J. Comp. & Math. Sci. Vol.

2 (4), 661-668 (2011)

Locating Sets of Circulant Networks G(n; ±{1, 2 … m})

CHRIS MONICA. M

Department of Mathematics, Loyola College, Chennai 600 034, India

ABSTRACT

Let S = {v1, v2 ... vℓ} be an ordered set of vertices in a graph G.

Then (d(u, v1), d(u, v2) ... d(u, vℓ)) is called the S-location of a
vertex u of G. The set S is called a locating set if the vertices of G
have distinct S-locations. A minimum locating set is a set S with
minimum cardinality. The cardinality of a minimum locating set
of G is called Location Number L(G). This concept has wide
applications in motion planning and in the field of robotics. This
paper determines a locating set of certain classes of circulant
networks.

Keywords: locating set, location number, circulant graphs,

diametrically uniform, digital devices.

1. INTRODUCTION build interconnection networks for

distributed and parallel systems. In the
Ring topologies have frequently seventies, circulant graphs constituted the
been used to implement local and campus basis for designing certain data alignment
area networks as well as other networks for complex memory systems. In
interconnection subsystems for diverse the eighties, several optimizations related to
digital devices. In order to improve their the diameter minimization of degree four
relatively poor performance and circulant graphs, enhancing their
survivability, extensions of rings have been applicability to the design of efficient
considered in the literature such as chordal interconnection networks. Nowadays, the
rings of degree three and four. Circulant analysis and characterization of circulant
graphs have deserved significant attention graphs and their applications still constitute
during the last decades4. The traditional ring active research areas.
and the complete graph topologies belong to
this class of graphs. From a theoretical point Let S = {v1, v2 ... vℓ} be an ordered
of view, there are thousands of publications set of vertices in a graph G. Then f(u) = (d(u,
analyzing their algebraic properties. From a v1), d(u, v2) ... d(u, vℓ)) is called the S-
more practical perspective, circulant graphs location of a vertex u of G. The set S is
have been employed in several applications18. called a locating set if no two vertices of G
In the sixties, these graphs were used to have the same S-location (that is, f(x) ≠ f(y)

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
662 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011)

if x ≠ y). In other words, if S is a locating set 2. TOPOLOGICAL PROPERTIES OF

then it is clear that for each pair of vertices u CIRCULANT NETWORKS
and v of V \ S, there is a vertex w ∈ S such
that d(u, w) ≠ d(v, w)9. A minimum locating Circulant networks have been
set is a set S with minimum cardinality19. known in graph theory for a long time. A
The cardinality of a minimum locating set of circulant undirected graph, denoted by G(n;
G is called location number and is denoted ±{1, 2 … j}), 1 < j ≤ n/2, n ≥ 3 is defined
by L(G). The minimum locating set problem as an undirected graph consisting of the
is to find a minimum locating set. The vertex set V = {0, 1 … n – 1} and the edge
members of a minimum locating set of G are set E = {(i, j): j − i ≡ s(mod n) , s ∈ {1, 2
called landmarks of G. Slater17, 19 describes
… j}}. In other words, a circulant graph is
applications of these concepts when working
Cayley graph whose adjacency matrix is a
with sonar and loran stations.
circulant20. These graphs are also known as
Independently, Harary and Melter8
star-polygon graphs, cyclic graphs,
discovered these concepts as well but used
distributed loop networks. Theoretical
the terms metric basis and metric dimension,
properties of circulant graphs have been
rather than locating set and location number.
studied extensively and are surveyed in1, 5. A
Chartrand et al.6 call a locating set as a
circulant graph is hamiltonian and is a
resolving set. The terminology of Slater is
Cayley graph. In addition, these graphs are
adopted in this paper.
regular, vertex-symmetric, maximally
connected and after an adequate
For the complete graph Kp, the cycle
transformation, they can be represented as
Cp and the complete bipartite graph Km, n,
mesh-connected topologies20.
Harary et al. [8] have shown that L(Kp) = p −
An interesting property of circulant
1, L(Cp) = 2 and L(Km, n) = m + n − 2. This
graphs, that is diametrically uniform11 is
problem has been studied for grids9, trees,
used in this paper.
multi-dimensional grids9, Petersen graphs3,
De Bruijn graphs15, Torus networks12, Benes For each vertex u of a graph G, the
and Butterfly networks13, Honeycomb maximum distance d(u, v) to any other
networks14 and circulant networks2. vertex v of G is called its eccentricity and is
The minimum locating set problem denoted by ecc(u). In a graph G, the
is proved to be NP-complete for general maximum value of eccentricity of vertices of
graphs by a reduction from 3-dimensional G is called the diameter of G and is denoted
matching7, 19. Recently Manuel et al.13 have by λ. Let G be a graph with diameter λ. A
proved that this problem is NP-complete for vertex v of G is said to be diametrically
bipartite graphs. opposite to a vertex u of G, if dG(u, v) = λ. A
In2, circulant networks G(n; ± {1, graph G is said to be a diametrically uniform
2}), G (n; ± {1, 2, 3} have been considered, graph if every vertex of G has at least one
these results are extended to circulant diametrically opposite vertex. The set of
networks G(n; ±{1, 2 … m}), for any m in diametrically opposite vertices of a vertex x
this paper. in G is denoted by D(x).

