Reprinted From: 2008 IEEE Sarnoff Symposium,

Princeton, NJ, USA; 28-30 April, 2008, Pages:1 - 5.

©2005 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for
advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or SM
lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

A Fundamental Limit on Antenna Gain for

Electrically Small Antennas
Andrew J. Compston, James D. Fluhler, and Hans G. Schantz

Ranging (NFER)2 that takes advantage of the fact that the

Abstract—A fundamental limit on an antenna’s gain is derived phase difference of a transmitting antenna’s electric and
and compared to measurements taken on a number of different magnetic fields goes to zero as the distance from the antenna
antennas. First, a propagation formula applicable in both the increases [2].
near and far fields is developed, and that result is used to
Often near-field applications like these use low frequencies
demonstrate that the gain of an antenna is limited by its electrical
size. in order to fully realize the advantages of the near field at
substantial distances. Compare the wavelength of a typical
Index Terms—Electrically Small Antennas, Antenna Gain, Wi-Fi signal (12.5 cm) with the wavelength of a typical
Antenna Measurements, Near Field. Q-Track NFER signal (1 MHz, 300 m). As a consequence,
most antennas used in near-field systems are much smaller
I. INTRODUCTION than a wavelength.
This paper exploits near-field phenomena in order to derive
T HE area around antennas is often split into two areas: the
near field, which extends out from the antenna at a
distance comparable to a wavelength, and the far field, which
a fundamental limit on the gain of an antenna versus its
electrical size. First, we derive a propagation formula similar
is the area beyond the near field. See [1] for an excellent to Friis’s formula for the far field. Using this formula, we
discussion of the properties of the near field, the far field, and show a fundamental limit on the gain of an antenna versus its
exactly where the boundary between the two lies. For most electrical size, and we compare the limit to a number of gain
antenna applications, especially those at high frequencies, the measurements taken on actual antennas.
behavior of the antenna’s electromagnetic fields in the near
field is often of little consequence. For example, a typical Wi- II. FRIIS’S PROPAGATION FORMULA AND THE FAR FIELD
Fi1 signal operating at 2.4 GHz has a wavelength of about 12.5 Harald Friis derived the propagation formula that bears his
cm. Because most Wi-Fi applications operate at ranges on the name in the following form [3]
order of meters, this system lends itself well to far-field PRX A A
analysis. This is the case for almost every electromagnetic = RX TX . (1)
PTX ( dλ ) 2
system in common use today.
However, the near field has some interesting properties that PRX is the power received by the receiving antenna, PTX is the
some systems exploit. As we will demonstrate, the power power transmitted by the transmitting antenna, ARX is the
transmitted by a near-field link rolls off much faster than in effective area of the receive antenna, ATX is the effective area
the far field, which means that often times signals will not of the transmit antenna, d is the distance between each
antenna, and λ is the wavelength of the electromagnetic wave.
transmit far enough to cause harmful interference. In turn,
Note that the antenna effective area or aperture A is related to
short distance, high data-rate communication is possible using
the antenna gain G by
a near-field communication link.
Another novel use of the near field is in real-time location λ2G
A= , (2)
systems (RTLS). The Q-Track Corporation has pioneered a 4π
RTLS technology known as Near-Field Electromagnetic so Friis’s formula can also be written as
2
PRX  λ  G A A G
Manuscript received February 1, 2008; revised April 2, 2008. This =  G RX GTX = RX 2TX = RX 2TX . (3)
material is based upon work supported by the National Science Foundation PTX  4πd  4πd 4πd
under Award Number: 0646339. Any opinions, findings, and conclusions or This is a very powerful formula, but because it assumes a
recommendations expressed in this publication are those of the authors and do
not necessarily reflect the views of the National Science Foundation. plane wave front, it is only applicable in the far field. In his
A. J. Compston is a senior electrical engineering undergraduate student at paper, Friis warns that his formula “is correct to within a few
the Georgia Institute of Technology and is with the Q-Track Corporation, percent when
Huntsville, AL 35816 (e-mail: [email protected]).
J. D. Fluhler is an electrical engineering undergraduate student at the 2a 2
University of Alabama, Huntsville and is with the Q-Track Corporation, d≥ , (4)
Huntsville, AL 35816 (e-mail: [email protected]).
λ
H. G. Schantz is with the Q-Track Corporation, Huntsville, AL 35816
(e-mail: [email protected]).
1 2
Wi-Fi is a registered trademark of the Wi-Fi Alliance. NFER is a registered trademark of the Q-Track Corporation.
 2

where a is the largest linear dimension of either of the

antennas” [3]. Specifically, “[t]his criterion has a phase error
of one-sixteenth of a wavelength” [1]. Also, Friis’s formula is
only strictly valid in free space.

