A Review of Ant Colony System Algorithm
Arslan Aslam FA18-MPE-004, Muhammad Hamza Amjad FA18-MPE-008
Department of electrical Engineering, Mirpur University of Science & Technology, Mirpur AJK.
Abstract- Ant colony optimization (ACO) is a meta-heuristic
approach that is used to solve various combinatorial optimization
problems. Real ant colonies behavior is main inspiration for this
meta-heuristic optimization technique. Decision making processes
of ant colonies as they search for food and find the most efficient
routes from their nests to food sources is the main component of
ACO for solution construction and find the most efficient routes
from their nests to food sources. This is a review report on ant
colony optimization with its algorithms in consecutive order and
Estimation-based ACO using TRAVELING SALESMAN
PROBLEM.
Keywords- Ant Colony Optimization, Meta-heuristics, ACO
Algorithms, Estimation-based ACO
I. INTRODUCTION
The routing protocols plays very important role for
solving different problems such as calculating, choosing and
selecting the relevant path for solving the problem efficiently.
There are routing algorithms [2] to find the shortest path and
also to increase the efficiency of the system. Combinatorial
optimization problems are fascinating due to the fact that they
are often easy to state but very difficult to solve. Many of the
problems are such hard that is it is strongly believed that they
cannot be solved. Hence, to practically solve such hard
problems one often has to use approximate methods to find
near-optimal solutions in a relatively short period of time. Such
kind of methods or algorithms are said to be heuristics. To
improve the system they use some problem-specific knowledge
which will help to find best optimal solution. We can say that
meta-heuristic is a set of algorithms that can be used to define
heuristic methods. The use of meta-heuristics has significantly
increased the ability of systems for finding very high quality
solutions to hard combinatorial optimization problems in a
reasonable time [14]
A. COMBINATORIAL OPTIMIZATION
Combinatorial optimization problems involve finding values
for discrete variables such that the optimal solution with
respect to a given objective function is found. According to the
problem order combinatorial optimization problem is either
maximized or a minimized to find best optimal solutions. More
formally, an instance of a combinatorial optimization problem
∏ is a triple (S, F, Ω), where S is the set of candidate solutions,
F is the objective function which assigns an objective function
value F(s) to each candidate solution.
S, and Ω is a set of constraints. The solution is belonging to the
set of candidate solutions that satisfy the constraints Ω are
called feasible solutions. The goal is to find a globally optimal
feasible solution s*.
B. NATURAL BEHAVIOUR OF ANTS
Real ants find their food by finding the best path from a food
source to their nest. While walking along that best part ants
place pheromone on the ground followed by pheromone
deposited by other ants previously. The important part of ACO
algorithms is the combination of promising solutions with
information about the structure of previously obtained good
solutions [10]. To find a best path, a moving ants lay some
pheromone on the ground, so an ant that following it can detect
and decide with high probability to follow. As a result, the
collective behavior that emerges is a form of a positive
feedback loop where the probability with which an ant chooses
a path increases with the number of ants that previously choose
the same path.
II. ACO ALGORITHMS
Ant colony optimization is a meta-heuristic optimization
technique which operates on mechanism in which a colony of
artificial ants cooperates with each other for finding good
solutions to difficult discrete optimization problems.
Cooperation is a key design component of ACO algorithms.
The process is to provide information to a set of relatively
simple agents (artificial ants) that communicate indirectly by
stigmergy, which is an indirect communication mediated by the
environment. Good solutions are an emergent property of the
agents’ cooperative interaction [1]. ACO algorithms can be
used to solve both static and dynamic combinatorial
optimization problems. Static problems are those in which the
characteristics of the problem are given once and for all when
the problem is defined, and do not change while the problem is
being solved [2].
An ACO algorithm is the combination of three procedures:
