Chapter 20

Minimal Spanning Tree

CONTENT
20.1 A MINIMAL SPANNING TREE
ALGORITHM

23610_ch20_ptg01_Web.indd 1 01/10/14 4:49 PM

 20-2 Chapter 20   Minimal Spanning Tree

In network terminology, the minimal spanning tree problem involves using the arcs of the
network to reach all nodes of the network in such a fashion that the total length of all the
arcs used is minimized. To better understand this problem, let us consider the communica-
tions system design problem encountered by a regional computer center.
The Southwestern Regional Computer Center must have special computer communica-
tions lines installed to connect five satellite users with a new central computer. The telephone
company will install the new communications network. However, the installation is an ex-
pensive operation. To reduce costs, the center’s management group wants the total length
of the new communications lines to be as short as possible. Although the central computer
could be connected directly to each user, it appears to be more economical to install a direct
line to some users and let other users tap into the system by linking them with users already
connected to the system. The determination of this minimal length communications system
design is an example of the minimal spanning tree problem. The network for this problem
with possible connection alternatives and distances is shown in Figure 20.1. An algorithm
that can be used to solve this network model is explained in the following subsection.

20.1 A Minimal Spanning Tree Algorithm

For a network consisting of

A spanning tree for an N-node network is a set of N 2 1 arcs that connects every node
N nodes, a spanning tree to every other node. A minimal spanning tree provides this set of arcs at minimal total arc
will consist of N 2 1 arcs. cost, distance, or some other measure. The network algorithm that can be used to solve the
mini­mal spanning tree problem is simple. The steps of the algorithm are as ­follows:

Step 1.  Arbitrarily begin at any node and connect it to the closest node in terms of
the criterion being used (e.g., time, cost, or distance). The two nodes are
referred to as connected nodes, and the remaining nodes are referred to as
unconnected nodes.
Step 2.   Identify the unconnected node that is closest to one of the connected nodes.
Break ties arbitrarily if two or more nodes qualify as the closest node. Add this
new node to the set of connected nodes. Repeat this step until all nodes have
been connected.
This network algorithm is easily implemented by making the connection decisions directly
on the network.

FIGURE 20.1  COMMUNICATIONS NETWORK FOR THE REGIONAL COMPUTER SYSTEM

5
40
2 30
50
20

Regional 40 3
40

computer 1
10

30
center
4
30

40 20
Miles of
communication
lines required
6
between locations

23610_ch20_ptg01_Web.indd 2 01/10/14 4:49 PM

 20.1   A Minimal Spanning Tree Algorithm   20-3

Referring to the communications network for the regional computer center and arbi-
trarily beginning at node 1, we find the closest node is node 2 with a distance of 20. Using
a bold line to connect nodes 1 and 2, step 1 of the algorithm provides the following result:

5
40
2 50 30
20
40 3

40
1

10
30
4

30
40 20

In step 2 of the algorithm, we find that the unconnected node closest to one of the con-
nected nodes is node 4, with a distance of 30 miles from node 1. Adding node 4 to the set
of connected nodes provides the following result:

5
40
2 50 30
20
40 3

40
1
10

30
4
30

40 20

Repeating the step of always adding the closest unconnected node to the connected
s­ egment of the network provides the minimal spanning tree solution shown in Figure 20.2.
Follow the steps of the algorithm, and see whether you obtain this solution. The minimal
length of the spanning tree is given by the sum of the distances on the arcs forming the
spanning tree. In this case, the total distance is 110 miles for the computer center’s com-
munications network. Note that while the computer center’s network arcs were measured
in distance, other network models may measure the arcs in terms of other criteria such as
cost, time, and so on. In such cases, the minimal spanning tree algorithm will identify the
Can you now find a
optimal solution (minimal cost, minimal time, etc.) for the criterion being considered.
minimal spanning tree for a The computer solution to the regional computer center’s problem is shown in Figure 20.3.
network? Try Problem 2. The minimal spanning tree solution is 110 miles.

23610_ch20_ptg01_Web.indd 3 01/10/14 4:49 PM

 20-4 Chapter 20   Minimal Spanning Tree

FIGURE 20.2  M
 INIMAL SPANNING TREE COMMUNICATIONS NETWORK
FOR THE REGIONAL COMPUTER CENTER

2 30
20
3
1 30

10
4
20

FIGURE 20.3  C
 omputer SOLUTION FOR THE REGIONAL COMPUTER CENTER
MINIMAL SPANNING TREE PROBLEM

**** NETWORK DESCRIPTION ****

6 NODES AND 11 ARCS

ARC START NODE END NODE DISTANCE

--- ---------- -------- --------
1 1 2 20
2 1 3 40
3 1 4 30
4 1 5 50
5 1 6 40
6 2 5 40
7 3 4 10
8 3 5 30
9 3 6 30
10 4 6 20
11 5 6 40