III. NEAR-FIELD PROPAGATION FORMULA

Whereas the far-field propagation formula developed by
Friis is the same regardless of the type of antenna used, in the Fig. 1. Infinitesimal dipole transmitting with fields at (r, θ, φ ).
near field one must distinguish between an electric antenna which means that the electric and magnetic fields can be
(like a dipole or whip) and its dual: a magnetic antenna (such rewritten as
as a loop). First the case of an electric transmit antenna is r 2 1  I∆lk 
2
r2 1
considered, followed by the case of a magnetic antenna. E =   PLlike , and H = ( I∆lk )2 PLunlike . (11)
 
6π  ε 0c  6π
A. Electric Transmit Antenna
Define the power densities of the electric and magnetic
Imagine an infinitesimal current element with length fields, respectively as WE and WH,
∆l << λ in free space. Suppose the current element has a
1 r 2 ( I∆lk ) 2
uniform current distribution I across its length. This structure, WE = ε 0 c E = PLlike , and (12)
first introduced by Heinrich Hertz [4], has since been analyzed 2 12πε 0 c
by a number of authors [5-7]. A theoretical idealization, it is 1 r 2 µ c( I∆lk ) 2
an accurate model for the electrically small antennas being WH = µ 0c H = 0 PLunlike . (13)
2 12π
considered.
where µ 0 is the magnetic permeability of free space.
The time harmonic electric field at the point (r, θ, ϕ)
Because the power transmitted by an ideal dipole through a
generated by the infinitesimal dipole is given by
closed sphere around the antenna as pictured in Fig. 1 is given
r I∆ lk 2   j  by
E=  + 1 − j  sin(θ )θˆ
4πε 0 c   kr ( kr ) 2 ( kr ) 3 
 1 r r r (I∆lk )2

 − jkr
, (5) PTX , E =
2 ∫∫ E × H ∗ ⋅ dS =
12πε 0 c
, (14)
 1 j  S
+ 2 −  cos(θ ) rˆ e
( kr ) 3  
2 the power densities (12) and (13) can be rewritten as
 ( kr ) 
WE =
(I∆lk )2 PL = PTX ,E PLlike , and (15)
where k = 2π / λ is the wave number, ε0 is the electric like
permittivity of free space, and c is the speed of light. The 12πε0c
magnetic field is µ0c(I∆lk) 2
r I∆lk 2   j WH = PLunlike = c2 µ0ε 0 PTX ,E PLunlike
  12π . (16)
H=   + 1  sin(θ )φˆ e − jkr . (6)

4π   kr (kr )  2   = PTX ,E PLunlike

The magnitudes squared of the fields are The power density at the receive antenna is equal to the
2 power it receives divided by its effective area, so
r 2  I∆lk 2   1 1 1  2
E =   − +  sin (θ ) P P
 4πε c    (kr)2 (kr)4 (kr)6  WE = RX , E = PTX , E PLlike ⇒ RX , E = Anf , RX PLlike . (17)
 0    Anf , RX PTX , E
, and (7)
 1 1  2  Similarly for a magnetic receive antenna
+ 4 +  cos (θ )
6
4  P P
 ( kr ) ( kr )   WH = RX ,H = PTX ,E PLunlike ⇒ RX ,H = Anf ,RX PLunlike . (18)
2 Anf ,RX PTX ,E
r 2  I∆lk   1
2
1  2
H =   +  sin (θ ) . (8)
 4π   ( kr ) 2 ( kr ) 4  It is important to realize that Anf,RX is not the effective area
   
in the traditional far-field sense because the far-field effective
For the sake of compactness and for reasons that will be
area assumes a plane wave front. This effective area must also
clearer momentarily, define the path-loss functions PLlike and
account for both the near- and far-field components (i.e., not
PLunlike as
just the θ- and ϕ- but also the r-component that is ignored in
3k 2   1 1 1  2 far-field analyses) and will therefore necessarily be different
PLlike (r,θ ,φ ) =  − +  sin (θ )
8π   (kr) (kr) (kr)6 
 2 4 in almost all cases.
, and (9)
 1 1  2  B. Magnetic Transmit Antenna
+ 4 + 6
 cos (θ )
4  Using the principle of duality, the electric and magnetic
 (kr) (kr)   fields of an infinitesimal current loop can be written as
3k 2  1 1  2 r I∆Sk 3   1  
PLunlike(r,θ ,φ ) =  +  sin (θ ) . (10)   − j  sin(θ )φˆ e − jkr , and
8π  (kr) (kr)4 
 2 E=
4πε 0 c   kr (kr) 2  
(19)

