I.

INTRODUCTION

Due to its simplicity of onboard implementation,

A Unified Approach to proportional navigation (PN) has attracted a
considerable amount of interest in the missile
Proportional Navigation guidance literature since its inception in 1940s. In its
nonlinear form, the PN schemes can be categorized
into two major classes: interceptor velocity referenced
class and the line-of-sight (LOS) referenced class.
The latter is superior to the former in the aspect of
mathematical tractability, and numerous papers have
CIANN-DONG YANG
appeared in the literature dealing with the analytical
CHI-CHING YANG
study of the LOS referenced system such as the
National Cheng Kung University
Taiwan true proportional navigation (TPN) [4, 5, 9, 11],
generalized true proportional navigation (GTPN)
[6, 13, 17], realistic true proportional navigation
(RTPN) [2, 3, 15], and ideal proportional navigation
In this paper, the two major classes of proportional navigation (IPN) [16]. Although less success has been achieved
(PN), namely, true proportional navigation (TPN) and pure in solving the nonlinear equations of interceptor
proportional navigation (PPN) are analyzed and solved by a velocity referenced system such as pure proportional
unified approach. The analytical tools used in the line-of-sight
navigation (PPN) [1, 7, 8, 10] and its variants, PPN
is believed to be more practically implementable with
(LOS) referenced systems such as TPN, realistic true proportional
no requirement on forward acceleration/deceleration
navigation (RTPN), generalized true proportional navigation
and with no constraints on the initial engagement. A
(GTPN) and ideal proportional navigation (IPN), are extended
thorough comparison between PPN and TPN can be
here to handle the interceptor velocity referenced systems such found in [12]. Augmented proportional navigation
as PPN and its variants. It is found that the above two branches (APN) [18] is a linearized version of RTPN with an
of guidance systems belong to a more general PN scheme which extra term to account for the maneuvering target.
defines the acceleration of the interceptor as being proportional Since APN makes use of extra information, namely,
to the LOS rate with direction normal to an arbitrarily assigned knowledge of the target maneuver, it is reasonable that
vector ~L. For example, ~L of TPN is LOS, and ~L of PPN is the this knowledge should enable the missile to maneuver
interceptor’s velocity. Every PN scheme associates with a specific in a more efficient manner with less requirement of
form of ~L. The optimal PN (OPN) problem which concerns the total acceleration.
determination of the optimal direction ~L is also addressed. Under
One of the challenges of investigating PN guidance
laws is how to formulate a standard framework under
the proposed general PN scheme, its six special cases, i.e., TPN,
which all PN schemes can be analyzed and solved in a
RTPN, GTPN, IPN, PPN, and OPN are solved in a unified way
unified manner. The concept of generalized guidance
from which many new relations among them can be revealed, and
law proposed in [14] was an attempt for this end,
their performances can be compared on a common basis. but was not successful enough to include PPN as its
special case. Up to now, it is still a common belief
that the relatively considerable success achieved in the
analysis of LOS-referenced guidance laws cannot be
made use of in treating the PPN problem. However,
the results of this work could somewhat change this
belief. Here, the problem of generalized guidance
law is revisited, and the possibility of analyzing
LOS-referenced guidance laws and interceptor
velocity referenced guidance laws in a unified
approach is verified.
The commanded acceleration of an interceptor
Manuscript received January 17, 1996; revised May 3 and June 28, guided by the generalized PN scheme can be
1996.
expressed by the following general form.
IEEE Log No. T-AES/33/2/03161.
~aI = ¸~L £ !
~ (1)
Authors’ address: Institute of Aeronautics and Astronautics,
National Cheng Kung University, Tainan, Taiwan, Republic of where ~L is the normal direction of the commanded
China. ~ = μ_~k is the LOS angular velocity, and
acceleration; !
¸ is the navigation constant. Different PN guidance
0018-9251/97/$10.00 °
c 1997 IEEE laws employ different forms of ~L. For example, ~L

IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 33, NO. 2 APRIL 1997 557
of TPN is along the LOS, while ~L of PPN is along each ~L can be expressed as a function of μ:
the interceptor’s velocity. For each PN scheme, we ~L = u(μ)~er + v(μ)~eμ : (3)
can assign a specific ~L to it, and this gives rise to the
optimal PN (OPN) problem that which of ~L is the The relative motion between an interceptor guided by
optimal selection. Of course, the answer depends on (1) and a nonmaneuvering target is described by the
the cost function being chosen. following nonlinear equations
Here, trajectory and performance of the interceptor
guided by the generalized PN law (1) are analyzed, r̈ ¡ rμ_2 = ¡¸μv(μ)
_ (4a)
and its six special cases, namely, TPN, RTPN, GTPN,
rμ̈ + 2r_ μ_ = ¸μu(μ):
_ (4b)
IPN, PPN, and OPN are solved analytically in a
unified framework under which the performances Without loss of generality, we assume that μ0 = 0 and
such as capture area, required time of interception, μ_0 > 0 by proper choice of reference direction. With
and cumulative velocity increment for six PN schemes change of independent variable from t to μ, (4) can be
can be compared on a common basis. Using this rewritten as
unified approach, new relations among the six PN
schemes are revealed; for example, we find that IPN Vr0 ¡ Vμ = ¡¸v(μ), Vr (0) = r_0 (5a)
is the extreme case of PPN, and that PPN resembles
Vμ0 + Vr = ¸u(μ), Vμ (0) = r0 μ_0 (5b)
the behavior of OPN, etc. This work concentrates
on the establishment of a unified framework for where primes denote the differentiation with respect
analyzing PN, and only nonmaneuvering target is to μ. To have the solution independent of the physical
considered. To account for maneuvering targets, (1) units, it is helpful to rewrite (5) in a dimensionless
can be augmented to the following form: form
0
~aI = ¸~L £ !
~ +~aT Vr ¡ Vμ = ¡¸v(μ), Vr (0) = ¡ sin Ã0 (6a)
0
where ~aT is the measured target’s acceleration. The Vμ + Vr = ¸u(μ), Vμ (0) = cos Ã0 (6b)
unified analysis given here can be applied to the case
where ~aT is a given constant, and a nonlinear analysis where the q variables with overline denote the division
of APN can thus be derived. Closed-form solution of by V0 = r_02 + (r0 μ_0 )2 ; for instance, Vr = Vr =V0 ,
nonlinear APN is possible, but the expression is rather etc. The dimensionless time ¿ is defined as ¿ =
tedious. The investigation of APN using the proposed t=(r0 =V0 ), andqthe parameter Ã0 is defined as Ã0 =
unified approach will be addressed in another paper. arc cos(r μ_ = r_ 2 + (r μ_ )2 ). For given functions of u(μ)
0 0 0 0 0
and v(μ) corresponding to a specific guidance law,
II. UNIFIED PN FORMULATION (6) can be solved to obtain Vr (μ) and Vμ (μ) which,
in turn, can be used to evaluate the capture area, the
Consider a planar relative motion described by cumulative velocity increment, and the required time
the polar coordinates (r, μ) with the origin located at of interception.
the interceptor, where r is the distance between the Once Vr (μ) and Vμ (μ) are found, the capture area
interceptor and the target, and μ is the aspect angle is determined from the conditions:
of LOS with respect to an inertial reference line. Let
Vr (μf ) · 0, Vμ (μf ) = 0 (7)
(~er ,~eμ ) be the unit vectors along the coordinate axes.
In terms of the polar coordinates, the ~L directions of and the normalized cumulative velocity increment is
the six guidance laws can be expressed, according to calculated as
their definitions, as Z ¿f ¯ ¯
¯ aI ¯
¢V = ¯ ¯
1) TPN: ~L = r_0~er (2a) ¯ V2 =r ¯ d¿
0 0 0
2) RTPN: ~L = r_~er (2b) Z μf q
~L = r_0~er + r0 μ_0~eμ =¸ u2 (μ) + v2 (μ) dμ: (8)
3) GTPN: (2c) 0

4) IPN: _e
~L = r_~er + rμ~ (2d) To derive the trajectory of the relative motion, we
μ
5) PPN: ~L = ¡V~I (2e) introduce the unit angular momentum h:

6) OPN: ~L = uopt (μ)~er + vopt (μ)~eμ (2f) h = r2 μ_ (9)

where RTPN can be conceived of the practically and the associated dimensionless form is h = h=(r0 V0 ).
implemented TPN with varying closing speed, and In terms of h, (6) is combined to yield
IPN can be considered as the practically implemented 0
GTPN with varying closing speed and with varying dh Vμ (μ) + Vr (μ)
= dμ: (10a)
tangential speed. It is shown in the next section that h Vμ (μ)

558 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 33, NO. 2 APRIL 1997
The integration gives the normalized capture area for TPN:
r
h(μ) = h0 e¡©(μ) (10b) 1
0 · h0 · 1 ¡ : (16)
where h0 = cos Ã0 . The function ©(μ) is called the 2¸
characteristic function of the guidance law, and is The name “capture area” originated from the fact that
defined as the admissible engagement conditions are described as
Z μ 0 an area in the (Vr0 , Vμ0 ) plane. However, we recognize
Vμ (Á) + Vr (Á)
©(μ) = ¡ dÁ: (11) that the name “capture length” is more suitable for the
0 Vμ (Á)
present approach, since the admissible engagement
Each guidance law possesses a unique characteristic conditions are expressed as a length interval of h0 .
function from which all the trajectory properties are To be consistent with the conventional usage, we
inherited. The characteristic functions for the six still adopt the name capture area in the following
guidance laws are derived in the next section. After discussion, but we want to remind the readers of
obtaining h(μ), we can find the LOS angular rate from the fact that when capture area is normalized to
the following relation: dimensionless form, it becomes a length not an area.
2 Capture length can be interpreted as a quantitative
dμ V measure of capture area and makes it possible the
= μ: (12)
d¿ h quantitative comparison of capture areas for different
Integrating both sides gives the required time of PN laws. Recall that the sample space of capture
interception length is fh0 j 0 · h0 · 1g, p hence, the capture length
Z μf Z μf ¡©(μ) of TPN with ¸ = 2 covers 3=2 portion of the whole
dμ e sample space. That is to say, when the engagement
¿f = = h0 2
dμ: (13)
0 dμ=d¿ starts with random initial condition,
0 V (μ) μ p the probability of
successful interception is about 3=2.
All the six guidance laws can be analyzed by the
The cumulative velocity increment for TPN comes
same procedures mentioned above. Uuder this unified
from (8) with u(μ) = ¡ sin Ã0 and v(μ) = 0
approach, the only difference in analyzing these
guidance laws is the vector ~L. By considering six Z μf q
types of ~L, the performances of the corresponding ¢VTPN = ¸ u2 (μ) + v2 (μ)dμ
0
six guidance laws can be evaluated and compared in
a standard framework. = ¸μf sin Ã0 : (17)

The characteristic function for TPN is found by

III. PERFORMANCE EVALUATION OF SIX GUIDANCE substituting (14) into (11)
LAWS μ ¶
μ+± ¼
tan +
A. TPN ¸ sin Ã0 2 4
©TPN = ln μ ¶ : (18)
R ± ¼
From the definition of TPN, we have ~LTPN = r_0~er , tan +
2 4
i.e., u(μ) = r_0 and v(μ) = 0, or in dimensionless form:
u(μ) = ¡ sin Ã0 , v(μ) = 0. Substituting u(μ) and v(μ) Substituting ©TPN into (13), we obtain the required
into (6), and solving the resulting coupled linear time of interception
equations, we obtain readily · ¸
h cot(±=2 + ¼=4) tan(±=2 + ¼=4)
Vr (μ) = R sin(μ + ±) ¡ ¸ sin Ã0 (14a) ¿fTPN = 0 + :
2R ¸ sin Ã0 + R ¸ sin Ã0 ¡ R
Vμ (μ) = R cos(μ + ±) (14b) (19)