Construct-Ants-Solutions, Update Pheromones, and Daemon
Actions [8].
a) ConstructAntsSolutionsmanages a colony of ants
that visit adjacent states of the considered problem by
moving through neighbor nodes of the problem’s
construction graph GC. They move by applying a
random probability distribution of local decision
policy that makes use of pheromone trails and
heuristic information [4]. In this way, ants build their
solutions to the optimization problem in a increasing
manner. Once an ant has built a solution, or while the
solution is being built, the ant evaluates the (partial)
solution that will be used by the Update Pheromones
procedure to decide how much pheromone to deposit
[6].
b) Update Pheromones is the process by which the
pheromone trails are modified. The trails value can
either increase, as ants deposit pheromone on the
components or connections they use, or decrease, due
to pheromone evaporation from a practical point of
view. The deposit of new pheromone increases the
probability that those components/connections that
were either used by many ants or that were used by at
least one ant and which produced a very good solution
will be used again by future ants. In this process
evaporation plays a useful part which results in
forgetting to avoid a too rapid convergence of
algorithm to a less optimal solution region and
encourages the algorithm for exploration of new areas
of the search space.
c) Daemon Actions procedure is used to implement
centralized actions which cannot be performed by
single ants. For example activation of local
optimization procedure for finding local best solutions
or going for global procedure to decide either it will
be useful or not to deposit additional pheromone to
make the search process to go for non-local
prospective. As a practical example, the daemon can
observe the path found by each ant in the colony and
select one or a few ants (e.g., those that built the best
solutions in the algorithm iteration) which are then
allowed to deposit additional pheromone on
components they used.
Figure 1 The working of the ACO metaheuristic
Algorithm:
Procedure ACO Metaheuristic
Schedule-Activities
Construct-Ants-Solutions
Update-Pheromones
Daemon-Actions % optional
End-Schedule-Activities
End-procedure
The ACO meta-heuristic is described in pseudo-code. Above
discussed three procedures are managed by ACO meta-
heuristic via the Schedule Activities construct: (1) management
of the ants’ activity, (2) pheromone updating, and (3) daemon
actions. The Schedule Activities does not specify how these
three activities are scheduled and synchronized. In other words,
it the designers if free to specify the way by which these
procedures should interact.
III. Max-Min Ant System
MAX–MIN Ant System (MMAS) introduces four main
modifications with respect to AS. The main modifications
introduced by MMAS with respect to AS are the following
[9].
a) It strongly exploits the best tours found: That is either
the ant that produced best tour in the current iteration
or the best so far is allowed to deposit pheromone. But
this can results in an inactive situation in which all the
ants may follow same path because of excessive
pheromone track on a good but not best optimal tour.
b) To counter this effect, a second modification
introduced by MMAS is that it limits the possible
range of pheromone trail values to the interval
[rmin,fmax].
c) Third, the pheromone trails are initialized to the upper
pheromone track limit, which increases the
exploration of tours at the start of the search due to
small pheromone evaporation rate.
d) Finally, in MMAS, pheromone trails are initialized
again each time when the system approaches inactive
situation [13].
A. Update of Pheromone Trails
After all ants have constructed a tour, pheromones are updated
by applying evaporation as in Ant system followed by the
deposit of new pheromone as follows:
rij ← rij + ∆rijbest,
Where ∆r ijbest=1/cbest, the ant which is allowed to add
pheromone may be either the best-so-far, in which case
∆rijbest=1/cbs, or the iteration-best, in which case ∆rijbest=1/cib,
where Cib is the length of the iteration-best tour. General, in
MMAS implementations both the iteration-best and the best-so-
far update rules are used, in an alternate way [9].
B. Pheromone Trail Limits
In MMAS, lower and upper limit rmin and rmax on the possible
pheromone values on any arc are imposed in order to avoid search
stagnation. In particular, the imposed pheromone trail limits
have the effect of limiting the probability p ij of selecting a city j