 3

Fig. 2. Transmit and receive antenna with

both at θ = 90°.
Fig. 3. Dependence on the near-field terms of the Fig. 4. Path-loss function for different θ.
propagation formula.
2
PRX PRX  k  A A
= =  ARX ATX = RX TX . (26)
r I∆Sk 3   1  PTX θ =90o ,like
PTX θ =90o ,unlike 
2πr  (λr )2
H=   − + j + 1  sin(θ )θˆ
 
4π   kr (kr ) 2 3
(kr )  Thus, the near-field propagation formula converges to
. (20) Friis’s formula in the far field in the horizontal plane.
 j 1  
+ 2 +  cos(θ )rˆ e − jkr B. Like Antenna Path-loss function and the Special Case of
 (kr )
2
(kr )3  
 θTX = 0°
where ∆S is the surface area of the loop. Using the same Fig. 4 shows a plot of PLlike as a function of kr for different
procedure outlined above for the electric transmitter, the values of θTX. The solid black line represents the special case
following can be proven for a magnetic transmit antenna of θTX = 90° that was previously demonstrated to follow the
PRX , H PRX , E Friis formula path-loss function for large r. As θTX goes to 0 in
= Anf , RX PLlike , and = Anf , RX PLunlike . (21)
PTX , H PTX , H the far field, the path-loss function gets smaller but still
Thus, the near-field propagation formula can be follows the θTX = 90° line. This is what one would expect from
summarized as below an ideal dipole with a donut power pattern in the far field.
Finally, at θTX = 0°, the path-loss function is smallest. The
PRX (r ,θ ,φ )  Anf ,RX PLlike (r ,θ ,φ ), like antennas θTX = 0° line does not follow the θTX = 90° line because in the
= , (22)
PTX  Anf ,RX PLunlike (r ,θ ,φ ), unlike antennas far field, that corresponds to the null of the antenna, where
where “like antennas” are taken to mean either two electric or there is ideally no power.
two magnetic antennas and “unlike antennas” are taken to However, the near field tells a different story. The path-loss
mean one electric and one magnetic antenna. function is largest for small r when θTX = 0°, which in turn
maximizes the PRX/PTX ratio. For θTX = 0°, (22) reduces to
IV. COMPARING THE NEAR-FIELD PROPAGATION FORMULA  3k 2  1 1 
WITH FRIIS’S FORMULA PRX  Anf , RX  + , like
=  2π  (kr )
4
(kr )6  . (27)
A. The Special Case of θ = 90° PTX θ = 0o 
0, unlike
For the special case where both antennas are at θ = 90°
Since for θTX = 0° the only component present in the field
(z = 0, the horizontal plane) relative to each other (as in Fig.
equations of an ideal dipole is the radial component, it is
2), see in (5) and (6) that the electric and magnetic fields have
convenient to define a near-field radial component pattern
no radial components. Therefore, the near-field effective areas
function of an ideal dipole as
of the antennas are equivalent to their far-field effective areas.
For an infinitesimal dipole, this is Fnf ,r (θ , φ ) = cos 2 (θ ) . (28)
3λ2 3π The near-field radial component directivity can also be
Anf ,inf = Anf ,TX = ATX = = 2, (23) calculated
8π 2k
which further reduces (24) to: Fnf , r (θ , φ )
Dnf , r (θ , φ ) = 4π 2π π
PRX k4  1 1 1 
= 2 ARX ATX 
PTX θ =90o ,like 4π 2
− 4
+
 (kr) (kr) (kr) 
 , and
6
(24) ∫∫ F
0 0
nf , r (θ , φ ) sin(θ ) dθdφ

. (29)
PRX k4  1 1  cos 2 (θ ) 2
= 2 ARX ATX  2
+ .
4
(25) = 4π 2π π
= 3 cos (θ )
PTX θ =90o ,unlike 4π  (kr) (kr) 
∫∫ cos (θ ) sin(θ )dθdφ
2

For large r, the 1/r4 and 1/r6 terms will go to zero much 0 0
faster than the 1/r2 terms, and they can be ignored for all This function is maximum when θRX = 0°, where the
practical purposes, as demonstrated in Fig. 3. This leaves directivity is 3. Because an ideal dipole was assumed, the
 4

