Professional Documents
Culture Documents
Dynamic High-Performance Decision-Making in Air Defense System Operation
Dynamic High-Performance Decision-Making in Air Defense System Operation
Abstract—Decision-making in critical situations which are One of the most critical decisions that must be done
time-constraint and high-impact requires comprehensive carefully is a decision in critical situations. Such situation is
information and a careful process. Such situations are always a decision that should be taken by a commander of an air
faced by an air defense commander who has the responsbility defense system. Consider an air defense system of a nation,
to secure the national airspace from any intruder which tries to
that is, a complex system which commonly consists of air,
enter the airspace anytime from anywhere. In a critical
situation where an unknown or unregistered object enters the land, and sea sensors and weapon systems as depicted in
airspace, the air defense commander has to make the right Fig.1 as an example. An air defense system is under
decision based on the data and the information from its command of an air defense commander and it commonly
subordinates at the soonest time. An air defense system is a consists of several air defense sector command which watch
combination of sensors, weapons, command and control (C2), each sector command’s airspace.
intelligence systems, communications, and personnel that forms
Command, Control. Communications, Computers,
Intelligence, Surveillance, and Reconnaissance (C4ISR)
operating under a designated air defense commander. Because
of its criticality in time and precision in decision, in this study
we proposed a solution to help the process of decision-making
in the air defense system by modeling the dynamics of decision-
making by using social network approach based on modified
Friedkin-Johnsen model. We found that this model is helpful
for an air defense commander in making a decision accordingly
to the information received from his subordinates, and shortens
the data processing in the decision-making process. Some air
defense decision-making scenarios with different sensory
apparatus configurations are also presented to show how it
works in a dynamic situation and its high- performance in
delivering a better decision.
Fig. 1. A simple example of air defense system.
Keywords—Air Defense System, C4ISR, decision-making,
opinion formation, social network Air defense system command is a combined defense
I. INTRODUCTION forces from three services that is, the army, the navy, and the
airforce. Commonly, the air defense forces are equipped with
Decision making has been an important thing since early warning surveillance or a tracking radar with an
humans were born and also an inseparable part of their daily advanced Command and Control (C2) system, fighter jets as
life. Decision-making is done in any situation, at anytime, the interceptor, warships as the sensors and weapon systems,
anywhere, and at any level whether in strategic, tactical, or guided missiles, and anti-air artillerry as well as groun-to-air
operational, from the simple things to the complex ones. missiles. This system can handle the threats robustly and
Decision-making and also its process cannot be done efficiently set the right weapons to weaken the target. Air
arbitrarily bearing in mind that the results of a decision will defense system operates 24/7. As its task and function, air
affect subsequent actions and impacts which should be defense system command protects the national airspace from
considered in advance. Decision-making must be right and any air violation and tresspassing of unknown or
precise on target, that is why the information used to build unregistered flying objects whether by mistake or by
the decision must be valid and correct. Decision-making in a intention. These objects can be any flying object such as
time-constraint and high-impact situation requires military fighter jet, military transport aircraft, commercial
comprehensive information and a careful process. airline, unmanned vehicle, helicopter, or other forms of
flying object which are indicated endangering the
40
Authorized licensed use limited to: University of Canberra. Downloaded on June 09,2020 at 03:23:43 UTC from IEEE Xplore. Restrictions apply.
Fig. 3. Graph representation [8].
of several scenarios of the decision-making process and
discussion. Finally, conclusions are stated in Section V. Definition 3 (Stochasticity and substochasticity). The
II. PRELIMINARIES nonnegative of matrix W is said as stochastic if the sum of
all rows is 1 and is said as substochastic if the sum of the row
A. Opinion, Agent, and Matrix Form is less than 1 [8].
The social network is a structure consisting of several
agents that interact with each other described by dyadic ties. B. The Friedkin-Johnsen Model (FJ Model)
The agent is the smallest part of this structure. The agent can The first model of opinion formation is presented by
be a person, community, institution, software agent, French-De Groot and it is called as Frenceh-DeGroot model.
hardware agent, system, or sensory apparatus depends on the Consider there are n agents in society. Their initial opinion
context of the topics [4]. The dyadic ties describe the
are stated as x1, x2 ,..., xn . Assume that the matrix of the
interaction or communication between the agent. Agent and
the dyadic ties cannot be separated as it is a result of social interaction between agent i and j , wij , so at some interaction
agent interaction. Opinion can be described as a cognitive k, the French-De Groot model is described as follows:
orientation towards some objects such as issues, events, or
individuals) [7]. The opinion of an agent or individual will
xi (k +1) = w xj (k)
ij
evolve throughout time and interaction. Many studies had (1)
proposed the model of opinion [2], [7], and [7-8]]. The
models quantify the opinion in a certain range e.g., [0,1] or
[-1,1]. The model describes the dynamics of the agent’s Notice that wij describes how the agent j’s influence
opinion. Models of social networks are divided into two toward agent i. High value of wij describes that agent j has
major groups, namely macroscopic and microscopic models.
