Real-Time Embedded Controller.. (Idris E. Putro and Harry Septanto)

REAL-TIME SIMULATION OF EMBEDDED CONTROLLER FOR
MISSILE
(SIMULASI WAKTU-NYATA PENGENDALI TERTANAM UNTUK MISIL)
Idris E. Putro1, Harry Septanto2
1 Pusat Teknologi Roket - LAPAN
2Pusat Teknologi Satelit - LAPAN
1e-mail: idris.ep@lapan.go.id

Diterima : 15 Juli 2019; Direvisi : 07 September 2019; Disetujui : 16 September 2019

ABSTRACT
This paper focuses on the development of real-time environment for embedded system to control
flight performance of missile. Real-time simulation is generated by using xPC target so that standalone
target boot can be developed, and target PCs can be simulated in real-time. Two PCs and single board
computers (SBCs) are required to configure the host and standalone target respectively. The host PCs
are used to control the target PCs while these target PCs run simultaneously in real-time. Target PCs
are divided into plant platform and controller platform. Plant platform represents missile dynamics
model, while control algorithm is compiled to the other system as controller platform. Linear control
synthesis will be implemented for maintaining flight stability of missile using optimal control approach
based linear quadratic regulator (LQR). Simulation results show the deviation time between real-time
system and non real-time under Matlab/Simulink only 0.001 second. This real-time system should
resemble the environment of future flight test.
Keywords: real-time simulation, xPC Target, single board computer (SBC), standalone target, LQR

ABSTRAK
Paper ini fokus pada pengembangan lingkungan real-time untuk sistem embedded berbiaya
murah yang digunakan untuk mengendalikan prestasi terbang dari missile. Simulasi real-time
dihasilkan dari metode xPC target sehingga penyalaan standalone target dapat dikembangkan, dan PC
sasaran dapat disimulasikan secara real-time. Dua PC dan dua SBC masing-masing diperlukan untuk
membuat host dan target mandiri. Host PC digunakan untuk mengendalikan PC sasaran dimana PC
sasaran ini bekerja serentak secara real-time. PC sasaran dapat dibagi menjadi Platform model wahana
dan platform kendalian. Platform model wahana mewakili model dinamika terbang missile, dan platform
yang lainnya merupakan hasil proses kompilasi dari algoritma kendali (kendalian). Implementasi dari
sintesa kendali linear akan digunakan untuk menjaga kestabilan terbang missile dengan menggunakan
pendekatan kendali optimal berbasis linear quadratic regulator (LQR). Hasil simulasi menunjukkan
bahwa perbedaan waktu antara sistem real-time dengan non real-time

oleh perangkat lunak

Matlab/Simulink hanya sebesar 0.001 detik. Sistem real-time ini diharapkan dapat mencerminkan
lingkungan nyata dari uji terbang di masa depan.
Kata kunci: simulasi real-time, xPC target, single board computer (SBC), standalone target, LQR

129

Jurnal Teknologi Dirgantara Vol.17 No.2 Desember 2019 : hal 129-140

1

INTRODUCTION
Development of unmanned aerial
vehicles have rapidly increased recently.
Wide range applications from civil and
safety purposes until military and
defense missions lead this vehicle needs
advanced control system. One of the most
considered vehicles to support defense
and military tasks is missile. The design
and implementation of autonomous
control system for missile is rather
challenging due to higher risk in flight
testing and increasing complexity in flight
dynamics for high speed mode. Hence
design and development of an accurate
control in real-time simulation for missile
become
important
to
do
before
implementing it in real flight test.
Designing and testing of control
system needs to be verified prior to the
flight test. Direct flight test without
ground simulation will risk either the
vehicles or environment due to an
expensive onboard system, a lot of time
wasted and more efforts to understand
the missile behavior. Flight and control
simulation are suitable to design and test
the control system associated with
missile dynamics. Using simulation,
every flight condition can be considered,
and then appropriate control algorithm
designed, analyzed, and adjusted so that
the precision controller can be developed
more accurate. After the controller is
ensured preconcerted accurate, then it
can be implemented as an embedded
system for the flight test.
The embedded control system
must be applied in real situation as
similar as the flight test. The responses of
the embedded controller should be able to
suffice deterministic time and meet the
real-time environment. There have been a
lot of method to develop real-time
simulation for testing flight controller.
The fanciest in the last decade is using
xPC target to construct real-time
hardware-in-the-loop simulation (HILS).