-75

-80
Gain (dBi)

-85

-90 Measured
Expected
-95 Fig. 6. Boundary spheres around two arbitrary antennas.
600 1000800 1200 1400
Frequency (kHz) PRX
Fig. 5. Model LP-105 measured gain and expected gain. G = 2kr , (33)
PTX
directivity is equal to the gain. Therefore, the PRX/PTX ratio for
which is what Friis’s formula also predicts.
the θTX = θRX = 0° case can be further defined for the case of a
general transmitter as B. One Antenna with Unknown Gain
 Gnf , RX Gnf ,TX  1 1  If two like antennas are tested and the gain of one is known,
PRX   + , like
= 2  4
(kr )6  . (30) the power transmitted and received can be measured and also
PTX  (kr ) substituted into (32). This results in
θ = 0o 
0, unlike 4 PRX
Gunknown =
Note that if a non-ideal dipole is assumed, the directivity is  1 1 1 
PTX Gknown  − + 
(kr )6  .
not equal to the gain, but the power transmitted in (14) will 2 4
also be multiplied by the antenna efficiency. An analysis  (kr ) (kr ) (34)
accounting for the antenna efficiency will discover the same 4 PRX ( β r )8
=
result of (30). PTX Gknown ((kr )6 − (kr )4 + (kr )2 )
Again, in the far-field limit as r gets large, the (kr)6 will
V. MEASURING FAR-FIELD GAIN OF ELECTRICALLY SMALL
dominate the denominator and (34) will reduce to
ANTENNAS IN THE NEAR FIELD
PRX
In the case of θ = 90°, the effective areas in the near-field Gunknown = (2kr )2 , (35)
PTX Gknown
propagation formula are equivalent to the far-field effective
areas. Therefore, the far-field gain can be measured in the near which again is what Friis’s formula predicts.
field for this special orientation. Equations (24) and (25) can C. Three Like Antennas
also be written in terms of gain as
For three antennas of the same type (electric or magnetic),
PRX G G  1 1 1  (34) can be used in the same method that Friis’s formula is
= RX TX  − + . (31)
PTX θ =90o ,like 4  (kr) (kr) (kr)6 
2 4 used in the far field to measure antenna gain of three unknown
antennas [8]. Measure the power transmitted by each antenna
Assuming that the two antennas are of the same type, (31)
and the power received by each of the other two antennas.
can be used to measure gain. Three cases are discussed below.
This produces three simultaneous equations with three
A. Two Identical Antennas unknowns (the three gains) that can be solved.
If the antennas are the same, their gains should be the same.
VI. FUNDAMENTAL LIMIT ON ANTENNA GAIN AS A FUNCTION
Therefore, by solving (31) for GRX GTX = G 2
OF ANTENNA SIZE
4 PRX
G2 = To derive the fundamental limit of antenna gain, first the
 1 1 1  concept of boundary spheres must be explained. They were
PTX  − + 
 (kr )
2
(kr ) 4
(kr )6  . (32) originally introduced by Wheeler [9]. One of us [10] has
applied them to the question of the maximum gain a given
PRX
⇒ G = 2(kr ) 4 antenna can realize. However, that analysis considered only
PTX ((kr ) − (kr ) 4 + (kr )2 )
6
the θ = 90° case; since the maximum PRX/PTX ratio occurs for
We used this method to calculate the gain of two Empire θ = 0°, the limit presented in [10] requires revision.
(Singer) Model LP-105 loop antennas with known antenna Imagine placing a boundary sphere around an arbitrary
factors. The measured results compared to the expected results antenna with a radius R that is the smallest distance to
are shown in Fig. 5. Sources of error in the measurement could completely enclose the antenna as shown in Fig. 6. Next,
include RF coupling through the power (coupling was imagine a second identical antenna next to the first one also
noticeably apparent for electric antennas) and a non free space surrounded by a boundary sphere. Absent any other sources,
environment. the power received by one antenna cannot exceed the power
Note in the far-field limit as r gets large, the (kr)6 will transmitted by the other in order to comply with the law of
dominate the denominator, and (32) will reduce to conservation of energy. Mathematically,
 5

PRX -60
≤1 . (36)
PTX
-65
Furthermore, the minimum separation between the antennas
such that neither boundary sphere intersects is -70
r = 2R . (37)
Take the maximum PRX/PTX ratio as (30). Combining (30), -75

(36), and (37) results in:

