Communications and Network, 2013, 5, 171-175
http://dx.doi.org/10.4236/cn.2013.53B2033 Published Online September 2013 (http://www.scirp.org/journal/cn)
A Swarm Intelligence Networking Framework for Small
Satellite Systems
Zijing Chen1, Yuanyuan Zeng1,2
1The Academy of Satellite Application, Beijing, China
2School of Electronic Information, Wuhan University, Wuhan, China
Received June, 2013
ABSTRACT
Recent development of technologies and methodologies on distributed spacecraft systems enable the small satellite
network systems by supporting integrated navigation, communications and control tasks. The distributed sensing data
can be communicated and processed autonomously among the network systems. Due to the size, density and dynamic
factors of small satellite networks, the traditional network communication framework is not well suited for distributed
small satellites. The paper proposes a novel swarm intelligence based networking framework by using Ant colony opti-
mization. The proposed network framework enables self-adaptive routing, communications and network reconstructions
among small satellites. The simulation results show our framework is suitable for dynamic factors in distributed small
satellite systems. The proposed schemes are adaptive and scalable to network topology and achieve good performance
in different network scenarios.
Keywords: Small Satellite Systems; Ant Colony Optimization; Swarm Intelligence; Network Reconstruction
1. Introduction
Recently, research on distributed small satellites attracts
a lot of focus. Small satellite has low manufacturing and
launch costs, while the network functionality is limited
for distributed and dynamic topology. The low data- rate
and limited transmission windows toward the ground
station bring challenges for small satellite communica-
tions. The small satellite systems enable complex sensing
tasks such as multipoint observation, co-observation, etc
[1]. Grouping is an efficient method to cluster small sat-
ellites and manage them. The grouping of small satellites
is usually executed according to the spatial characters.
Some realistic examples of grouping include formation
flying and constellation of small satellites for missions.
In space related networking, the disruptive tolerant char-
acteristic is a hot topic. The reconstruction capability
with recovery is necessary especially for the intermittent
network situation by changing orbits for small satellites
in missions. The complexity of distributed small satellite
systems has led the research interests of methodologies
in self-organized and self-adaptive network architecture
and configuration. In this paper, we present a novel
swarm intelligence network framework for distributed
small satellite systems by using ant colony optimization
(ACO). The objective is to build a heuristic network
communication framework that achieves good resource
allocation with self-adaptive reconstructing ability.
Section II presents the background and related work.
Section III is the system model. In Section IV, we pre-
sent the ant colony optimization based small satellites
network framework. Section V is simulations and
evaluations. Section VI concludes this paper.
2. Background and Related Work
Recent research on the small satellite systems illustrate
the advantages of distributed and inexpensive small sat-
ellites to design, build, launch and operate when com-
pared with the traditional satellite missions. Burlacu et al
[2] analyze the small satellite systems on the challenging
open issues. They propose that the future hop research
topic on small satellites routing; and suggest utilize the
communication schemes of ad hoc networks. The dis-
tributed small satellite system mission examples include
UWE-1 project. It was initiated in 2004 to develop a
pico-satellite platform for telecommunication experi-
ments [3]. The successor UWE-2 was developed in 2007
to demonstrate the small satellite capabilities in terms of
attitude determination[4]. The future prospects of UWE
project include the increase of robustness and rapid re-
sponse by using satellite formations and swarms for ad-
vanced space missions.
*Corresponding author. Many researchers envision satellite networking topol-
C
opyright © 2013 SciRes. CN