where
q B. RTPN
R = k 2 sin2 Ã0 + cos2 Ã0 , tan ± = k tan Ã0 :
RTPN is the practical implementation of TPN
The definition k = ¸ ¡ 1 is used throughout this work. using instantaneous closing speed in the interceptor’s
The condition Vμ (μf ) = 0 gives the terminal aspect acceleration. For RTPN, we have ~LRTPN = r_~er , i.e.,
angle μf as u = Vr and v = 0 which are then substituted into (6) to
¼ yield the governing equations of RTPN
μf = ¡ ± = arccot(k tan Ã0 ): (15)
2 0
Vr ¡ Vμ = 0 (20a)
Using above μf in the capture condition Vr (μf ) · 0
0
and expressing the resultant in terms of h0 , we obtain Vμ + Vr = ¸Vr : (20b)

YANG & YANG: A UNIFIED APPROACH TO PROPORTIONAL NAVIGATION 559

These coupled linear differential equations can be v(μ) = r0 μ_0 , or in dimensionless form: u(μ) = ¡ sin Ã0 ,
solved easily, leading to the following solutions v(μ) = cos Ã0 . Substituting u(μ) and v(μ) into (6) yields
cos à p p the governing equations of GTPN
Vr (μ) = p 0 sinh kμ ¡ sin Ã0 cosh kμ (21a) 0
k Vr ¡ Vμ = ¡¸ cos Ã0 (28a)
p p p
Vμ (μ) = cos Ã0 cosh kμ ¡ k sin Ã0 sinh kμ: 0
Vμ + Vr = ¡¸ sin Ã0 (28b)
(21b)
whose solutions can be derived easily as
The terminal aspect angle μf evaluated from the
Vr (μ) = k sin(μ + ±) ¡ ¸ sin Ã0 (29a)
condition Vμ (μf ) = 0 is found as
p Vμ (μ) = k cos(μ + ±) + ¸ cos Ã0 (29b)
1 k sin Ã0 + cos Ã0
μf = p ln p (22) where ± = ¼ ¡ Ã0 . The condition Vμ (μf ) = 0 gives the
2 k k sin Ã0 ¡ cos Ã0
terminal aspect angle μf as
which can be used in the capture condition Vr (μf ) · 0 μ ¶
¡¸
to derive the capture area of RTPN μf = arccos cos Ã0 ¡ ¼ + Ã0 (30)
r k
1
0 · h0 < 1 ¡ : (23) and the inequality Vr (μf ) · 0 is reduced to the
¸ following form
As expected, the capture area of RTPN is smaller 1
0 · h0 · 1 ¡ (31)
than that of TPN, when we compare (23) with (16). ¸
Using (8), the cumulative velocity increment of RTPN which is the capture area of GTPN. An alternative
becomes definition of GTPN [6] is u(μ) = C1 and v(μ) = C2 ,
Z μf
¸ where constants C1 and C2 are independent of
¢VRTPN = ¡¸ Vr (μ) dμ = cos Ã0 : (24)
0 k initial conditions. Since one more free parameter is
introduced in this case, the capture area can be larger
Next we consider the required time of interception
than that derived above. As to the cumulative velocity
of RTPN. The characteristic function of RTPN is
increment for GTPN, we have from (8)
determined by substituting (21) into (11)
Z μf q
¸ p p p ¢VGTPN = ¸ u2 (μ) + v2 (μ) dμ = ¸μf (32)
©RTPN (μ) = ¡ ln jcosh kμ ¡ k tan Ã0 sinh kμj 0
k
(25) where μf has been derived in (30). Substituting (29)
which, in turn, is applied in (13) to obtain ¿fRTPN as into (11) yields the characteristic function of GTPN
Z μf
Z μf p ¸ sin Ã0
¡1=¸¡1 ©GTPN (μ) = dμ
¿fRTPN = h0 ( cos Ã0 cosh kμ 0 k cos(μ + ±) + ¸ cos Ã0
0
p p = F(μ + ±) ¡ F(±) (33)
¡ k sin Ã0 sinh kμ)(2¡¸)=¸¡1 dμ:
(26) where
¸ sin Ã0
A more concise expression of ¿fRTPN can be found by F(x) = p
k2 ¡ ¸2 cos2 Ã0
exploiting the following relation between μ and the ¯ ¯
normalized relative distance ½ ¯ (k ¡ ¸ cos à ) tan(x=2) + pk 2 ¡ ¸2 cos2 à ¯
¯ 0 0¯
q £ ln ¯ p ¯:
p p
kμ k 2 2 ¯ (k ¡ ¸ cos Ã0 ) tan(x=2) ¡ k 2 ¡ ¸2 cos2 Ã0 ¯
(cos Ã0 ¡ k sin Ã0 )e = h0 ½ ¡ h0 ½2k ¡ ¸h0 + k:
Having obtained ©GTPN (μ), we proceed to find the
In terms of ½, the integration in (26) becomes
required time of interception of GTPN by using
Z 1 p
k d½ ©GTPN (μ) in (13)
¿fRTPN = q : (27)
0 2 2k 2 Z
h0 ½ ¡ ¸h0 + k μf
e¡©GTPN (μ)
¿fGTPN = h0 dμ: (34)
0 (k cos(μ + ±) + ¸ cos Ã0 )2
C. GTPN
D. IPN
GTPN is another generation of TPN, where the
interceptor’s acceleration keeps a fixed angle with Just like the fact that RTPN is the implemented
respect to the normal direction of LOS. We have from version of TPN, one can conceive of IPN as the
~0 = r_0~er + r0 μ_0~eμ , i.e., u(μ) = r_0 and
(1c) that ~LGTPN = V implemented version of GTPN. Instead of ~LGTPN = V ~0