-80

Gain (dBi)
PRX Gnf , RX Gnf ,TX  1 1 
=  +  ≤1
PTX θ = 0o 2  4
(2kR)6  -85
 (2kR)
2 2(2kR)6 . (38) -90
⇒ Gnf , RX Gnf ,TX ≤ =
 1 1  1 + (2kR)
2
  -95 Theoretical Limit
 (2kR)4 + (2kR)6  EMCO
  EMCO Expected
Empire Loop
-100 Empire Loop Expected
Because both antennas are assumed to be identical, their gains Terk AM Advantage
Multiple Frequencies
are also identical, so -105
Single Frequency

Gnf , RX Gnf ,TX = Gnf 2 0.0001 Rλ 0.001

Fig. 7. Measured antenna gains compared to the theoretical limit.

2(2kR)6 2 , (39)
⇒ Gnf ≤ = ( 4πRλ ) 3
analysis that can often yield interesting results. For example,
1 + (2kR)2 1 + (4πRλ )2
the gain limit derived above has profound implications on
where Rλ is taken to mean R in units of wavelength (so R/λ). electrically small antennas, which are becoming more and
Finally, note that the near-field gain accounts for all more common as near-field applications are increasing in
components of the fields, whereas the far-field gain only popularity. However, we suspect that we are only beginning to
accounts for two components. Therefore, one would expect scratch the surface on this fascinating topic.
that the near-field gain must be at least greater than or equal to
the far-field gain. Therefore, in terms of the far-field gain REFERENCES
6
2(2kR) 2 [1] C. Capps, “Near Field or Far Field,” EDN, August 16, 2001, pp. 95-102.
G ff ≤ Gnf ≤ 2
= (4πRλ )3 2
. (40) Available: http://www.edn.com/contents/images/150828.pdf. [Accessed
1 + (2kR) 1 + (4πRλ ) Feb. 1, 2008].
[2] H. Schantz, “A Real-Time Location System Using Near-Field
To check the limit suggested above, we measured the gains Electromagnetic Ranging,” IEEE Antenna and Propagation Society
of a number of different magnetic antennas and compared International Symposium, 2007. Available:
them against their theoretical limit for their size (see Fig. 7). http://www.q-track.com/Technology.aspx?ID=25. [Accessed Feb. 1,
2008]. See http://www.q-track.com/ for more on NFER.
For the antennas with data taken at multiple frequencies, we [3] H. Friis, “A Note on a Simple Transmission Formula,” Proc. IRE, 34,
measured the power received by an Empire LP-105 loop 1946, pp. 254-256.
antenna with a known gain and changed the transmit antenna. [4] H. Hertz, Electric Waves, English ed. New York: Dover Publications,
We measured the gain of EMCO Model 6509 antenna using 1962, pp. 141-151.
[5] C. A. Balanis, Antenna Theory: Analysis and Design, 2nd ed. United
this method and compared it to its expected values based on its States: John Wiley & Sons, 1997, pp. 133-143.
antenna factor. For the antennas with data at a single [6] J. D. Kraus, Antennas, 2nd ed. United States: McGraw-Hill, 1988, pp.
frequency, we first measured the power transmitted out to a 201-213.
[7] W. L. Stutzman and G. A. Thiele, Antenna Theory and Design, 2nd ed.
known distance by the EMCO Model 6509 loop antenna with United States: John Wiley & Sons, 1998, pp. 20-24.
a known gain. We then transmitted the exact same power for [8] IEEE Standard Test Procedures for Antennas, IEEE Std 149-1979, New
each antenna at the same distance, and the relative difference York: IEEE, p. 96.
[9] H. A. Wheeler, “Fundamental Limitations of Small Antennas,” Proc.
of the power received by this new antenna was added to the IRE, 35, 1947, pp. 1479-1484.
gain of the EMCO. [10] H. G. Schantz, “A Near-Field Propagation Law & A Novel Fundamental
Some antennas seem to perform better than the theoretical Limit to Antenna Gain Versus Size,” IEEE Antenna and Propagation
Society International Symposium, 2005, Vol. 3A, pp. 237-240.
limit would predict. However, sources of error in the Available: http://www.q-track.com/Technology.aspx?ID=25. [Accessed
measurements, including RF coupling and non-free-space Feb. 1, 2008].
conditions, can account for this discrepancy. Recall that the
Empire LP-105 antenna’s measured gain was as high as 5 dB
off of the expected value. Accounting for a measurement error
of 5 dB, it is entirely plausible that all of the data fall below
the expected limit.

VII. CONCLUSION
The near fields are an often overlooked aspect of antenna