World Academy of Science, Engineering and Technology 53 2009

Tabu Search Approach to Solve Routing Issues

in Communication Networks
1
Anant Oonsivilai , 2Wichai Srisuruk , 3Boonruang Marungsri , 4Thanatchai Kulworawanichpong

 In [4] the problem of optimal routing in shortest-path data

Abstract—Optimal routing in communication networks is a networks is addressed using open shortest path first (OSPF),
major issue to be solved. In this paper, the application of Tabu Search routing information protocol (RIP), and other interior gateway
(TS) in the optimum routing problem where the aim is to minimize protocols (IGPs). A proposed combinatorial algorithm for
the computational time and improvement of quality of the solution in solving the optimal shortest-path routing problem consists of
the communication have been addressed. The goal is to minimize the finding the link-distance metric assignment to all links in a
average delays in the communication. The effectiveness of Tabu
network.
Search method is shown by the results of simulation to solve the
shortest path problem. Through this approach computational cost can
In [5], a description of the Hopfield/Tank model used for
be reduced. routing problem is shown.
In [6] a method that combines Hopfield networks (HN) and
Keywords—Communication networks, optimum routing a genetic algorithm (GA) to solve the problem of optimal
network, tabu search algorithm, shortest path. routing in computer networks is used.
. In [13]-[14] neural network approach is applied to solve the
shortest path problem.
I. INTRODUCTION In all methods, it has been reported that the performance has
been improved. Limited global search, long computation time
I N a communication networks, particularly, a packet
switched communication network, routing is a major issue
that has a significant impact on the system performance. The
to solve large-scale problem, are the main problem of the most
above listed algorithms. The application of Tabu Search to the
routing problem has been motivated by the idea of taking
main objective of the routing optimization algorithm is to find advantages of being very successful in achieving near-optimal
the best path for data transmission within a short time to (and sometimes optimal) solutions to a variety of hard
satisfy the demand. By optimizing the routing, we improve problems [7].
network throughput defined by the mean packet delay. This This paper is organized as follows: section I presents the
paper deals with shortest path computation in real time. In this introduction. Section II presents the details on the problem
context, shortest path is referred to the minimization of the formulation. Tabu Search algorithm background is presented
propagation distance, number of nodes, link capacity, or in section III. Section IV discusses the results obtained.
congestion level of the link [1-2]. Section V gives relevant conclusion.
The routing problem is one of the important issues to be
studied in communication networks. Several research interests II. PROBLEM FORMULATION
in searching for solution have been studies.
In [3] a method of self-healing communication networks Please A given directed graph ‫ ܩ‬ൌ ሺܸǡ ‫ܧ‬ሻ and set of ܰ
that allow re-routing of demands through switching processes nodes of graphܸሺ‫ܩ‬ሻ ൌ ሼ݊௜ ሽ݅ ൌ ͳǡʹǡ ǥ ǡ ܰ, and that of edges
at designated nodes is presented. In this paper, it is shown how connecting nodes inܸ is depicted in Fig. 1. Edge set of
network utilization, demand throughput and reliability of such ‫ܧ‬ሺ‫ܩ‬ሻ ൌ ሼሺ݅ǡ ݆ሻȁ݅ǡ ݆ ‫ܸ א‬ሽ representing edges of a graph‫ܩ‬.
networks can be studied simultaneously to achieve an optimal Corresponding to each edge there is a cost ‫ܥ‬௜௝ representing the
design. The use of linear programming approximation to the cost from node ݊௜ to node݊௝ .
multi-commodity network flow formulation to solve the In this problem, the total length is equal to the sum of all
problem for medium and large network sizes in moderate costs of links within the path. The conditions to be satisfied in
computational times has been addressed. this problem of finding the shortest path from the source to the
destination are: each node is available only once in a path and
one node at a time. We employ tabu search to solve the
problem of finding the shortest path of a given communication
network between a source and destination nodes.
All authors are with the Alternative and Sustainable Energy Research Unit,
Power and Control Research Group, School of Electrical Engineering,
Institute of Engineering, Suranaree University of Technology, 111 University
Street, Muang District, Nakhon Ratchasima, 30000 Thailand.
1
[email protected] , [email protected], [email protected] ,
4
[email protected]

1174
 World Academy of Science, Engineering and Technology 53 2009

shows the results of optimal paths from source node 1 and

destinations node 4 and node 5. The optimal path for source
node 1 and destination node 4 is: node 1-to- node 2 -to– node
4 (1-2-4). The optimal for source node 1 and destination node
5 is: node 1-to-node 2 –to-node 5 (1-2-5).