560 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 33, NO. 2 APRIL 1997
used in GTPN, IPN employs the instantaneous relative
~ (μ) to generate the interceptor’s
velocity ~LIPN = V
acceleration. Hence, in the case of IPN, we have
_ and (6) becomes
u(μ) = r_ and v(μ) = rμ,
0
Vr ¡ Vμ = ¡¸Vμ (35a)
0
Vμ + Vr = ¸Vr : (35b)
Fig. 1. Planar pursuit geometry.
Solving for Vr (μ) and Vμ (μ), we obtain
Vr (μ) = ¡ sin(kμ + Ã0 ), Vμ (μ) = cos(kμ + Ã0 ): via ® = ¸μ + ®0 . Now the normal direction ~LPPN
(36) of the acceleration vector ~aI can be expressed
in the LOS-based coordinates as ~LPPN = ¡V ~I =
The terminal aspect angle μf evaluated from the ¡VI cos(kμ + ®0 )~er ¡ VI sin(kμ + ®0 )~eμ , i.e.,
condition Vμ (μf ) = 0 is
u(μ) = ¡VI cos(kμ + ®0 ), v(μ) = ¡VI sin(kμ + ®0 ):
1 ³¼ ´
μf = ¡ Ã0 (37) (41)
k 2
and the capture condition Vr (μf ) = ¡1 · 0 is always Substituting (41) into (6) yields the governing
satisfied for any initial conditions of engagement, as equations of PPN
long as ¸ > 1. Hence, the capture area of IPN covers 0
all range of h0 , i.e., 0 · h0 · 1. Next we consider the Vr ¡ Vμ = ¸VI sin(kμ + ®0 ), Vr (0) = ¡ sin Ã0
cumulative velocity increment of IPN. From (8), we (42a)
have
¸ ³¼ ´ 0
Vμ + Vr = ¡¸VI cos(kμ + ®0 ), Vμ (0) = cos Ã0 :
¢VIPN = ¸μf = ¡ Ã0 : (38)
k 2 (42b)
It can be seen that ¢VIPN ! 0, as Ã0 ! ¼=2. This These coupled linear differential equations can be
corresponds to the tail-chase condition. On the other solved readily for Vr (μ) and Vμ (μ) as
hand, ¢VIPN achieves its maximum as Ã0 approaches
0, which corresponds to the case that the target Vr (μ) = (Vr (0) + VI cos ®0 ) cos μ
escapes initially along the normal direction of LOS.
As to the characteristic function of IPN, we substitute + (Vμ (0) + VI sin ®0 ) sin μ ¡ VI cos(kμ + ®0 )
(36) into (11) to obtain (43a)
¡¸ cos(kμ + Ã0 ) Vμ (μ) = (Vr (0) + VI cos ®0 ) sin μ
©IPN (μ) = ln : (39)
k cos Ã0
+ (Vμ (0) + VI sin ®0 ) cos μ ¡ VI sin(kμ + ®0 ):
We can determine the required time of interception of (43b)
IPN from the (13) with Vμ (μ) and ©(μ) given by (36)
and (39), respectively. Above solutions are in terms of the initial conditions
Vr (0) and Vμ (0). To compare with the conventional
E. PPN result where solutions are expressed in terms of the
~I and V
initial orientations of V ~T , we need to relate
The relative motion of PPN adopts the Vr (0) and Vμ (0) to ®0 and ¯0 . Referring to (1), we
interceptor’s velocity as the reference line. Therefore, have
it is a common belief that PPN cannot be analyzed
using the LOS-based coordinates. However, we will Vr (0) = VT cos ¯0 ¡ VI cos ®0 (44a)
show below that the unified methodology, which Vμ (0) = VT sin ¯0 ¡ VI sin ®0 (44b)
employs the LOS-based framework and is applied
successfully to the above four guidance laws, is also Using (44) in (43) leads to the familiar result:
applicable to the PPN case. The first step in this
~I
formulation is to express the interceptor’s velocity V Vr (μ) = VT cos(¯0 ¡ μ) ¡ VI cos(μ ¡ ®) (45a)
in terms of the LOS-based coordinates. Referring to
Vμ (μ) = VT sin(¯0 ¡ μ) + VI sin(μ ¡ ®): (45b)
Fig. 1, we have
~I = VI cos(μ ¡ ®)~er ¡ VI sin(μ ¡ ®)~eμ
V (40) Though the identical result can also be obtained from
direct observation of the kinematic diagram, here
where ® is the aspect angle of V ~I with respect to we have rederived the key equations of PPN using a
the inertial reference line. From the definition of unified framework where LOS-based guidance laws
PPN, the magnitude of the interceptor’s acceleration such as TPN, RTPN, GTPN, and IPN have been
_ Hence, ® is related to μ
aI is aI = VI ®_ = ¸VI μ. analyzed.

YANG & YANG: A UNIFIED APPROACH TO PROPORTIONAL NAVIGATION 561

 Define two parameters ³ and ´ such that ¯0 = (12). The second term in I measures the energy
³®0 , and VI = ´VT , then it can be shown that VT = consumption during the interception. The optimal
1=(´ ¡ 1), VI = ´=(´ ¡ 1), and ®0 = Ã0 ¡ ¼=2. u(μ) and v(μ) can be determined using the standard
Accordingly, (45) can be rewritten as variational approach. We rewrite u and v in the way
1 ´ u = ¡M cos ¾, v = ¡M sin ¾. Hence, the governing
Vr (μ) = ¡ sin(μ ¡ Ã0 ) ¡ sin(kμ + Ã0 ) equations (6) become
´¡1 ´¡1
0
(46a) Vr ¡ Vμ = ¡¸v = ¸M sin ¾ (51a)
¡1 ´ 0
Vμ + Vr = ¸u = ¡¸M cos ¾ (51b)
Vμ (μ) = cos(μ ¡ Ã0 ) + cos(kμ + Ã0 )
´¡1 ´¡1
(46b) and the corrresponding Hamiltonian is