when an ant is in city i to the interval
[pmin,pmax] , with 0<pmin≤pij ≤pmax≤1.
Only when an ant k has just one single possible choice for the
next city, that is
│Nik│=1, we have pmin = pmax = 1.
C. Pheromone Trail Initialization and Re-
initialization
At the start of the algorithm, the initial pheromone trails are set
to an estimate of the upper pheromone trail limit. This way of
initializing the pheromone trails, in combination with a small
pheromone evaporation parameter, causes a slow increase in the
relative difference in the pheromone trail levels, so that the
initial search phase of MMAS is very explorative. As a further
means of increasing the exploration of paths that have only a
small probability of being chosen, in MMAS pheromone trails
are occasionally reinitialized. Pheromone trail re-initialization
is typically triggered when the algorithm approaches the
stagnation behavior (as measured by some statistics on the
pheromone trails) or if for a given number of algorithm
iterations no improved tour is found [9].
IV The Travel Salesman Problem
The Travelling Salesman Problem is a problem in
combinatorial optimization where the list of cities and their
distances are given. Our task is to find a shortest possible tour
that visits each city exactly once. It is one of the most
intensively studied problems in optimization. The problem was
first defined in the 1800s by the Irish mathematician W.R.
Hamilton and the British mathematician Thomas Kirkman.It
was however first formulated as a mathematical problem only
in 1930 by Karl Menger.The name Travelling Salesman
Problem was introduced by American Hassler Whiteney.
The travelling salesman problem is the problem of finding a
shortest closed tour which visits all the cities in a given set.
Travelling salesman problem is one of the well-known and
extensively studied problems in combinational optimization.
Now days these traditional methods are not adaptive to real
time. So as to make it dynamic, artificial intelligence techniques
like Genetic algorithm, Neural Network, Ant Colony
Optimization and Particle Swarm Intelligence can be used.
TSP can be defined as: If there are N cities and salesman starts
from his house and have to travel all cities only once then the
order in which salesman tour will cost minimum is our required
solution, The data consist of weights assigned to the edges of a
finite complete graph and the objective is to find a Hamiltonian
cycle.A cycle passing through all the vertices of the graph while
having the minimum total weight. In the Travel Salesman
Problem hamiltonian cycles are commonly called tours. There
are two cases of Travel salesman Problem based on
arrangement of distance between cities.
Symmetric Travel Salesman Problem is the case of the TSP
where the distances between the cities are independent of the
direction or distance between two cities is same in each
direction forming an undirected graph. Whereas in,
Asymmetric Travel salesman Problem distance between two
cities is different or there might be no path exists in both
direction forming a directed graph. Hence number of tours in
STSP and ASTSP are (n-1)!/2 and (n-1)! respectively.
A complete weighted graph G= (N, E) can be used to represent
a TSP, where N is the set of n cities and E is the set of edges
(paths) fully connecting all cities. Each edge (i,j)∈E is assigned
a cost dij, which is the distance between cities i and j. dij can be
defined in the Euclidean space and is given as follows:
dij = [(xi_xj)2 +(𝑦𝑖_𝑦𝑗)2]½
.
Initially, m artificial ants are placed on randomly selected
cities. At each time they move to new cities and modify the
pheromone trail on the edges used this is known as local trail
updating.
When all the ants have completed a tour the ant that made the
shortest tour modifies the edges belonging to its tour known as
global trail updating by adding an amount of pheromone trail
that is inversely proportional to the tour length.
An ant k currently at city i choose to move to city j by applying
the following probabilistic transition rule: The probability rule
between two nodes i and j, called Pseudo-Random-Proportional
Action Choice Rule, depends on two factors: the heuristic and
meta-heuristic.
where τ is the pheromone, η is the inverse of the distance
between the two cities, q is a random variable uniformly
distributed over [0, 1], q0 is a tunable parameter in the interval is useable for solving problems occurring in practical
[0, 1], and J belongs to the candidate list and is selected based
applications.
on the above probabilistic rule.