The macroscopic model focuses on subgroup among the high influence to the agent i, and vice versa. The high value
social network, while the microscopic one focuses to of wii means that agent i is highly believe his own opinion.
investigate each agent’s opinion [4].
The Friedkin-Johnsen model [11] gives a little
The dynamic of the opinion is affected by other agents’ modification to the French-DeGroot model. They introduced
opinions and social events in the society. Opinions from another variable, that is u as a prejudiced opinion. They also
other agents will be weighted to describe the influence of the defined a diagonal matrix Λ = diag(λ1, λ2 ,..., λn ) where
agents. The weight of the agents may differ from one agent
to others. The bonding in society is described by a graph. λ1 ∈[0,1] is described as a susceptibility of agent i to social
process in society. The Friendkin-Johnsen model is shown as
Definion 1 (Graph). Graph is a pair described as
follows:
G = (V , E) , where V = {V1,...,V2} , n describes the
number of the agent. The elements vi are called as the nodes xi (k +1) = ΛWxj (k ) + (1− Λ)u (2)
The influence or connection among the agents is defined Definition 4 (Social Input). Social input is a publication
by a weight of some real positive value. The graph can also of input so that it will give impact in a social network such
be represented as a matrix. The graph is depicted in Fig. 3. apologize, clarification, investigatiom, help, etc in order to
Definition 2 (Matrix). Consider a graph G = (V , E) , a control the public opionion in society.
non-negative matrix W = (wij )i, j∈V is applied to G . It also In order to describe the social input that happens in
society, a additional variable is added to the ordinary FJ
can be said that W is a matrix for G if (i, j) ∈ E when model. The modified FJ models is stated in (3).
i 1 j 2 3
weighting factor of agent i to agent j. The sum of of
weighting factor of an agent to another agent is 1. where Λ3 = 1− Λ1 − Λ2 and Λl , l = 1, 2, 3 are nonnegative
41
Authorized licensed use limited to: University of Canberra. Downloaded on June 09,2020 at 03:23:43 UTC from IEEE Xplore. Restrictions apply.
III. AIR DEFENSE SYSTEM OPERATION
The national defense of a nation or state consists of
ground force (army), naval force, and airforce componemts.
To organize all components, a state defense system is
created. In such system, decisions regarding the efforts on
maintaining and sustaining the state defense are made. Every
information from all subordinates in every force will affect
42
Authorized licensed use limited to: University of Canberra. Downloaded on June 09,2020 at 03:23:43 UTC from IEEE Xplore. Restrictions apply.
the decision that the commander will make. So it is necessary only to identify what the object is whether it is a jet fighter
to make a model and algorithm to help the commander in or a groups of fighter jets flying in a formation, a helicopter,
making the decision in such complex system. One a private aircraft, tactical ballistic missiles, commercial
components in a state defense system is air defense command airplanes, or military air transporters, etc. Such agent is
which is a combined forces formed from three services. Air chosen to anticipate if target X is hostile and attacks the
defense command runs an air defense system and is agent. If this happens, we only loose a weapon system but
responsible for its 24/7 operation in defending the nation or not the personnel which is invaluable. If the agent confirms
state airspace from any air violation or tresspassing of that target X is indicated hostile and shows hostility, the
endangering flying objects or target X. agent then sends the information to the C2 center. Having
this information, the C2 center then dispatches the jet fighters
As stated in Section I, reference [3] also developed the according to the information delivered by the two agents, the
decision-making loop model called the SIDA systems that is UAV and the radar. In the opposite situation, the C2 center
Sense-Inference-Decide-Act cycle in a framework called will not dispatch any agent and only carries out continuous
C4ISR. In the first stage of C4ISR, the sensory apparatus monitoring the situation of the national airspace.
such as drones, radars, satellites, UAVs, etc, gather all the
information related to state defense [4]. The next state in IV. CASE STUDY
C4ISR is the interference and decision formulation. At this The mechanism of the air defense system in dealing with
step, the model of opinion formation is utilized. The situations occured in national airspace from the early
information from the sensory apparatus is collected and warning until the engagement of agents to target X in
applied to the model to obtain the final consensus as a basis airspace has been explained in the previous section. In this
for the commander in making the decision. After the section, the simulation of the opinion model for air defense
consensus is achieved, the commander or national leader has system operation decision making is presented. The
to decide what to do. The last stage of the C4ISR is act, which simulation of opinion model will be presented with three
is the stage when the commander executes the decision such different scenarios.