Start

Initialize TS ሺ‫ݏ‬Ǣ ݇ ൌ ͳǢ ̴݇݉ܽ‫ݔ‬ሻ

Fig. 1: Eight (8) node, Twelve (12) edge network model

.

III. TABU SEARCH ALGORITHM

Identify neighbor set ܰሺ‫ݏ‬ሻ
The Tabu Search metaheuristic has been defined in [8, 15,
Identify aspirant set ܶሺ‫ݏ‬ǡ ݇ሻ
16]. The basic idea behind tabu search is that, adding short-
term memory to local search, improves its ability to locate
optimal solutions. It is an is an iterative search that starts from
some initial feasible solution and attempts to determine the
best solution in the manner of a hill-climbing algorithm. The ‫ ݏ‬ᇱ ‫ܰ א‬ሺ‫ݏ‬ǡ ݇ሻ ൌ ሼܰሺ‫ݏ‬ሻ െ ܶሺ‫ݏ‬ǡ ݇ሻሽ ൅ ‫ܣ‬ሺ‫ݏ‬ǡ ݇ሻǤ
Choose the best ‫ݏ‬Ԣ ݇ ൌ ݇ ൅ ͳ
algorithm keeps historical local optima for leading to the near-
global optimum fast and efficiently [8].
The idea is to make sure that there will be no same local
optimum happening again in the process. Backtracking
process starts from stepping back to some local optimum in YES
tabu list and then searching a new optimum in different
Does ‫ݏ‬Ԣ converge?
directions. Backtracking is performed when the backtracking
criterion (BC) is encountered. In this paper, we address the use
of tabu search for solving the routing problem. More details on End NO
tabu search can be found in [8-11]. In these publications, it is
stated that the main two components of the tabu search
algorithm are the tabu list restrictions and the aspiration ݇ ൌ݇൅ͳ
criterion.
In applying the Tabu Search algorithm, to solve a shortest
path routing problem, the basic idea is to choose a feasible NO
solution at random and then obtain a neighbor to this solution.
݅‫ ݇ ݏ‬ൌ ݇௠௔௫ ǫ
During these search procedures the best solution is always
updated and stored aside until the stopping criterion is
satisfied [12]. Fig. 2 shows the flow chart for the basic Tabu
Search algorithm process, followed by its pseudo-code in Fig.
3.
End

IV. RESULTS
The Tabu Search method has been used to perform an
example of 5-node network shown in Fig. 3 used by [6, 13]. In
Fig. 2: Flow chart for the basic Tabu Search process [8].
[6], Hopfield-genetic approach was applied for the similar
problem while in [13-14] neural network approach was used.
Both methods converge to an optimal solution. Table 1 shows
the 5-node, 8-links, network and their random costs. Table 2

1175
 World Academy of Science, Engineering and Technology 53 2009

Initiate ݇ ൌ ͳ. Table 2: The source destination optimal path for the five (5)-node, eight (8)-
links network
generate initial solution
WHILE the stopping condition is not met Source Destination Optimal path Cost
DO 1 1-2-4
4 0.793612
Identify ܰሺ‫ݏ‬ሻǤ(Neighbourhood set)
1 1-2-5
Identify ܶሺ‫ݏ‬ǡ ݇ሻ. (Tabu set) 5 0.799612
Identify ‫ܣ‬ሺ‫ݏ‬ǡ ݇ሻ. (Aspirant set)
Choose the best ‫’ݏ‬ The similar approach was applied to an eight (8)-node, twelve
‫ݏ‬Ԣ ‫ܰ א‬ሺ‫ݏ‬ǡ ݇ሻ  ൌ  ሼܰሺ‫ݏ‬ሻ  െ ܶሺ‫ݏ‬ǡ ݇ሻሽ ൅ ‫ܣ‬ሺ‫ݏ‬ǡ ݇ሻǤ (12)-links, network shown in Fig. 4 with link costs presented
‫ ݏ‬ൌ ‫’ݏ‬. by Table 3.
݇ ൌ ݇ ൅ ͳ.
END WHILE

Fig. 3: Tabu Search process pseudo-code.

Fig.4: An eight (8)-node, twelve (12)-links network

Table 3: The links and their random costs for eight (8)-node, twelve (12) links
network

Fig. 3: A five (5)-node, eight (8)-links network

Link Cost
1-to-2 0.7919
1-to-7 0.9218
Table 1: The links and their random costs for five (5)-node, eight (8)-links
network 1-to-8 0.7382
2-to-3 0.1763
Link Cost
2-to-7 0.4057
1-to-2 0.377483
3-to-5 0.9355
1-to-3 0.76667686
3-to-8 0.9169
2-to-3 0.03693263
4-to-5 0.4103
2-to-4 0.415129
4-to-6 0.8936
2-to-5 0.42212904
4-to-7 0.0579
3-to-4 0.45564041
5-to-6 0.3529
3-to-5 0.03693263
6-to-8 0.8132
4-to-5 0.87610209