Local trail updating:

REFERENCES
As the agent moves between cities it updates the amount of
pheromone on the edge as
[1] Pardeep Kaur,” Ant colony optimization: An optimization

ij (t)  (1 ).ij (t 1)  .0 technique”, International Journal of Advance Foundation and

Where, ρ is the evaporation constant. The value τ0 is the initial Research in Science & Engineering (IJAFRSE) Volume 1,

value of pheromone trails and can be calculated as
0 (n.Lnn) -1
Where, n is the number of cities and L the length of the tour
produced by one of the construction heuristics.
Global trail updating:
When all agents have completed a tour the agent that finds the
shortest route updates the edges in its path using the following
equation
ij (t)  (1 ).ij (t 1)  /L + international journal of advanced computer technology, 3 (12),
Where, L+ is the length of the best tour generated by one of the
agents. Fig. 2 shows a pseudo code of the general procedure in
an Ant Colony Optimization meta-heuristic.
V. CONCLUSION
Special Issue, ICCICT 2015.
[2] Vittrio Maniezzo, Alberto Colorni Marco Dorigo "The
Ant System: Optimization by a colony of cooperating agents,"
[3] Debasmita Mukherjee, SriyankarAcharyya,” Ant Colony
Optimization Technique Applied in Network Routing Problem”,
International Journal of Computer Applications (0975 - 8887) Volume
1 – No. 15, 2014.
[4] Khushbu Khatri,Vinit Kumar Gupta,” A Survey Paper on
Solving TSP using Ant Colony Optimization on GPU”, An
December-2014.
[5] Ms. Meenakshi R Patel, Ms. BabitaKubde,” A survey paper on
Ant Colony Optimization Routing algorithm for selecting Multiple
Feasible Paths for Packet Switched Networks” IJCSI International
Journal of Computer Science Issues, Vol. 9, Issue 2, No 3, March 2012.
[6] Vijayalaxmi, Dr. Chandrashekhara, S. Adiga “Ant Colony,

We have successfully reviewed the ant colony for the Optimization Techniques for network Routing Problem”,

optimization problem. It is clearly understood from this paper
how ant colony originated, the basic concepts and the working
of ant colony optimization. Ant colony optimization is one of
the well- known optimization technique. This report presents an
approach for solving travelling salesman problem based on ant
colony algorithm. It can be concluded that the quality of
International Conference on Innovations in Engineering and
Technology (ICIET'2013) Dec. 25-26, 2013.
[7] Ismail Musirin, Muhammad Khayat Idris.“Ant Colony
Optimization (ACO) Technique in Economic Power Dispatch
Problems” , International Multi Conference of Engineers and

solutions depends on the number of ants. The lower number of Computer Scientists 2008 Vol II IMECS 2008, 19-21 March, 2008.
ants allows the individual to change the path much faster. The
[8] Zar Chi Su SuHlaing,May Aye Khine,” Solving Traveling
higher number of ants in population causes the higher growth
Salesman Problem by Using Improved Ant Colony Optimization
of pheromone on edges and thus an individual keeps the path
Algorithm”, International Journal of Information and Education
with higher concentration of pheromone with a high
probability. Technology, Vol. 1, No. 5, December 2011.

The great advantage over the use of particular methods is that [9] Dweepna Garg, Saurabh Shah,” ANT COLONY
ACO algorithm provides relatively good results by a
OPTIMIZATION FOR SOLVING TRAVELING SALESMAN
comparatively low number of iterations and is therefore able to
find an acceptable solution in a short time interval. Therefore it
PROBLEM” International Journal of Computer Science and System

AnalysisVol. 5, No. 1, January-June 2011, pp. 23-29.

[10] BhanuPratap Singh, SohanGarg,”A Characteristics Study of Ant

Colony Optimization Algorithms for Routing Problems”,International

journal Of Advanced Research in Computer Science And Software

Engineering,Volume 3, Issue 3, March 2013.

[11] E.-G. Talbi, O. Roux, C. Fonlupt,” Parallel Ant Colonies for the

quadratic assignment problem”: International journal Of Advanced

Research in Computer Science And Software Engineering, pp 441–

449,2001.

[12] Guo Hong,” A Hybrid Ant Colony Algorithm for Quadratic

Assignment Problem”,the Open Electrical & Electronic Engineering

Journal, 2013, 7, (Supple 1: M5) 51-54.

[13] M. Dorigo, "The ant system: optimization by a colony of

cooperating agents," IEEE Trans Syst. Man, Cybern, vol. 26, pp. 1-13,

1996.

[14] L. Maria Gambardella, "Ant colonies for the quadratic

assignment problem," J. Oper. Res. Soc., vol. 50, pp. 167-176, 1999.

[15] L. Bianchi, L.M. Gambardella et M.Dorigo, An ant colony

optimization approach to the probabilistic traveling salesman

problem, PPSN-VII, Seventh International Conference on Parallel

Problem Solving from Nature, Lecture Notes in Computer Science,

Springer Verlag, Berlin, Allemagne, 2002.