130

For instance, Valavanis et al reveal in his
book that Ates utilized xPC target to
simulate various air vehicle type in realtime such as jet Fighters F-16 and F-4.
(Valvalanis, 2008). Peng Lu (Lu, 2011)
has constructed hardware-in-the-loop
(HIL) to demonstrate real-time simulation
UAV system based on xPC target under
Matlab/Simuling. He has successfully
validated that the responses of real-time
simulation accurate enough compared to
real flight. Extensive work has also been
performed for autonomous flight of rotary
wing UAV (Budiyono et al, 2010). Realtime HIL through xPC target is used to
simulate various flight condition such as
hover, cruise, climb, descent, and turning
maneuver flights. Further improvement
uses combined programming between
xPC
target
and
LabVIEW/SIT
environment
to
build
real-time
simulation system of the orbital transfer
vehicle (OTV) model (Ma et al, 2014).
Optimal control has been widely
used in aerospace application. Devan
extend the use of lqr control to design roll
autopilot of missile by cascading the
algorithm
(Devan,
2018).
Further
extensive research work has been
conducted (Zheng, 2018) by using lqr to
perform
tracking
guidance
law
simulation for target missile.
This paper presents real-time
simulation based on xPC target for
embedded controller to simulate flight
stability of missile. The missile model
either nonlinear or linear will be run
simultaneously
and
the
controller
gathers flight data and stabilize the
dynamic of missile. All the real-time
simulation based on xPC target are
carried out under Matlab/Simulink
software by using Real Time Workshop
(RTW). Optimal control based linear
quadratic regulator is proposed for
controller target PC to keep stability of
missile
flight
in
a
presence
of
disturbances.

Real-Time Embedded Controller.. (Idris E. Putro and Harry Septanto)

2

METHODOLOGY
Real-time simulation developed in
this research uses methodology as the
following flow chart.

Picture 2-2: Missile configuration design

Missile dynamic constructed from eulernewton formulation.
𝑑𝑉
=𝐹
𝑑𝑡 %

(2-1)

𝑑𝜔
=𝑀
𝑑𝑡 %

(2-2)

𝑚
𝐼

Picture

2-1:

Flow

chart

of

real-time

simulation uses xPC target
method

Real-time system is built for target
PC which will be used as embedded flight
controller. Embedded flight controller
must run in real-time environment thus
there is no significant delayed time
between inputs and responses in order to
control the flight stability of missile. A
real-time Operation System (OS) in this
research is created based on xPC Target
Kernel, since xPC gives the ability to
convert
a
designed
model
in
Matlab/Simulink to a real-time OS
(Mohammed, 2013)
Details explanation regarding host
PC and target PC associated with missile
dynamic, control method, and xPC target
will be revealed in the next sub-chapter
respectively.
2.1

Missile Dynamic Model
Missile model in this research
uses configuration of double stages
rocket developed by LAPAN.

Equation above derived from the law of
conservation of linear and angular
momentum, with 𝐹 = 𝑋, 𝑌, 𝑍 / , and 𝑀 =
𝐿, 𝑀, 𝑁 / , are the vector of total forces and
moments acting on the missile center of
gravity. Total forces and moments are
generated by control surfaces, i.e.
elevator, aileron, rudder, and throttle.
Gravitational force in three axes are also
considered.
Further derivation of forces and
moments can be seen in (Putro et al,
2012) yields missile dynamic presented
as 6 degree of freedom. Missile dynamic
in translational motions of 3 axis 𝑥, 𝑦, and
𝑧 are respectively represented as equation
below.
𝑚𝑢 = 𝐹6 − 𝑞𝑤 + 𝑟𝑣 − 𝑔 sin 𝜃
𝑚𝑣 = 𝐹B − 𝑟𝑢 + 𝑝𝑤
+ 𝑔 cos 𝜃 sin 𝜙
𝑚𝑤 = 𝐹G − 𝑝𝑣 + 𝑞𝑢
+ 𝑔 cos 𝜃 cos 𝜙