where for comparison purpose, we only consider the H = °M 2 + ¸Vr (Vμ + ¸M sin ¾)
case of ³ = 1. When other values of ³ are concerned,
they can be treated in the same way. The terminal + ¸Vμ (¡Vr ¡ ¸M cos ¾) (52)
aspect angle μf is determined from the condition where the adjoint variables ¸Vr and ¸Vμ satisfy
Vμ (μf ) = 0, i.e.,
@H
cos(μf ¡ Ã0 ) = ´ cos(kμf + Ã0 ): (47) ¸0V = ¡ = ¸Vμ , ¸Vr (μf ) = 0 (53a)
r @Vr
It was shown by qualitative methods [7] that Vr (μf ) · @H
0 is always satisfied for any initial conditions, ¸0V = ¡ = ¡¸Vr , ¸Vμ (μf ) = free: (53b)
provided ¸ > 2 and ´ > 1. Hence, the capture area of
μ @Vμ
PPN covers all range of h0 , i.e., 0 · h0 · 1. Regarding The solutions of (53) can be easily derived as
to the cumulative velocity increment of PPN, we have
the following result from (8) and (41) ¸Vr (μ) = A sin(μ ¡ μf ),
Z μf q (54)
¸Vμ (μ) = A cos(μ ¡ μf )
¢VPPN = ¸ u2 (μ) + v2 (μ) dμ
0
where A and μf are to be determined. The conditions
¸´ @H=@M = 0 and @H=@¾ = 0 give, respectively, the
= ¸VI μf = μ : (48)
´¡1 f optimal M and ¾ as
The characteristic function of PPN can be found by ¸
substituting (46) into (11), leading to M¤ = (¸ cos ¾ ¡ ¸Vr sin ¾),
2° Vμ
Z μ (55)
¸´ sin(kμ + Ã0 ) ¸Vr
©PPN (μ) = dμ ¤
tan ¾ = ¡ :
0 ´ cos(kμ + Ã0 ) ¡ cos(μ ¡ Ã0 ) ¸Vμ
(49)
The combination of (54) and (55) reveals that
©PPN (μ) is then exploited to determine the required M ¤ = ¸A(2°)¡1 and ¾¤ = μf ¡ μ. The optimal pair of
time of interception by using (13). (u(μ), v(μ)) turns out to be
¸A
F. OPN u¤ (μ) = ¡ cos(μ ¡ μf ),
2°
The interceptor’s acceleration of PN guidance (56)
¸A
laws possesses the general form: ~aI = ¸~L £ ~! , where v¤ (μ) = sin(μ ¡ μf ):
2°
~L = u(μ)~er + v(μ)~eμ . In the above discussion we have
considered five PN guidance laws, with each guidance The optimal trajectory is governed by (51) with u and
law associated with a specific pair of (u(μ), v(μ)). v replaced by (56)
It is natural to raise the question that which pair of
(u(μ), v(μ)) is the best choice? The answer depends on 0 ¸2 A
Vr ¡ Vμ = + sin(¡μ + μf ) (57a)
the performance index used for the optimization. Here 2°
we adopt the following performance index: ¸2 A
0
Z μf Vμ + Vr = ¡ cos(¡μ + μf ): (57b)
2°
I = μf + ° (u2 (μ) + v2 (μ)) dμ (50)
0 The similarity between (42) and (57) manifests the
where the first term in I takes into account the time close connection between PPN and OPN. This point is
of interception, since μ is a monotonically increasing investigated in the next section. Solving (57) with the
or decreasing function of time t as can be seen from initial conditions Vr (0) = ¡ sin Ã0 and Vμ (0) = cos Ã0

562 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 33, NO. 2 APRIL 1997
and with the terminal conditions Vμ (μf ) = 0 and A5) PPN:
H(μf ) = ¡1, we obtain ´
u(μ) = ¡ sin(kμ + Ã0 ),
´¡1
Vr (μ) = + sin(μf ¡ Ã0 ) cos(μ ¡ μf )
´
v(μ) = cos(kμ + Ã0 )
¸2 A ´¡1
¡ (μ ¡ μf ) sin(μ ¡ μf ) (58a)
2°
A6) OPN:
Vμ (μ) = ¡ sin(μf ¡ Ã0 ) sin(μ ¡ μf ) ¡1
u(μ) = cos(μf ¡ Ã0 ) cos(μ ¡ μf ),
¸2 A ¸μf
¡ (μ ¡ μf ) cos(μ ¡ μf ) (58b)
2° 1
¡1
v(μ) = cos(μf ¡ Ã0 ) sin(μ ¡ μf )
2
where A = 2°(¸ μf ) cos(μf ¡ Ã0 ) and μf is the ¸μf
smallest root satisfying
Three remarks can be addressed from the
¸2 2 trajectories of ~L(μ) within the inertial X ¡ Y
μ = μf sin 2(μf ¡ Ã0 ) + cos2 (μf ¡ Ã0 ): (59) coordinates for various guidance laws. 1) The
° f
trajectories of ~LPPN with varying speed ratio ´
The capture area described by Vr (μf ) · 0 becomes converge to the trajectory of ~LIPN , as ´ approaches
0 · h0 · cos μf : (60) infinity. This observation reflects the fact that IPN is
the extreme case of PPN. 2) The trajectory of ~LGTPN
The cumulative velocity increment of OPN is coincides with the trajectory of ~LIPN . In fact, both
calculated from (8) and (56) trajectories lie on the unit circle, because of u2 (μ) +
Z μf q v2 (μ) = 1 for both case. However, it should be noted
¢VOPN = ¸ u2 (μ) + v2 (μ) dμ that at any instance the orientations of ~LGTPN and ~LIPN
0 may not coincide. 3) Of most significance is that the
2
¸ A orientation of ~LOPN is fixed in the inertial coordinates.
= μ = cos(μf ¡ Ã0 ): (61) This observation can be checked theoretically by
2° f
showing ~LOPN is a constant vector.
The substitution of (58) into (11) yields the B) Terminal Aspect Angle: μf .
characteristic function of OPN as B1) TPN:
Z μf
cos(μf ¡ Ã0 ) cos(μ ¡ μf )
©OPN (μ) = dμ: μf = arccot(k tan Ã0 )
0 μf cos(μ ¡ Ã0 ) ¡ μ cos(μf ¡ Ã0 ) cos(μ ¡ μf )
(62) B2) RTPN:
p
1 k sin Ã0 + cos Ã0
IV. COMPARISON AND RELATIONSHIP μf = p ln p
2 k k sin Ã0 ¡ cos Ã0
In the previous section, the performance and B3) GTPN:
trajectories of the six guidance laws have been μ ¶
analyzed in a unified framework. For the sake of ¡¸
μf = arccos cos Ã0 + Ã0 ¡ ¼
comparison and for the convenience of deducing the k
relationship among the guidance laws, it is better to B4) IPN:
put together the characteristics of the various guidance 1 ³¼ ´
laws. μf = ¡ Ã0
k 2
A) Acceleration Normal Direction: ~L = u(μ)~er +
v(μ)~eμ . B5) PPN:
A1) TPN: cos(μf ¡ Ã0 ) = ´ cos(kμf + Ã0 )
u(μ) = ¡ sin Ã0 , v(μ) = 0 B6) OPN:
A2) RTPN: ¸2 2
cos à p p μ = μf sin 2(μf ¡ Ã0 ) + cos2 (μf ¡ Ã0 )
u(μ) = p 0 sinh kμ ¡ sin Ã0 cosh kμ, v(μ) = 0 ° f
k
C) Cumulative Velocity Increment: ¢V.
A3) GTPN: C1) TPN:
u(μ) = ¡ sin Ã0 , v(μ) = cos Ã0 ¢V = ¸μf sin Ã0