Z. J. CHEN, Y. Y. ZENG
172
ogy technologies as a promising approach to realize new
innovative collaborative space missions. The main to-
pology technologies of distributed satellite systems in-
clude formations, constellations, clusters and swarms.
Formation is a cost-effective system approach that
solves the problem of the size and mass for a single sat-
ellite with limited resources. Formation maintains the
relative distance between entities in the same or very
similar orbit. Braukhane et al [5] propose a scalable for-
mation flying system named FormSat, which consists of
transparent communication satellites connected via in-
ter-satellite links to a hub satellite by allowing in-orbit
resources management and optimization.
Constellation is a multi-satellite system with the enti-
ties distributed in different orbits. The Global Positioning
System (GPS) and Iridium satellite constellation are two
classical examples of constellation. Surrey Satellite
Technology Ltd. designed, built and launched the
world’s first constellation to provide daily global earth
observation coverage with the telecommunication and
earth disaster monitoring constellation[6].
Clusters and swarms are methods for small satellite to
cooperate with each other and achieve a common goal.
Qin et al propose a weight based dominating set cluster-
ing algorithm for small satellite networks to group the
satellites by the spatial characteristics. Swarm schemes
are originated from flocks, schools, etc. The swarm based
networking topology help the communications and tasks
achieve global optimization from the convergence of
individual behaviors.
Recently, another emerging topic in distributed space
missions named fractional spacecraft is proposed. The
American Defense Advanced Research Projects Agency
announced F6 system program aiming at the feasibility
and benefits of fractionated satellite that fulfilled with a
fractionated cluster of free-flying, wireless intercon-
nected modules[7].
Our work will utilize the swarm intelligence and build
a resource optimized network framework that enable
low-cost communications and reconstructing capability
for in-orbit satellites considering intermittent network
scenarios.
3. System Model
In the paper, we consider about the Low Earth Orbit
(LEO) small satellite. As the small satellite fly close to
earth, usually between 500 km and 850 km altitude, we
approximate the orbit as a circular model other than the
ellipse orbit to simply the problem formulation. In this
case, the satellite moves in an even circular orbit model.
An ad hoc networking can be used for small satellite
network framework building while each satellite is a
node. A specific ID is used to denote it. To achieve the
resource optimization when building the framework,
each node collects the neighborhood information by pe-
riodic HELLO messages: the neighbor satellite ID,
neighbor inclination Inc, right ascension of the ascending
neighbor Ra, the orbit information Orb, the time of dura-
tion with the neighbor Dur, and the neighbor resource
metric Res.
In the networking framework, each node will try to
search the optimized node as the neighborhood and form
a resource optimized network architecture for communi-
cations in missions. According to this, the stability is one
of the main factors in building the framework. The rela-
tive static satellite is preferred to form neighborhood
with temporal and spatial stability. The same orbital node
or the node with similar orbit is considered. For the satel-
lite within a short distance, we can still approximate con-
sider their position as static, i.e., for two satellite i and j,
if dij<dmax within Dur(i, j), we say i and j keeping stabil-
ity. The distance can be calculated by:
22
2cos
ijiji jij
dRRRR

 (1)
In Formula (1), Ri and R
j denote the orbit radius of
node i and j, respectively. θij is the angle of the two nodes
towards the earth.
Beyond the stability, the resource optimization is an-
other factor in the framework. The neighbor resource
metric Res can be defined as the weighted combinations
of power, link data rate, and the maximal time window
length toward the earth ground station with the unit as
time slots.
Our motivation is to build a ACO-based small satellite
networking topology and communication routing frame-
work to form the network topology and communication
links among satellites that achieve the global resource
optimization.
4. ACO Based Network Framework Schemes
4.1. Network Framework Using ACO
In this paper, we utilize Ant Colony Optimization[8] to
form and provide a self-organized network framework
for communication among small satellite. The framework
provides autonomous inter-satellite communications and
self-adaptive topology with reconstruction capability
when part of the link fails. To achieve this, the virtual
agents (artificial ants) in ACO algorithm is utilized to
search the optimal solutions based on decentralized co-
ordination among the artificial ants. During the candidate
solution searching process, each artificial ant k on current
hop i will decide the outgoing link toward node j accord-
ing to Formula (2).
()
()
0
p
ij ij
p
kil il
ij lNs
if lNs
p
otherwise