(2-3)
(2-4)
(2-5)

Rotational motions of missile dynamics in
𝑥, 𝑦, and 𝑧 axis can be formulated as
below.
𝐼6 𝑝 = 𝐿 − 𝑞𝑟 𝐼G − 𝐼B + 𝐼6G 𝑝𝑞 + 𝐼6G 𝑟

(2-6)

131

Jurnal Teknologi Dirgantara Vol.17 No.2 Desember 2019 : hal 129-140

𝐼B 𝑞 = 𝑀 − 𝑞𝑟 𝐼6 − 𝐼G + 𝐼6G 𝑝 H − 𝑟 H 𝐼B (2-7)
𝐼G 𝑟 = 𝑁 − 𝑞𝑟𝐼6G − 𝐼B − 𝐼6 𝑝𝑞 + 𝐼6G 𝑟

(2-8)
Completing equation (2-3) to (2-8) to
govern 6 degree of freedom, kinematic
equations can be arranged as the
following equations for each 𝑥, 𝑦, and 𝑧
axis.
𝜙 = 𝑝 + 𝑞 sin 𝜙 tan 𝜃 + 𝑟 cos 𝜙

(2-9)

𝜃 = 𝑞 cos 𝜙 − 𝑟 sin 𝜙

(2-10)

𝜓 = 𝑞 sin 𝜙 + 𝑟 cos 𝜙 sec 𝜃

(2-11)

2.2

Optimal Control Design
Optimal Control is used to
improve
the
attitude
stabilization.
Optimal control approach in this
research uses linear quadratic regulator
(LQR). LQR controller has been known as
ubiquitous controller which provides
effectively fast response with minimum
settling time (Devan, 2018). This paper
proposes lqr approach to keep the
angular angle of missile always in stable
performance even small disturbances
occur. This can be done by setting the
angular rate errors tend to zero when
disturbance applied thus the angle still
can be kept after steady state.
The LQR method try to find the
control inputs that can minimize the
chosen output by minimized 𝑄 and 𝑅
weighting matrices (Kumar, 2016). This
approach must find state-feedback law
𝑢 = −𝑘𝑥 as it may minimizes the
quadratic cost function as the following
equation in time domain.
R

𝐽 𝑢 =
S

𝑥 / 𝑄𝑥 + 𝑢 / 𝑅𝑢 + 2𝑥 / 𝑁𝑢 𝑑𝑡 (2-12)

The optimal gain matrix then can be
found by solving the matrix differential of
Riccati equation.

Eventually the optimal control gain
matrix can be obtained from the following
equation.
(2𝐾Y ≜ −𝑅 WX 𝐵 / 𝐾
14)
2.3

Host PC
The host PCs are developed using
Matlab/Simulink
software
under
Windows operating system. Under
Matlab/Simulink, the host PC will run
missile dynamic model and controller
model and generate compiler files for
target PCs to be controlled during
simulation.
This host PCs will be connected to
target PCs through 100 Mbps TCP/IP
connection for downloading executable
files into target computer using xPC
target in Matlab. This high-speed
connection allows xPC target explorer in
the host computer become console for
monitoring and controlling the target
computer during the process of the real
time simulation. The host PCs also will
be used to display and analyze all of the
flight simulation data, covering missile
dynamic model and controller model.
Asus Strix ROG series notebook is
used as a host computer to run the
missile model, meanwhile desktop
computer is used to run controller model.
Communication between host PCs and
target PCs are connected using UDP port.
This high-speed connection will establish
xPC target spy in host PC while the
simulation is running. xPC target spy is
target screen spy in host PC while the
target PC was running. This command is
commonly used to display the target PC
simulation in host PC thus we can
monitor the simulation of target PC on
the host PC and directly controlled it from
host PC.
2.4

𝐾 = 𝐾𝐴 + 𝐾𝐵𝑅

132

WX

/

/

𝐵 𝐾−𝑄−𝐴 𝐾

(2-13)