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011) 663
3. MINIMUM LOCATING SET OF a
CIRCULANT NETWORKS b1 a1

The following result due to Khuller b2 a2

9
et al. is crucial to the result in this paper.
b3 a3
Theorem 1: Let G be a graph with
minimum metric dimension 2 and let {u, v}
b4 a4
∈ V be a metric basis in G. Then the
following conditions are true:
a) There is a unique shortest path between b5 a5
u and v.
b6 a6
b) The degree of each u and v is at most 3.
Figure 2. G(13; ±{1, 2})
Since G(n; ±{1, 2 … m}), 1 < m ≤
n/2 is 2m-regular, the following Lemma is Fix a ∈ V. If || is odd, then the
an application of Theorem 1. mirror through a passes through one of the
members of D(a). This vertex is denoted by
a*. See Figure 1. If || is even, then the
a
b1 mirror through a cut an edge of the circulant
a1
network G(n; ± {1, 2 … m}). This is shown
in Figure 2. So the mirror through the vertex
b2 a2 a divides the vertex set of a circulant
network G(n; ± {1, 2 … m}) into two sets S1
and S2. Let S1 = {a1, a2 ... at}, S2 = {b1, b2 ...
b3 a3
bt}, where t is an integer depending upon the
circulant network. Then ai and bi are images
b4 a4 of each other with respect to the mirror
through a.
b1 a
a5 a1
b5 b2
a* a2

b3 a3
Figure 1: G(12; ±{1, 2})
b4 a4

Lemma 1: L(G(n; ±{1, 2 … m})) > 2, 1 < m

a5
≤ n / 2. b5
a6
b6
Definition 1: For any vertex x of G(n; ± {1, b7 a7
a*
2 … m}), the geometric diameter of C n Figure 3. G(16; ± {1, 2, 3, 4}),
through x is called a mirror. D(a) = {a*, a5, a6, a7, b5, b6, b7}

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
664 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011)

Proposition 1: [11] For any vertex a of G(n; Proof: Let u1, u2 ... um ∈ S1 such that d(a, ui)
± {1, 2 … m}), = d(a, uj), 1 ≤ i, j ≤ m, i ≠ j. Then u1, u2 ... um

 , ௧ିሺ௜ିଵሻ , ௧ିሺ௜ିଵሻ , 1    ,
are consecutive vertices on Cn. The shortest

   02
paths aa1am+1a2m+1 … ami – (m – 1), aa2am+2a2m+2


௧ିሺ௜ିଵሻ , ௧ିሺ௜ିଵሻ , 1    ,
… ami – (m – 2) … aama2ma3m … ami( = a*),

    12
1      are of equal length.

