Syst, Vol. 12, No. 3, November 2023: 403-413
Int J Reconfigurable & Embedded Syst ISSN:2089-4864  413

[25] D. Z. Zhu, P. L. Werner, and D. H. Werner, “Design and optimization of 3-D frequency-selective surfaces based on a
multiobjective lazy ant colony optimization algorithm,” IEEE Transactions on Antennas and Propagation, vol. 65, no. 12, pp.
7137–7149, Dec. 2017, doi: 10.1109/TAP.2017.2766660.
[26] J. Liu, Y. Wang, G. Sun, and T. Pang, “Multisurrogate-assisted ant colony optimization for expensive optimization problems with
continuous and categorical variables,” IEEE Transactions on Cybernetics, vol. 52, no. 11, pp. 11348–11361, Nov. 2022, doi:
10.1109/TCYB.2021.3064676.
[27] R. Baldacci, A. Mingozzi, and R. Roberti, “New state-space relaxations for solving the traveling salesman problem with time
windows,” INFORMS Journal on Computing, vol. 24, no. 3, pp. 356–371, Aug. 2012, doi: 10.1287/ijoc.1110.0456.
[28] K. Doerner, M. Gronalt, R. F. Hartl, M. Reimann, C. Strauss, and M. Stummer, “Savings ants for the vehicle routing problem,” in
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 2002, pp. 11–20. doi: 10.1007/3-540-46004-7_2.
[29] E. Demir, T. Bektaş, and G. Laporte, “An adaptive large neighborhood search heuristic for the pollution-routing problem,”
European Journal of Operational Research, vol. 223, no. 2, pp. 346–359, Dec. 2012, doi: 10.1016/[Link].2012.06.044.
[30] D. C. Paraskevopoulos, P. P. Repoussis, C. D. Tarantilis, G. Ioannou, and G. P. Prastacos, “A reactive variable neighborhood tabu
search for the heterogeneous fleet vehicle routing problem with time windows,” Journal of Heuristics, vol. 14, no. 5, pp. 425–455,
Oct. 2008, doi: 10.1007/s10732-007-9045-z.

BIOGRAPHIES OF AUTHORS

Patil N. Siddalingappa working as Assistant Professor in Department of

Information Science, BIET, Davangere. My teaching experience is 11 years and my area of
interest is graph database, big data, database management system, and data mining. He can be
contacted at email: patilns_12@[Link].

Poornima Basavaraj is a professor and Head, Dept. of Information Science and

Engineering at Bapuji Institute of Engineering and Technology, Davangere, Karnataka. She
has received her [Link]. (VTU, Belagavi, 2001) and Ph.D. (Kuvempu University, India,
2012) on “Studies in Fuzzy Graphs” under the Guidance of Dr. V. Ramaswamy. Her research
interests are in the areas of data mining, big data analytics, fuzzy systems and image
processing. She is also principal investigator of VGST funded project “Novel integrated filter-
based approaches for image abstraction to investigate the hidden research issues in image
processing domain” under KFIST-L2 for a grant of 40 lakhs. She can be contacted at email:
poornimateju@[Link].

Preethi Basavaraj working as Assistant Professor in Department of Computer

Science and Engineering, BIET, Davangere. My teaching experience is 8 years and my area of
interest is image processing, artificial intelligence, and machine learning. She can be contacted
at email: Preethib027@[Link].

Premasudha B. Gowramma received her B.E. (Electronics) from Bangalore

University, [Link] (CSE) from JNTU, MCA degree from Bangalore University and Ph.D.
from Dr. MGR University, Chennia. She is at present, working as Professor and head in the
Department of Master of Computer Applications, Siddaganga Institute of Technology,
Tumkur. She has 30 years of teaching experience in the area of computer science. She had
delivered technical talk and conducted workshops in various organizations. She can be
contacted at email: bgpremasudha@[Link].

Route optimization via improved ant colony algorithm with graph network (Patil N. Siddalingappa)