Target PC
The target PCs are the computer
which are running the executable files

Real-Time Embedded Controller.. (Idris E. Putro and Harry Septanto)

representing missile dynamic model and
controller model and will be run
simultaneously as a real-time kernel
application
on
free-DOS
operating
system. Target PCs can be controlled
through host PCs under DOS loader
target boot mode and also can be run
automatically under standalone target
boot mode.
In this research, both of those
configurations have been done. The DOS
loader mode is used to tune the missile
model and controller parameters, and
then those optimized models are used to
generate compiler files for running under
standalone mode. The last configuration
of controller model on target computer
resemble the embedded controller system
that can be used for real flight test.
The target PCs were using a 3.5”
single board computer (SBC) powered by
Intel Celeron 600MHz and 512MB DDR
memory (SBC: WAFER-8522). The main
specifications of target PC can be seen as
the following table.
Tabel 2-2: SBC WAFER-8522 SPECIFICATION

CPU
Memory

I/O

Dimension
Weight

On-board Intel®
Celeron®M 600MHz 512K
cache
DDR226 512 MHz
5x RS-232
Dual Realtek RTL8110SC
Ethernet
4x USB
1x IDE
1x PS/2 for KB/MS
1x TPM v.1.2 module onboard
146 mm x 102 mm
GW: 700 g and NW: 175 g

The SBC performance allows the
process running in high-frequency as in
desktop PC under a small compact and
mobile package that possible to be used
as flight control computer (FCC) in the

future. The integrated VGA on SBC is
used to display the running behavior of
each model in an LCD monitor. This
feature serves observation, therefore
simulation process in standalone mode
can be analyzed. Furthermore the host
computers can be eliminated for
controller model to simplify the system
after all of configuration has been fixed
previously by using DOS loader mode.

Picture 2-3: Single board computer WAFER8522

SBC missile dynamic model is
connected to SBC controller model by two
channels of RS-232 at 115,200 bps, 8
data bits and 1 stop bits. Output data
from missile dynamic model is sent
through RS-232 com-1 port and will be
received as input data by SBC controller
model
at
com-1
port.
Reversely,
command data from SBC controller
model is sent through RS-232 com-2 port
and received as input data at com-2 port
by SBC missile model.
Since the SBC has six RS-232 ports,
using these two channels separately can
give
advantages
to
avoid
serial
communication problems such as CPU
overload that might be occurred when
using only one serial port alternately.
3

RESULTS AND DISCUSSION
Real
time
simulation
is
established for both missile dynamic
model and controller model of the target
PCs. These model are obtained from xPC
target compiling process.

133

Jurnal Teknologi Dirgantara Vol.17 No.2 Desember 2019 : hal 129-140

Plant
target
pc
which
is
represented missile dynamic model
generated from equation (2-3) - (2-11) in
matlab/simulink software. By inserting
missile configuration data in table 2-1
into equations, thus missile dynamic can
be
formulated
and
simulated
in
Matlab/Simulink
Plant target pc is connected to the
controller target pc through serial cable
connection on RS232 port. The simulink
block diagram of those target PCs are
shown in Figure 3-1.
Plant target and Controller target
PCs above are compiled using C compiler
on
Real
Time
Workshop
under
Matlab/Simulink
Environment
and
generate executable files to be run on
single board computer. These executable
files running on SBC-target computer are
called target PCs.
Simulation is conducted by
running missile dynamic model equipped
with optimal control as defined in chapter
2.2. Optimal gain controller is obtained
for Lapan double stage rocket in equation
(2-14) 𝐾Y as the following below:

2.28 −10.32 10.24 −2.95
0.77 −0.12 −1.54 −7.98
Missile configuration (Mugia, 2012) can
be seen from table 2-1.
𝐾Y =

Tabel2-1: MISSILE CONFIGURATION.