  ,    22


  %
2

 , ௧ିሺ௜ିଵሻ , ௧ିሺ௜ିଵሻ , 1    ,
Let S = {a, a1}. Then
 2
    2,   4,6 … 2  2
d(a1, ami – (m – 1)) = d(a, ami – (m – 1)) – 1


௧ିሺ௜ିଵሻ , ௧ିሺ௜ିଵሻ , 1      1,
= d(a, ami – (m – j)) – 1
 2
= d(a1, ami – (m – j)) – 1
    2,   3, 5 … 2  1
≠ d(a1, ami – (m – j)),

for 2 ≤ j ≤ m, 1    .


where   
 1/2.
Proposition 1 is illustrated in Figure 3 and
Figure 4. But the shortest paths aa2am + 2a2m + 2 … ami –
(m – 2), aa3am + 3a2m + 3 … ami – (m – 3) …

aama2ma3m … ami, 1     are of equal

a
b1 a1
b2
length; the shortest paths aa2am + 2a2m + 2 …
a2 ami – (m – 2) and b1a2am + 2a2m + 2 … ami – (m – 2) are
b3 also of equal length. Include b1∈ S.
a3

d(b1, ami) = d(a, ami) + 1

b4 a4
= d(a, ami – (m – j)) + 1
= d(b1, ami – (m – j)) + 1
b5 a5 ≠ d(b1, ami – (m – j)),
whenever 2 ≤ j ≤ m, 1    .


a6
b6
a7
b7 a8 Further, the shortest paths aa2am +
b8
2a2m + 2 … ami – (m – 2), aa3am + 3a2m + 3 … ami –
Figure 4. G(17; ± {1, 2, 3, 4}), (m – 3), aa4 am + 4a2m + 4 … ami – (m – 4), … aam –
D(a) = { a5, a6, a7, a8, b5, b6, b7, b8}
1a2m – 1 a3m – 1 … ami – 1(= at), 1      are



Lemma 2: Let G(n; ±{1, 2 … m}) be an of equal length. Proceeding as before and
undirected circulant network where n ≡ including a2, b2 … am – 2, bm – 2, am – 1 into S,
0(mod 2m). Let S = {a, a1, b1, a2, b2 … am – 2, it is found that there exists no two vertices in
bm – 2, am – 1} (here b0 = a). Then any two S1 – a* having the same S-coordinates. The
vertices in S1 – a*(or S2 – a*) have distinct proof is similar for any two vertices in S2 –
S-coordinates. a*.

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011) 665

a a1 S = {a, a1, b1, a2, b2 … b(m – 2) / 2, am / 2}, and if

a2
a3 m is odd S = {a, a1, b1, a2, b2 … a(m – 1) / 2, b(m
a4 – 1) / 2}.

Then two vertices in S1 – a*(or S2 – a*) have

distinct S-coordinates.