as sending the troops. This action will affect the field
situation in positive or in a negative way. From the effects of A. The Hierarcy of An Air Defense System Network
the action applied to the field, C4ISR starts to sense again the A simple hierarchy of an air defense system used in this
situation, and this cycle repeats. study is shown di Fig. 4. There are nine agents where six
agents namely, radar, UAV, and fighter jet act as sensory
In this study, the concept of SIDA C4ISR is adopted. The
apparatus, while the other three ones act as information
case that is used is the air defense of an area. The air defense
inferencing fusion agents in this scheme. Jet fighter agent
system consists of several layers of protection with different
always flies in formation of three aircraft. It is a standard
agents. Commonly, an air defense system is divided into
procedure in air defense system in case of there might be an
three layers. The first layer or the most outer layer is called
air engagement with the detected object. The agents are
as area air defense which is protected by fighter jets.The
teamed up as three teams. These agents will gather the
second layer is called as terminal air defense which is
information related to the national airspace situation, and
protected by guided missile, and the last one is point air
deliver the comprehensive information to the air defense
defense which is protected by anti-air artillery. The use of
commander as the basis for making decision.
UAV as an unknown object observer is not common, but
modern combat UAVs can fly like fighter jets and is
equipped with sophisticated weapons. So it is valuable to
involve this kind of weapon system into our scenarios.
Consider a case when air defense system detects target X
entering the national airspace in a certain area. The
information from the radar team is then sent to the C2 center.
If the detected object is unknown, then C2 center will Fig. 4. An example of hierarcy of an air defense system network.
dispatch jet fighters or UAVs to carry out visual observations
to ensure the identity of the detected object. Commonly, in The first team consists of three early warning radars
the time where UAV is not a part of weapon systems and not which also function as ground control intercepting radars,
as weaponized and high-performance as today, C2 center namely Radar 1, Radar 2, and Radar 3. The first team
always dispatches jet fighters considering their high speed so monitors the national airspace dynamics and collects
they can reach at the target X’s coordinate in a short time information about all flights which pass through it. They also
followed with intercepting the object. Another one important monitor any indication of flight which try to violate or
is the jet fighters can directly do engagement with the target violates or passes through the national airspace without
X if it shows hostility. having permission from our authority. This team tracks every
object that enters the airspace area. The information from the
Dispatching a team of UAVs as the forward observer to first team will be the basis for the air defense commander to
identify target X is a challenge but it is worth to try in our decide what the next step should be taken in the case the
scenarios. Using UAVs in an air defense system are aimed to detected object is not identified and may be hostile.
reduce the risk of personnel loss because of air engagement Information detected by radar form a detected object consists
with hostile target X. Back to our scenario, the C2 center of elevation or altitude, speed or velocity, and range, which
dispatches a UAV agent to the coordinate sent by the early from these data the first team commanders
warning radar of the target X’s location. Its task is
43
Authorized licensed use limited to: University of Canberra. Downloaded on June 09,2020 at 03:23:43 UTC from IEEE Xplore. Restrictions apply.
may be able to identify the type of target X. To ensure that dangerous threat. From this consensus, the air defense
his identification is correct, he sends a report to the air commander could decide to carry out necessary
defense commander. measurement to object as previously explained in Section IV
A.
Based on the first team commanders’ reports, the air
defense commander will give instruction to the second team The second scenario of the opinion formation model
commander to dispatch his agents, namely the UAV to do simulation is a scenario with the hierarchy as depicted in Fig.
direct visual observation. At this stage, the UAVs are sent to 4. The simulation results from the second scenario are
identify the type of target X whether it is a threat or not. explained in Fig. 6. The initial condition of the nine agents
When the second team confirms that it is a threat and shows was 0 as there was no information about target X
hostility more over it is weaponized, the air defense approaching the national airspace. In minute 5, Radar 2
commander will instruct the third team commander to detected this target approaching the national airspace. Radar
dispatch his agents to not only reconfirm the type of the 2 gave social input to society and affected other agents. As a
threats but also to be prepared in air engagement which may response, the remaining radar within the range increased as
occur. When target X is reconfirmed as a threat and hostile, well as the UAVs opinion. Same as stated before that the
the third team’s agents will intercept the object and after opinion of the fighter did not affect as the radar is less trusted
reporting the situation to air defense commander, the agent than the fighter. To follow up the social input from radar 2,
will do necesssary measurement such as force down, or do the commander sends the UAV to identify target X. In min-
air engagement to drive it out from the national airspace. 20, the UAV sent to target X’s location and confirmed that
target X was a threat.