A4) IPN: C2) RTPN:

¸
u(μ) = ¡ sin(kμ + Ã0 ), v(μ) = cos(kμ + Ã0 ) ¢V = cos Ã0
k

YANG & YANG: A UNIFIED APPROACH TO PROPORTIONAL NAVIGATION 563

 C3) GTPN: E6) OPN:
¢V = ¸μf Z μf
e¡©OPN (μ)
¿f = h0 μf2 dμ
C4) IPN: [μf cos(μ ¡ Ã0 ) ¡ μ cos(μ ¡ μf ) cos(μf ¡ Ã0 )]2
¸ ³¼ ´ 0

¢V = ¡ Ã0 where the first four characteristic functions ©TPN (μ),

k 2
©RTPN (μ), ©GTPN (μ), and ©IPN (μ) have been expressed
C5) PPN:
¸´ in terms of the elementary functions of μ. Based on
¢V = μ the above lists of performance characteristics, some
´¡1 f
relations among the various guidance laws can be
C6) OPN: revealed immediately.
¢V = cos(μf ¡ Ã0 ) Relation 1): lim´!1 PPN(´) = IPN. The first
relation we have is that when the interceptor has an
D) Capture Area. absolute speed superiority over the target, then PPN
D1) TPN: and IPN become indistinguishable. The performance
( r )
of PPN depends on the value of ´ which is the speed
1
h0 j 0 · h0 · 1¡ ratio of the interceptor over the target. We denote
2¸
this performance dependence by PPN(´). If we let
D2) RTPN: ´ ! 1, all the PPN formula listed above converge
( r ) to those of IPN. Mathematically speaking, we have
1 lim´!1 PPN(´) = IPN. Although IPN may not be of
h0 j 0 · h0 · 1¡
¸ practical use, the study of IPN can, at least, tell us the
limits of the performance of PPN.
D3) GTPN: Relation 2): TPN = RTPN = GTPN = IPN, as
½ ¾
1 Ã0 ! ¼=2. The second relation says that under the
h0 j 0 · h0 · 1 ¡ tail-chase situation, the four guidance laws, namely,
¸
TPN, RTPN, GTPN, and IPN, become identical. It is
D4) IPN: noted that when Ã0 = ¼=2, h0 = cos(¼=2) = 0. From
fh0 j 0 · h0 · 1g the definition of h0 , h0 = 0 represents the situation
D5) PPN: where the target escapes initially along the LOS. From
fh0 j 0 · h0 · 1g the performance lists A—E, we can observe that when
Ã0 ! ¼=2, all the formula of TPN, RTPN, GTPN,
D6) OPN: and IPN become identical. It is also noted that μ(t) =
fh0 j 0 · h0 · cos μf g μf = 0, t ¸ 0, for the four guidance laws, if Ã0 = ¼=2.
This is simply the case that the interceptor follows
As capture area is concerned, we have PPN = IPN ¸ a straight-line course during the interception for a
TPN ¸ RTPN ¸ GTPN. The capture area of OPN is nonmaneuvering target under tail-chase condition.
not fixed, depending on the weighting factor °. Relation 3): limμ!μf OPN = PPN. The third
E) Time of Interception: ¿f . relation says that toward the end of interception,
E1) TPN: the behavior of OPN and PPN becomes closer and
Z μf ¡©TPN (μ)
h e closer. The behavior of PPN depends on the speed
¿f = 0 dμ ratio ´, while the behavior of OPN depends on the
R 0 cos2 (μ + ±)
weighting factor °. It is possible for certain choice
E2) RTPN: of ´ and ° that the behavior of OPN and PPN can
Z μf
e¡©RTPN (μ) be similar. From (42) and (58), the interceptor’s
¿f = h0 p p p dμ acceleration according to PPN and OPN can be
2
0 (cos Ã0 cosh kμ ¡ k sin Ã0 sinh kμ)
described, respectively, as
E3) GTPN: _ sin(kμ + ® )~e + cos(kμ + ® )~e ]
Z (~aI )PPN = ¸VI μ[¡ 0 r 0 μ
μf
e¡©GTPN (μ)
¿f = h0 dμ (63a)
0 [k cos(μ + ±) + ¸ cos Ã0 ]2
¸2 A _
E4) IPN: (~aI )OPN = μ[¡ sin(¡μ + μf )~er + cos(μ ¡ μf )~eμ ]:
Z 2°
μf
e¡©IPN (μ) (63b)
¿f = h0 dμ
0 cos2 (kμ + Ã0 )
Under certain design conditions such as ° = ¸A=(2VI ),
E5) PPN: ®0 = ¡μf , k = ¸ ¡ 1 = 1, we have limμ!μf (~aI )PPN =
Z μf _e . The interpretation of this
e¡©PPN (μ) lim (~a ) = ¸V μ~
¿f = h0 (´ ¡ 1)2 dμ μ!μf I OPN I μ
0 [¡ cos(μ ¡ Ã0 ) + ´ cos(kμ + Ã0 )]2 result is that PPN is an optimal guidance law from the