Parameters

Values and
Units

Diameter(d)

200 mm

Total length(l)

2353 mm

Total mass (m)

65.26 kg

CoG position(cg)

1050 mm

Thrust(T)
- Booster
- Sustainer

500 kgf
80 kgf

Wing area (S)

0.0488 m2

Wing chord (c)

0.325 m

Inertial Moments
(I)
- 𝐼66
- 𝐼BB

-

𝐼GG

Picture3-1: Simulink blocks of missile dynamic and controller models

134

0.012 kgm2
84.43 kgm2
84.43 kgm2

Real-Time Embedded Controller.. (Idris E. Putro and Harry Septanto)

Roll Angle response
0.35

0.35

Simulink

Realtime

0.3

0.3

0.25

0.25

0.2

0.15

Roll f (rad)

Roll f (rad)

0.2

0.1

0.15

0.1

0.05
0.05
0
0
-0.05
-0.05
-0.1

0

2

4
time (sec)

6

8

0

2

4
time (sec)

6

8

Picture3-2: Comparison of simulink and target pc simulation: Roll angle response

Roll rate response
1
Simulink

Real time

0.8

0.8
0.6
0.6

Roll rate (rad/s)

Roll rate(rad/s)

0.4
0.4

0.2

0

0

-0.2

-0.2

-0.4

0.2

0

2

4
time (sec)

6

8

-0.4

0

2

4
time (sec)

6

8

Picture3-3: Comparison of simulink and target pc simulation: Roll rate response

135

Jurnal Teknologi Dirgantara Vol.17 No.2 Desember 2019 : hal 129-140

Simulation result under real time
environment can be seen from figure 3-2
until 3-7. Figure 3-2 and 3-3 show the
attitude responses on lateral mode. There
is a very small different response time
between simulink result and realtime
result. The difference value only within
0.001 sec since the plant model was very
simple.
From real-time simulation also
can be seen that the controller is able to
work properly. The controller can

Pitch Angle response
0

0

-0.2

-0.2

-0.4

-0.4

-0.6

-0.6

Pitch q (rad)

Pitch q (rad)

maintain the roll rate close to zero
𝑝 ≈ 0 𝑟𝑎𝑑/𝑠 and keep the roll angle to
φ = −0.074 𝑟𝑎𝑑 for cruise flight.
Figure 3-4 to 3-7 show the same
behavior with lateral mode. The controller
in real time situation can keep pitch and
yaw angle near the operating points of
cruise flight, θ = −0.0767 𝑟𝑎𝑑 and ψ =
0.3077 𝑟𝑎𝑑
Meanwhile the model also can
maintain the rate attitude tend to zero,
q = 0 rad/s and r = 0 rad/s.

-0.8

-0.8

-1

-1

-1.2

-1.2

Real time

Simulink
-1.4

0

2

4
time (sec)

6

8

-1.4

0

2

4
time (sec)

6

Picture 3-4: Comparison of simulink and target pc simulation: Pitch angle response

136

8

Real-Time Embedded Controller.. (Idris E. Putro and Harry Septanto)

Pitch rate response
Real time

0.9

0.9

0.8

0.8

0.7

0.7

0.6

0.6
Pitch rate (rad/s)

Pitch rate(rad/s)

Simulink

0.5
0.4

0.5
0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

0

-0.1

0

2

4
time (sec)

6

-0.1

8

0

2

4
time (sec)

6

8

Picture 3-5: Comparison of simulink and target pc simulation: Pitch rate response

Yaw Angle response
0.4

0.4

Real time

0.35

0.35

0.3

0.3

0.25

0.25

Yaw y (rad)

Yaw y (rad)

Simulink

0.2

0.2

0.15

0.15

0.1

0.1

0.05

0

2

4
time (sec)

6

8

0.05

0

2

4
time (sec)

6

8

Picture 3-6: Comparison of simulink and target pc simulation: Yaw angle response

137

Jurnal Teknologi Dirgantara Vol.17 No.2 Desember 2019 : hal 129-140

Yaw rate response
0.2

0.2

Real time

0.15

0.15

0.1

0.1

Yaw rate (rad/s)

Yaw rate(rad/s)

Simulink

0.05

0

-0.05

0.05

0

0

2

4
time (sec)

6

8

-0.05

0

2

4
time (sec)