 (2)

Z. J. CHEN, Y. Y. ZENG 173
In which, τij is the pheromone associated with the out-
going link (i, j). N(sp) is the set of feasible next hop ex-
ploration set. In the packet routing, the next hop will be a
node not visited yet to prevent loops. The parameter α
and β control the relative importance of τij and ηij. ηij is
the heuristic information on link (i, j), which is given by
Formula (3). In which, γ is the weight factor parameter.
1(1 )
ij j
ij
Res
d
   (3)
After the local search for the current hop, the ant up-
dates local pheromone of the recent link traversed ac-
cording to Formula (4). In which, φ is the decay coeffi-
cient between 0 and 1. τ0 is the initial setting value of the
pheromone, which is defined by upper layer application.
In this way, we can prevent the searching stop from the
local optimization.
0
(1 )
ij ij

   (4)
After finishing a network-scale search process, the
pheromone can be updated according to the global solu-
tions, as shown in Formula (5). This pheromone update
process helps the convergence of solutions.
(1 )
ij ij
best
C

   (5)
Where, Cbest is the total best path found from the cur-
rent iteration with the optimized weighted combination
of stability and resource metrics.
The ACO based network framework construction
scheme can be illustrated as Algorithm 1. Line 1-Line 6
is network initialization with network model, pheromone
and route path cost. Line 7-Line 15 is the ACO-based
networking process. It consists many steps. In each step,
it includes the route searching, local pheromone update
and global pheromone update. Line 10 is the ant local
route searching. The searching probability is decided by
Formula (2). Line 11 is the local pheromone update ac-
cording to the search route. Line 13-Line 14 is the global
update according to the current best route path when the
complete route path is found.
Algorithm 1: ACO based network framework con-
struction
1 Initialize ()
2 {
3 Network model G(V,E);
4 Pheromone matrix for each link as T=[τij]N×N with
τijτ0;
5 C
best∞;
6 }
7 for each ACO step do
8 update local neighborhood and information;
9 for each artificial ant k=1 to m do
10 select the next hop j according to Formula
(2);
11 update τij according to Formula (4);
12 endfor
13 update Cbest of the current step;
14 update τij for links on the best path according to
Formula (5);
15 endfor
5. Simulations
Our simulation is implemented by C++. The simulation
scenario is deployed in a 2-D plane, i.e., assume that all
the satellites have the same inclinations and right ascen-
sions of the ascending node. Satellites are randomly dis-
tributed from 500 km to 550 km altitude orbit. The dis-
tance between two adjacent orbits is set to 10 km, so
there are 6 orbits in total. The orbit of satellite in our
simulation is illustrated as shown in Figure 1. Figure 2
shows the random position deployed in orbit of 100 small
satellites. According to the satellite coverage area and
earth radius, we choose dmax to 55 km. We can get the
generated possible links from the parameter dmax with
distance between two small satellites that no more than
dmax. In the simulations, we consider the grounding capa-
bilities of the satellite node (such as orbit height and dis-
tance) and use it as a metric in routing and topology for-
mation.
Figure 1. The orbit of LEO small satellite.
-600 -400 -2000200400600
-600
-400
-200
0
200
400
600
Coordinate Y
Coordinate X
Figure 2. The random 100 satellite nodes in orbits.