564 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 33, NO. 2 APRIL 1997
 Fig. 2. Relations among 6 guidance laws.

Fig. 4. Time of interception with ¸ = 6.

Fig. 3. Time of interception with ¸ = 2:5.

sense of the optimization of the performance index I

in (50) with propor choice of weighting factor °. This
fact may somewhat explain the superior performance
of PPN. Fig. 2 summarizes the relations among the six
guidance laws. Fig. 5. Cumulative velocity increment with ¸ = 2:5.

V. NUMERICAL RESULTS

Using the mathematical formula derived

previously, the six guidance laws can be compared
on the same basis, and more physical insights can
be gained from numerical studies. The influence of
parameters ¸, Ã0 , ´, and °, are discussed separately.

A. The Effect of ¸

Two cases with ¸ = 2:5 and ¸ = 6 are considered

here. It is observed that when ¸ is changed from 2.5 Fig. 6. Cumulative velocity increment with ¸ = 6:0.
to 6, the required time of interception ¿f is reduced
for every guidance law (see Figs. 3 and 4), and at the
same time, the cumulative velocity increment of every to the worst-case scenario where the target escapes
guidance law is also decreased (see Figs. 5 and 6). As purely in the tangential direction. When the target’s
expected, the navigation constant ¸ has a similar effect motion is near the tail-chase condition (i.e., small h0 ),
on each PN scheme, and the performance gets better there is no significant difference in the behavior of the
with an increasing ¸. A ¸ between 3 and 6 would be a six PN schemes as can be seen from Figs. 3 and 5.
good choice for global performance. A ¸ greater than However, when h0 is increased toward its maximum
6 is not recommanded to avoid instability caused by allowable value, drastic degradation of performance
time lags and sensor noises. is observed in most PN schemes. The maximum
allowable h0 for each PN scheme is determined by
the capture conditions listed in property D1—D6.
B. The Effect of Ã0 (h0 ) According to the performance at (h0 )max , the six
The PN schemes can be divided into two groups. The
q variable Ã0 is defined as h0 = cos Ã0 = first group consists of TPN, RTPN, and OPN with
r0 μ_0 = r_02 + r02 μ_02 . The value of h0 reflects the tendence large °. The required time of interception in this
of the target escaping in the tangential direction. group exhibits an abrupt ascent when h0 is increased
Hence, Ã0 = ¼=2 (h0 = 0) stands for the tail-chase toward (h0 )max as shown in Figs. 3 and 4. However,
condition where the target escapes along the radial the associated increase in the cumulative velocity
direction (LOS), while Ã0 = 0 (h0 =1) corresponds increment ¢V is relatively small in this group, as

YANG & YANG: A UNIFIED APPROACH TO PROPORTIONAL NAVIGATION 565

can be seen from Figs. 5 and 6. The second group
which possesses the reverse properties of group 1
comprises PPN, IPN, GTPN, and OPN with small °.
In second group, the increase in the required time of
interception is insignificant even near the worst-case
condition ((h0 )max ! 1); however, the cumulative
velocity increment grows precipitously (see Figs. 5
and 6). Among the six PN schemes, OPN is the only
one that is able to compromise the required time of
interception with the required energy consumption by
proper choice of the weighting factor °. An OPN with Fig. 7. Trajectories of h(μ) with h0 = 0:5.
large ° puts main concern on energy consumption,
and the resulting performance is similar to TPN
(group 1); while an OPN with small ° put major
penality on the time of interception, and the resulting
performance is close to PPN (group 2). The OPN
with ° = 1 having equal weights on both requirements
seems to be a good compromise (see Figs. 3 and 5).