B. Simulating Air Defense System Operation using Opinion
Formation Model
The model of opinion formation in the air defense system
operation as explained previously is simulated using three
different scenarios. Suppose there is target X approaching the
border of national airspace. The first scenario is a scenario
with the hierarchy as depicted in Fig. 4 but without the second
team. So the agents involved in the simulation are four agents
namely, Radar 1, Radar 2, Radar 3, and Jet Fighters. The
simulation’s result of the first scenario is shown in Fig. 5.
44
Authorized licensed use limited to: University of Canberra. Downloaded on June 09,2020 at 03:23:43 UTC from IEEE Xplore. Restrictions apply.
From the simulation, scenario 1 and 2 resulted in a
consensus so it can be used as a basis for the commander to
make the right decision. Scenario 3 did not result in a
consensus as there was two ‘stubborn’ agents giving their
social inputs at the same time. Scenario 3 was not
recommended for a commander in the process of decision-
making. However, the commander still had to make the
decision wisely and do not depend only from the results of
the algorithm. Experienced, wisdom, and humanity of the air
defense commander will never be replaced by any algorithm
in the process of decision-making. For the next future work,
the decision making for air defense operation by the
commander can be enhanced by using Artificial Intelligence
Fig. 7. The simulation results of Scenario 3.
(AI) based method.
REFERENCES
As shown in Fig. 6 in min-30 the UAV and the fighters
confirmed that target X approaching the national airspace
was a threat. But it can be seen that the UAV and the fighters [1] Sumari, A.D.W. and Ahmad, A.S. “The Application of Cognitive
Artificial Intelligence within C4ISR Framework for National
gave different social input. The UAV remained at Resilience”. In 4th Asian Conference on Defence Technology (ACDT)
0.75 and the fighter remained at 0.9. The opinion of the 2017.
fighters were more trusted as the fighters were manned by [2] Arifin, M. S., et al, “Opinion Formation for Decision Making Process
human. A human could identify exactly the type of threat. in Command, Control, Communications, Computers, Intelligence,
The UAV was less trusted than the fighters because even Surveillance, and Reconnaissance” In 5th Asian Conference on
Defence Technology (ACDT) 2018.
though the UAV could approach closer to the object, UAV
[3] Naseem, A., et al, “Decision support system for optimum decision
only controlled by the command center from a distance and making process in threat evaluation and weapon assignment: current
the sensor of the UAV might be false. status, challenges and future directions”, Annual Reviews in Control,
2017, p. 1-19.
The social input from the UAV and the fighter did not
[4] Karako, T. and Rumbaugh, W., “Distributed Defense”, CSIS Missile
make any consensus despite the consensus of the agent from Defense Project, 2018
team 1. The radars made a consensus at around 0.85. This [5] French, J., Jr., A formal theory of social power. Physchological
scenario did not help the commander to make a decision on Review, 63, 181–194, 1956.
how to treat the threat because there were 3 subgroups [6] Proskurnikoc, A.V. and Tempo, R., “A tutorial on modeling and
opinion about target X. These subgroup opinion occurred analysis of dynamic social networks: Part I”, Annual Reviews in
because the UAVs and the fighter had their own highly trust Control43, 65–79, 2017.
so they act as a ‘stubborn’ agent. [7] Taylor, M., Towards a mathematical theory of influence and attitude
change.Human Relations, 21(2), 121–139, 1968.
V. CONCLUSION [8] Krause, U., A discrete nonlinear and non-autonomous model of
consensusformation. Communication in Difference Equations, 227–
This study has proposed a dynamic of high-performance 236, 2000.
decision making for the air defense system operation in [9] Constrantine, R.W., “Air Defense Systems Department: An
protecting national airspace. It has been shown that opinion Overview”, Johns Hopkins APL Technical Digest, Volume 22,
of agents in the air defense system operation can be modeled Number 3, 2001.
using modified FJ model. The model is applied in a simple [10] Arifin, M.S., et al, “Switching Model of a Dynamic Social Network”,
Internetworking Indonesia Journal, Vol.1, No.1, 2018.
case of air defense system operation. The dynamic of the
high-performance decision-making process was presented in
different scenarios.
45
Authorized licensed use limited to: University of Canberra. Downloaded on June 09,2020 at 03:23:43 UTC from IEEE Xplore. Restrictions apply.