Z. J. CHEN, Y. Y. ZENG
174
We then simulate the performance of our framework
schemes in networking topology. The connection ratio is
defined as the average connection degree between two
nodes in the network by using our ACO based frame-
work schemes to form the communication topology. We
simulation the connection ratio as increases the number
of satellite number in the network system. The results are
shown in Figure 3. It shows that our framework can
adaptively build the network topology with good com-
munication connectivity especially as the node number
grows big enough. The framework schemes help to
maintain a network communication structure with good
connectivity that benefits the distributed satellite com-
munications among missions. Since each satellite may
change the orbit around the earth to execute the missions.
We then simulate the network framework with random
changing orbit of node #5 till node #15. The results are
shown in Figure 4. The results show the framework
scheme can achieve a good connection ratio when the
node changes the orbit. Because the scheme is intelligent
to make adaptive re-routing and form the self-organized
topology according to communication requirement.
20406080100 120 140 160 180 200 220
60
65
70
75
80
85
90
95
100
105
with dmax =55km
Connection Ratio (% )
Node number (#)
Figures 3. The connection ratio with dmax=55 km.
20406080100 120 140 160180 200 22
0
88
90
92
94
96
98
100
Connection Ratio (%)
Node Number (#)
with dmax =55km
Figure 4. The connection ratio when changing the orbit.
6. Conclusions and Future Work
Distributed small satellite systems are the main focus for
the future deep space detection and monitoring applica-
tions. The networking topology and reconstruction is
challenged for the dynamic and varied space application
scenarios. The traditional networking framework is not
well suited for the space application-specific situations.
We then propose a ACO based swarm intelligence net-
work framework in the paper to form the heuristic,
self-adaptive and self-organized network structure. The
framework scheme is with fast reconstruction capability
that guarantees the good performance of the small satel-
lite systems.
We make the preliminary simulations to testify the
performance of the framework schemes. We will simu-
late the framework scheme in different network scenarios
and parameters in our future work.
7. Acknowledgements
This work was supported by the Open Research Fund of
Scientific Research Foundation for the Returned Over-
seas Chinese Scholars, State Education Ministry, sup-
ported by the Open Research Fund of The Academy of
Satellite Application, Chinese National Science Founda-
tion Project (no. 61103218), The Natural Science Foun-
dation of Jiangsu Province Youth Project (No. BK-
2012200), Hubei Province National Science Foundation
Project (no.2011CDB446).
REFERENCES
[1] J. R. Wertz and W. J. Larson, Space Mission Analysis and
Design, Space Technology Library, Microcosm Press and
Springer, 3rd edition, 2007.
[2] M.-M Burlacu and P. Lorenz, “A Survey of Small Satel-
lites Domain: Challenges, Applications, and Communica-
tion Key Issues,” ICST’s Global Community Magazine,
2010, pp. 1-11.
[3] M. Schmidt and F. Zeiger, “Design and Implementation
of In Orbit Experiments for the Pico Satellite UWE-1,”
International Astronatical Congress, 2007.
[4] M. Schmidt, K. Ravandoor, O. Kurz, S. Busch and K.
Schiling, “Attitude Determination for the Nano-Satellite
UWE-2,” Space Technology, Vol. 28, 2009, pp. 67-74.
[5] M. Arza, M. Bacher, M. Calaprice, H. Fiedler, V. Koehne,
H. R. McGuire and J. J. Rivera, “FormSat, A Scalable
Formation Flying Communication Satellite System,”
IEEE Aerospace Conference, 2010, pp. 1-18.
[6] P. Stephens, J. Cooksley, A. Curlel, L. Boland, S. Jason, J.
Northham, A. Brewer, J. Anzaichi, H. Newell, C. Un-
derwood, S. Machin, W. Sun and M. Sweeting, “Launch
of The International Disaster Monitoring Constellation:
The Development of a Novel International Partership in
Space,” Proceedings of RAST: Recent Advances in Space
Technologies, 2003, pp. 525-535.

Z. J. CHEN, Y. Y. ZENG
175
[7] J. M. Lafleur and J. H. Saleh, “Exploring the F6 Fraction-
ated Spacecraft Trade Space with GT-FAST,” Proceed-
ings of AIAA, 2009.
[8] M. Dorigo, M. Birattari and T. Stutzle, “Ant Colony Op-
timization – Artificial Ants as a Computational Intelli-
gence Technique,” IEEE Computational Intelligence
Magazine, 2006, pp. 28-39.
doi:10.1109/MCI.2006.329691