C. The Effect of ´
The variable ´ is the speed ratio of the interceptor
over the target. Two speed ratios ´ = 2 and ´ = 5
are considered in the numerical comparison. It is Fig. 8. Trajectories of h(μ) with maximum allowable h0 .
found that the PPN with ´ = 2 demonstrates shorter
interception time than that of PPN with ´ = 5 (see
Figs. 3 and 4), but requires more energy consumption case of h0 = 0:5, indicating that the convergence of the
(see Figs. 5 and 6). As has been proved in the last relative motion is more slow.
section, IPN is the extreme case of PPN as ´ ! 1.
Hence, we can see from Figs. 3 and 4 that the ¿f
curve of IPN constitutes an upper envelop of the VI. CONCLUSIONS
curves of PPN with various speed ratios, and from
Figs. 5 and 6 we can observe that the ¢V curve of An attempt has been made to unify the
IPN constitutes a lower envelop of the various PPN mathematical tools used in analyzing the existing PN
curves. Because the distinction between the PPN with guidance laws. With the introduction of a generalized
´ = 2 and the PPN with ´ = 1 (IPN) is not apparent, PN scheme, the six guidance laws, namely, TPN,
a PPN scheme employing large speed ratio seems RTPN, GTPN, IPN, PPN, and OPN, are included as
impractical and unnecessary. its special cases, and are solved in a unified manner.
Along this approach, the experiences gained fron
D. The Effect of ° the analytical study of TPN-type guidance laws can
be equally applied to the PPN-type guidance laws.
The weighting factor ° has a remarkable influence The concept proposed in this paper may hopefully
on OPN. The numerical results show that the other pave the way for combining the advantages of TPN in
five PN schemes can be roughly approximated by mathematical tractability with the advantages of PPN
OPN with certain choice of °. Generally speaking, in practical implementation.
the behavior of OPN with small ° is close to the
behavior of PPN, while the OPN with large ° is close REFERENCES
to TPN. Fig. 7 shows that the h(μ) trajectory of OPN
with ° = 1 almost coincides with that of PPN with [1] Becker, K. (1990)
Closed-form solution of pure proportional navigation.
´ = 5, and Fig. 8 shows that the trajectory of OPN IEEE Transactions on Aerospace and Electronic Systems,
with ° = 8:5 is close to that of TPN. Figs. 7 and 8 26 (Apr. 1990), 526—533.
also manifest the difference in the convergent speed [2] Cochran, J. E., No, T. S., and Thaxton, D. G. (1991)
of the relative motion for different h0 . In Fig. 7 h0 is Analytical solutions to a guidance problem.
set to 0.5 for all six guidance laws, and in Fig. 8 h0 is Journal of Guidance, Control, and Dynamics, 14 (Jan.—Feb.
1991), 117—122.
set to its maximum allowable value that corresponds
[3] Dhar, A., and Ghose, D. (1993)
to the worst-case condition of each guidance law. It Capture region for a realistic TPN guidance law.
can be observed that under the worst-case condition, IEEE Transactions on Aerospace and Electronic Systems,
the trajectories of h(μ) are more sluggish than the 29 (July 1993), 995—1003.

566 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. 33, NO. 2 APRIL 1997
 [4] Ghose, D. (1994) [12] Shukla, U. S., and Mahapatra, P. R. (1990)
True proportional navigation with maneuvering target. The proportional navigation dilemma–Pure or true?
IEEE Transactions on Aerospace and Electronic Systems, IEEE Transactions on Aerospace and Electronic Systems,
30 (Jan. 1994), 229—237. 26 (Mar. 1990), 382—392.
[5] Ghose, D. (1994) [13] Yang, C. D., Yeh, F. B., and Chen, C. H. (1987)
Capture region for true proportional navigation guidance The closed-form solution of generalized proportional
with nonzero miss-distance. navigation.
Journal of Guidance, Control, and Dynamics, 17 Journal of Guidance, Control, and Dynamics, 10
(May—June 1994), 627—628. (Mar.—Apr. 1987), 216—218.
[6] Ghose, D. (1994) [14] Yang, C. D., Hsiao, F. B., and Yeh, F. B. (1989)
On the generalization of true proportional navigation. Generalized guidance law for homing missiles.
IEEE Transactions on Aerospace and Electronic Systems, IEEE Transactions on Aerospace and Electronic Systems,
30 (Apr. 1994), 545—555. 25 (Mar. 1989), 197—212.
[7] Guelman, M. (1971) [15] Yuan, P. J., and Chern, J. S. (1992)
A qualitative study of proportional navigation. Solutions of true proportional navigation for maneuvering
IEEE Transactions on Aerospace and Electronic Systems, and non-maneuvering targets.
AES-7 (July 1971), 638—643. Journal of Guidance, Control, and Dynamics, 15 (Jan.—Feb.
[8] Guelman, M. (1972) 1992), 268—271.
Proportional navigation with a maneuvering target. [16] Yuan, P. J., and Chern, J. S. (1992)
IEEE Transactions on Aerospace and Electronic Systems, Ideal proportional navigation.
AES-8 (May 1972), 364—371. Journal of Guidance, Control, and Dynamics, 15
[9] Guelman, M. (1976) (Sept.—Oct. 1992), 1161—1165.
The closed-form solution of true proportional navigation. [17] Yuan, P. J., and Chern, J. S. (1993)
IEEE Transactions Aerospace and Electronic Systems, Exact closed-form solution of generalized proprotional
AES-12 (July 1976), 472—482. navigation.
[10] Mahapatra, P. R., and Shukla, U. S. (1989) Journal of Guidance, Control, and Dynamics, 16
Accurate solution of proportional navigation for (Sept.—Oct. 1993), 963—966.
maneuvering targets. [18] Zarchan, P. (1994)
IEEE Transactions on Aerospace and Electronic Systems, Tactical and Strategic Missile Guidance (2nd ed.), Vol.
25 (Jan. 1989), 81—89. 157: Progress in Astronautics and Aeronautics.
[11] Murtaugh, S. A., and Criel, H. E. (1966) New York: AIAA Inc., 1994.
Fundamental of proportional navigation.
IEEE Spectrum, 3 (Dec. 1966), 75—85.

Ciann-Dong Yang received his B.S., M.S., and Ph.D. degrees in aeronautics and
astronautics from National Cheng Kung University, Taiwan, in 1983, 1985, and
1987, respectively.
Since 1985, he has been with the Department of Aeronautics and
Astronautics, National Cheng Kung University, where he is currently a full
Professor. His research interests are in H1 robust control and missile guidance.
Dr. Yang is the recipient of the 1990 M. Barry Carlton Award and the
Outstanding Paper Award of the IEEE Aerospace and Electronic Systems. He
is also the author of Engineering Analysis and is the coauthor of Post Modern
Control Theory and Design.

Chi-Ching Yang was born in Lukang, Taiwan, on Aug. 18, 1969. He received
his B.S. degree in mechanical engineering from Chung Yuan University, Taiwan,
in 1991 and the M.S. and Ph.D. degrees in aeronautics and astronautics from
National Cheng Kung University, Taiwan, in 1993 and 1996.
He is currently a postdoctorate researcher in the Institute of Aeronautics and
Astronautics, National Cheng Kung University, Taiwan. His research interests are
mainly in missile guidance and control theory.

YANG & YANG: A UNIFIED APPROACH TO PROPORTIONAL NAVIGATION 567