Theorem 2: Let G(n; ±{1, 2 … m}), n ≥ 8

be an undirected circulant network. Then
(a) 2 < L(G(n; ±{1, 2 … m})) ≤ 2m – 1,
a when n ≡ 0(mod 2m).
a i -1i -2
ai +1 ai (b) 2 < L(G(n; ±{1, 2 … m})) ≤ m + 1,
when n ≡ 2, 3, 4, 5(mod 2m).
Figure 5. Shortest paths aa1a3a5 … ai and aa2a4a6
(c) 2 < L(G(n; ±{1, 2 … m}) ) ≤ ||,
… ai + 1, with i odd when n ≡ k(mod 2m), for any vertex a
and k = 1, 6, 7 … 2m – 1.
The shortest path from the vertex a of G(n;
±{1, 2}) to at is shown in Figure 5. Proof: Let n ≡ 3(mod 2m) and let m be even.
The proof of Lemma 3 and 4 is similar to Let a any vertex of G(n; ±{1, 2 … m}). Then
Lemma 2.
the vertices at distance i (< λ) are ami – (m – j),
Lemma 3: For an undirected circulant bmi – (m – j), 1 ≤ j ≤ m, 1    /. The
network G(n; ±{1, 2 … m}), n ≡ 1, 3, 5 … vertices at, bt are at distance λ from a. Let S
2m – 1(mod 2m). = {a, a1, b1, a2, b2 … b(m – 2) / 2, am / 2}. By
Lemma 3, any two vertices in S1(or S2) have
distinct S-coordinates. But ⁄
and
(a) For n ≡ 1, 7, 9, 11 … 2m – 1(mod 2m), ⁄ , 1    / are equidistant
let S = {a, a1, b1, a2, b2 … am – 1, bm – 1}. from each vertex of S.
(b) For n ≡ 3, 5(mod 2m), let S = {a, a1, b1, Include bm/2. It readily follows that
a2, b2 … b(m – 2) / 2, am / 2}, if m is even and d(a(3m/2)+1, bm/2) ≠ d(b3m/2, bm/2).
S = {a, a1, b1, a2, b2 … a(m – 1) / 2, b(m – 1) / For i > 2,
2} if m is odd.
d(ami – (m/2) + 1, bm/2) = d(ami – (m/2) + 1, a(3m/2)+1) +
Then two vertices in S1(or S2) have distinct d(a(3m/2)+1, bm/2)
S-coordinates. = i – 2 + d(a(3m/2)+1, bm/2).
Similarly, for i > 2,
Lemma 4: For an undirected circulant d(bmi – (m/2), bm/2) = d(bmi – (m/2), b3m/2) +
network G(n; ±{1, 2 … m}), n ≡ 2, 4, 6 … d(b3m/2, bm/2)
2m – 2(mod 2m). = i – 2 + d(b3m/2, bm/2).
(a) For n ≡ 6, 8, 10 … 2m – 2(mod 2m), let Then d(ami – (m/2) + 1, bm/2) ≠ d(bmi – (m/2), bm/2),
S = {a, a1, b1, a2, b2 … am – 1, bm – 1}. for 2    ⁄.
(b) For n ≡ 2, 4(mod 2m), if m is even let Thus 2 < L(G(n; ±{1, 2 … m})) ≤ m + 1.

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
666 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011)

b1 a a1 The proof of remaining cases are similar.

b2 a2 A locating set for circulant networks G(20;
±{1, 2 3, 4, 5, 6}) and G(25; ±{1, 2 3, 4, 5,
b3 a3 6, 7}) are shown in Figure 6 and Figure 7.
b4 a4 b1 a a1
b2 a2
b5 a5 b3 a3

b4 a4
b6
a6 b5 a5

b7 b6 a6
a7
b8 a8 b7
a7
b9 a9 b8
a* a8
b9 a9
Figure 6. G(20; ± {1, 2 3, 4, 5, 6}) with S = {a, a1, b1, a10
b10
a2, b2, a3, b3}}(only edges adjacent to vertex a are b11 b a12 a11
shown) 12

Figure 7. G(25; ± {1, 2 3, 4, 5, 6, 7}) with S = {a, a1,

We next prove the case for n ≡ b1, a2, b2}(only edges adjacent to vertex
a are shown)
1(mod 2m).
For any vertex a of G(n; ±{1, 2 … m}), Theorem 3: Let G be an undirected
vertices ami – (m – j), bmi – (m – j), 1 ≤ j ≤ m, circulant graph G(n; ± {1, m}), n ≥ 8. Then
1     are at a distance i (< λ) from a.


the location number of G satisfies 2 < L(G(n;
The vertices at distance λ from a are ± {1, m})) ≤ m+2 when n ≡ 0, 1, 2 … 2m –
1(mod2m).


 , 

 , 1    . Let
See Figure 8.
  , 
, 
,  ,  …  ,  , 
.
a
Then any two vertices in S1 have
distinct S-coordinates, but in S2 the vertices b2
a2
bt – 1, bt are equidistant from each vertex of S.
The symmetry of the graph implies that any
two vertices in S2 have distinct S′-
coordinates where S′ = {a, a1, b1, a2, b2 …
a4
bm – 2, am – 1, bm – 1}. Include bm – 1 in S. Now
at, bt are the only vertices equidistant from
vertex each of S. By taking am into S,
!
,      1 and !
,    .