MINIMAL SPANNING TREE

*********************

START NODE END NODE DISTANCE

---------- -------- --------
1 2 20
1 4 30
4 3 10
4 6 20
3 5 30

TOTAL LENGTH 110

23610_ch20_ptg01_Web.indd 4 01/10/14 4:49 PM

 Problems 20-5

NOTES AND COMMENTS

1. The Management Science in Action, EDS De- available at that stage. Following this strategy
signs a Communication Network, describes an at each successive stage will provide the overall
interesting application of the minimal spanning optimal solution. Cases in which a greedy algo-
tree algorithm. rithm provides the optimal solution are rare. For
2. The minimal spanning tree algorithm is consid- many problems, however, greedy algorithms
ered a greedy algorithm because at each stage are excellent heuristics.
we can be “greedy” and take the best action

MANAGEMENT SCIENCE IN ACTION

EDS DESIGNS A COMMUNICATION NETWORK*

EDS was a global leader in information technology EDS’s task was to span the network by find-
services that was originally headquartered in Plano, ing the most cost-effective way to link the 64
Texas, and later acquired by Hewlett-Packard Co. customer locations with each other and with ex-
The company provided hardware, software, com- isting EDS data centers. The arcs of the network
munications, and process solutions to many com- represented communication links between pairs
panies and governments around the world. of nodes in the network. In cases where land
EDS designed communication systems and communication lines were available, the arcs
information networks for many of its customers. consisted of fiber-optic telephone lines. In other
In one application, an EDS customer wanted to cases, the arcs represented satellite communica-
link ­together 64 locations for information flow tion connections.
and ­ communications. Interactive transmission Using cost as the criterion, EDS developed the
involving voice, video, and digital data had to be information network for the customer by solving
accommodated in the information flow between a minimal spanning tree problem. The minimum
the various sites. The customer’s locations included cost network design made it possible for all cus-
approximately 50 offices and information centers tomer locations to communicate with each other
in the continental United States; they ranged from and with the existing EDS data centers.
Connecticut to Florida to Michigan to Texas to
California. Additional locations existed in Canada, *The authors are indebted to Greg A. Dennis of EDS
Mexico, Hawaii, and Puerto Rico. A total of 64 loca- for providing this application.
tions formed the nodes of the information network.

GLOSSARY

Minimal spanning tree  The spanning tree with the minimum length.

Spanning tree    N21 arcs that connect every node in the network with all other nodes
where N is the number of nodes.

PROBLEMS

1.  evelop the minimal spanning tree solution for the following emergency communication
D
network.

23610_ch20_ptg01_Web.indd 5 01/10/14 4:49 PM

 20-6 Chapter 20   Minimal Spanning Tree

Distance
5 6 in miles
1 4
2 3
2

2
3 7
4 3

2 5

2
3 5

7
4 8

2.  he State of Ohio recently purchased land for a new state park, and park planners identi-
T
fied the ideal locations for the lodge, cabins, picnic groves, boat dock, and scenic points of
interest. These locations are represented by the nodes of the following network. The arcs
of the network represent possible road connections in the park. If the state park designers
want to minimize the total road miles that must be constructed in the park and still permit
access to all facilities (nodes), which road connections should be constructed?

2 12
6 12
1 8
6 5 3
8 3 7
11
5 6
8 11 4
15
7
3 9
2
4 4
2
2

1
6

1 4 6
3 5
9 8 10
4 14
2

3 5
7

3 4
3

6
9 13

3.  idwest University is installing a high-speed intranet system. The following network

M
shows the possible network connections among the offices. Distances between offices are
shown in thousands of feet. Develop a design for the intranet system that will connect all
offices. Provide the design that minimizes the total length of connection among the eight
offices.

23610_ch20_ptg01_Web.indd 6 01/10/14 4:49 PM

 Problems 20-7

3 6
2
2

1
2

5
0.5

1.
2.5
1 7 8
3 3

0.5
1 3 1.2
5
4

1.6
1 4

4.  he Metrovision Cable Company just received approval to begin providing cable televi-
T
sion service to a suburb of Memphis, Tennessee. The nodes of the following network
show the distribution points that must be reached by the company’s primary cable lines.
The arcs of the network show the number of miles between the distribution points. De-
termine the solution that will enable the company to reach all distribution points with the
minimum length of primary cable line.

4 3
2 8
9
4 4
3
2
3

4 7 7
3 11
3 6
4
1 4
2

2 10
5
4 3
6

5
4

23610_ch20_ptg01_Web.indd 7 01/10/14 4:49 PM

 Self-Test Solutions and Answers
to Even-Numbered Problems

Chapter 20
2. Connect Distance
1–6 2
6–7 3
7–8 1
7–10 2
10–9 3
9–4 2
9–3 3
3–2 1
4–5 3
7–11 4
8–13 4
14–15 2
15–12 3
14–13 4
Total  37
4. 1–4, 2–3, 3–4, 4–5, 4–6, 6–7, 7–8, 8–9, 9–11, 11–10
Minimum length 5 28 miles

23610_ch20_ptg01_Web.indd 8 01/10/14 4:49 PM