6

8

Picture 3-7: Comparison of simulink and target pc simulation: Yaw rate response

4

CONCLUSION
Real-time Simulation of embedded
control has been conducted. Missile
dynamic model and controller model can
simulate simultaneously in real-time
based on xPC target. It was demonstrated
that Plant target PC and Controller target
PC
were successfully run under
standalone mode and maintain its
operating points for stabilization flight.
The small deviation time 0.001 second
between real-time and non real-time
simulation is reasonable since the
controller has only handled the data from
missile dynamic model in attitude
performances.
Further development can be
considered by implementing this optimal
control based linear quadratic regulator
(LQR) for way-points tracking and
navigation control in a real-time
hardware prior to flight test.

138

ACKNOWLEDGMENT
This paper and the research
behind it would not have been possible
without the exceptional support of my
Professor, Kwang Joon Yoon, for the
chance working together as his research
assistant during master degree program.
I am also grateful for the anonymous peer
reviewers
at
Journal
Teknologi
Dirgantara
since
their
insightful
comments lead this manuscript greatly
improved.
Eventually
I
immensely
grateful to Pustekroket– LAPAN for
providing me a huge facility to support
this research.
REFERENCES
Valavanis, K.P., et al.(eds), 2008.
Unmanned Aircraft System. Springer
Publisher, DOI: 10.1007/978-1-40209137-7_1.
Budiyono, A., Idris E. Putro, et al., 2010.
Real-time Hardware Simulation of a

Real-Time Embedded Controller.. (Idris E. Putro and Harry Septanto)

Small-Scale
Helicopter
Dynamics,
International Journal of Aircraft
Engineering
and
Aerospace
Technology Vol. 82, Issue 6, pp. 360371.
Peng Lu, QingboGeng, 2011. Real-time
Simulation System for UAV based on
Matlab/Simulink,
IEEE
2nd
International
Conference
on
Computing, Control and Industrial
Engineering
(CCIENG),20-21
Aug
2011, ISBN 978-1-4244-9600-6.
Endang Mugia GS., 2012. Pengaruh Nilai
Koefisien Aerodinamika 𝐶q r dan 𝐶q s
pada Kestabilan Terbang Gerak
Periode Pendek (Short Period) RKX-200
LAPAN, Jurnal Teknologi Dirgantara
Vol. 10, Issue 2, December, pp. 133138.
Idris E. Putro, Herma Y. Irwanto, Ahmad
Riyadl, 2012. Control Simulation of
Fixed Wing UAV based on First
Principle Approach, 6th International
UAV World Conference (AIRTEC), 6 – 8
November 2012.
Ma, Yehan., Ling, M., et al., 2014. Design
and Implementation of Real-time
Simulation System of OTV based on
xPC Target and SIT, IEEE 11th World

Congress on Intelligent Control and
Automation (WCICA), 29 June – 4 July
2014, ISBN 978-1-4799-5825-2.
E. Vinodh Kumar, J. Jerome, 2016.
Algebraic Riccati equation based Q and
R matrices selection algorithm for
optimal LQR applied to tracking control
of 3rd order magnetic levitation system,
Archive of Electrical Engineering
Vo.65, Issue 1, pp. 151-168.
Devan, Viswanath., Chandar TS., 2018.
Cascaded LQR Control for Missile Roll
Autopilot, International Conference on
Control,
Communication,
and
Computing (IC4), 5-7 July 2018,
Trivandrum, India
Mohammed, Abdullah. A., Ali. Dia M.,
2013. Creating Real-Time Operation
System Based on xPC Target Kernel,
International Journal of Recent
Technology and Engineering (IJRTE),
Volume-2, Issue-4, September 2013,
ISSN: 2277-3878.
Zheng, X., et al., 2018. Design and
Simulation of Trajectory Tracking
Guidance Law Based on LQR for Target
Missile,
IOP
Conference
Series:
Materials Science and Engineering
435
(2018)
012021
DOI:
10.1088/1757-899X/435/1/012021

139

Jurnal Teknologi Dirgantara Vol.17 No.2 Desember 2019 : hal 129-140

140