Hence d(at, am) ≠ d(bt, am). Since || 

2  ||, 2 < L(G(n; ±{1, 2 … m})) ≤
||. Figure 8. G(18; ± {1, 4}), with its S-coordinates

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011) 667
4. CONCLUSION 7. Garey M. R., Johnson D. S.,
Computers and Intractability: A Guide
In this paper, a locating set for to the Theory of NP Completeness, W.
circulant networks G(n; ±{1, 2 … m}), for H. Freeman and Company, (1979).
any m has been obtained. By imposing 8. Harary F., Melter R. A., “The metric
conditions on locating set, the problem is dimension of a graph”, Ars
under investigation for other interconnection Combinatorica, 2, 191-195 (1976).
networks. 9. Khuller S., Ragavachari B., Rosenfeld
A., “Landmarks in Graphs”, Discrete
REFERENCES Applied Mathematics, 70, 217-229
(1996).
1. Bermond J. C., Comellas F., Hsu D.
10. Melter R. A., Tomcscu I., “Metric
F., “Distributed loop computer
bases in digital geometry”, Computer
networks: A survey”, Journal of
Vision, Graphics, and Image
Parallel and Distributed Computing,
processing, 25, 113-121 (1984).
24, 2-10 (1995).
11. Paul Manuel, Bharati Rajan, Indra
2. Bharati Rajan, Indra Rajasingh, Chris
Rajasingh, Amutha Alaguvel, “Tree
Monica M, Paul Manuel, “On
Spanners, Cayley Graphs and
Minimum Metric Dimension of
Diametrically Uniform Graphs”,
Circulant Networks”, Journal of
LNCS, 2880, 334-345(2003).
Computer and Mathematical Sciences,
12. Paul Manuel, Bharati Rajan, Indra
2(1), 155-162 (2010).
Rajasingh, Chris Monica. M,
3. Bharati Rajan, Indra Rajasingh, J. A.
“Landmarks in Torus Networks”,
Cynthia and Paul Manuel, “On
Journal of Discrete Mathematical
Minimum Metric Dimension”, The
Sciences & Cryptography, 9(2), 263-
Indonesia-Japan Conference on
271 (2006).
Combinatorial Geometry and Graph
Theory, September 13-16, Bandung, 13. Paul D. Manuel, Mostafa I. Abd-El-
Indonesia (2003). Barr, Indra Rajasingh, Bharati Rajan,
4. Boesch F. T., Tindell R., “Circulants “An Efficient Representation of Benes
and their connectivity”, Journal of Networks and its Applications”,
Graph Theory, 8, 487-499, (1984). Journal of Discrete Algorithm, 6(1),
5. Elspas B., Turner J., “Graphs with 11-19 (2008).
circulant adjacency matrices”, Journal 14. Paul Manuel, Bharati Rajan, Indra
of Combinatorial Theory, 9, 297-307 Rajasingh, Chris Monica. M, “On
(1970). Minimum Metric Dimension of
6. Gary Chartrand, Linda Eroh, Mark A. Honeycomb Networks”, Journal of
Johnson, Ortrud R. Oellermann, Discrete Algorithm, 6(1), 20-27 (2008).
“Resolvability in graphs and the 15. Paul Manuel, Bharati Rajan, Indra
metric dimension of a graph”, Discrete Rajasingh, Cynthia J. A., “On
Applied Mathematics, 105, 99-113 Minimum Metric Dimension of De
(2000). Bruijn Graph”, Proc. National

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)
668 Chris Monica. M, J. Comp. & Math. Sci. Vol.2 (4), 661-668 (2011)

Conference on Computational properties and communication

Intelligence, India, 40-45, March algorithms”, Discrete Applied Math.,
2005. 77, 281-305 (1997).
16. Slater P. J., “Leaves of trees”, Congr. 19. Teresa W. Haynes, Stephen
Number., 14, 549-559 (1975). Hedetniemi, Slater P. J., “Funda-
17. Slater P. J., “Dominating and mentals of Domination in Graphs”,
reference sets in a graph”, Journal of CRC Press, 1, (1998).
Math. Phys. Sci., 22, 445-455 (1988). 20. Xu J., Topological Structures and
18. Stojmenovic I., “Multiplicative Analysis of Interconnection Networks,
circulant networks. Topological Kluwer Academic Publishers (2001).

Journal of Computer and Mathematical Sciences Vol. 2, Issue 4, 31 August, 2011 Pages (581-692)