1176
 World Academy of Science, Engineering and Technology 53 2009

[12] B. Marungsri, B., Meeboon, N., and Oonsivilai, A. “Dynamic Model

Identification of Induction Motors using Intelligent Search Techniques
with taking Core Loss into Account” , WSEAS Trans. on Power
Table 4: The source destination optimal path for the eight (8)-node, twelve
Systems, Vol. 1, No. 8, August 2006, pp. 1438 – 1445.
(12) links, and network
[13] Ali, M.K., and Kamoun, F. “Neural Network for Shortest Path
Computation and Routing in Computer Networks.” IEEE Transaction
Source Destination Optimal path Cost Neural Networks, Vol. 4, pp. 941-953, Nov. 1993.
1 4 1-7-4 0.9797 [14] Araujo. F. and Ribeiro, B. “Neural Networks for Shortest Path
1 1-7-4-5 Computation and Routing in Computer Networks.” IEEE Transactions
5 1.39 on Neural Networks, Vol. 4, Issue 5, pp. 941–953, November 1993.
2 2-7-4-5-6 [15] Glover, F. “Future Paths for Integer Programming and Links to Artificial
6 1.2268
Intelligence”, Comp. Oper. Res., Vol. 13, pp. 533-549, 1986.
3 3-2-7
7 0.582 [16] Hansen, P. “The Steepest Ascent Mildest Descent Heuristic for
6 6-5-4-7 Combinatorial Programming”, Congress on Numerical Methods in
7 0.8211 Combinatorial Optimization, Capri, Italy, 1986.

V. CONCLUSION
It has been observed that the performance of tabu search is
robust and fast. In this paper a novel strategy for routing in
computer networks has been presented. Results obtained for
the test cases suggest that the simulated with Tabu Search
algorithm allocation is a stout approach for this problem, and
was always able to find good quality possible solutions.

REFERENCES

[1] Baransel, C., Dobosiewicz, W., and Gburzynski, P. "Routing in Multi

Packet Switching Networks: Gbps Challenge." IEEE Network, May-
June 1995, pp. 123-138.
[2] Bertsekas , D., and Gallager, R. " Data Networks", Second Eddition,
Prentice-Hall, 1992.
[3] Saniee, I. “Optimal Routing Designs in Self-healing Communications
Networks.” International Transactions in Operational Research, Vol. 3,
Issue 2, pp. 187 – 195, Aug 2006.
[4] Ramakrishnan, K. G. and Rodrigues M. A. “Optimal Routing In
Shortest-Path Data Networks.” Bell Labs Technical Journal, Vol. 6 Issue
1, pp. 117 – 138, Aug 2002.
[5] Liu Rong , L. Ze-Rin , L. and Zheng, Z. “Neural Network Approach for
Communication Network Routing Problem.” IEEE TENCON '93/
Beijing, pp. 649 -662. 1993.
[6] Hamdan M. and El-Hawary, M.E. “Hopfield-Genetic Approach for
Solving the Routing Problem In computer Networks.” Proceedings of
the 2002 IEEE Canadian Conference on Electrical and Computer
Engineering, pp. 823-827.
[7] Ikonomovska, E., Gjorgjevik, D., and Loskovska, S. “Using Data
Mining Technique for Coefficient Tuning of an Adaptive Tabu Search.”
Proceedings of International Conference on Computer as a Tool
EUROCON 2007, pp. 706-713, Warsaw, September 9-12, 2007.
[8] Anant Oonsivilai, A. and Pao-La-Or, P. “Application of Adaptive Tabu
Search For Optimum PID Controller Tuning AVR System” WSEAS
Transactions on Power Systems Vol. 3, Issue 6, Pp. 495-506, June 2008.
[9] Puangdownreong, D., Sujitjorn, S., and Kulworawanichpong, T.
“Convergence Analysis of Adaptive Tabu Search”, International Journal
of ScienceAsia, Vol. 38, No.2, 2004, pp. 183-190.
[10] C. Chaiyaratana, C., and Zalzala, A.M.S. “Recent Developments in
Evolutionary and Genetic Algorithms: Theory and Applications”,
Second International Conference on Genetic Algorithms in Engineering
System: Innovations and Applications, 2-4 September 1997, pp. 270 –
277.
[11] Puangdownreong, D., Areerak, K., Srikaew, A., Sujitjorn, S., and
Totarong, P. “System Identification via Adaptive Tabu Search”, IEEE
International Conference on Industrial Technology, Vol. 2, 11-14
December 2002, pp. 915 – 920.

1177