RTLS in Multipath and AWGN Environments Location Error Analysis of an Active RFID-Based

(1)

In this paper, we analyze the location accuracy of real- time locating systems (RTLSs) in multipath environments in which the RTLSs comply with the ISO/IEC 24730-2 international standard. To analyze the location error of RTLS in multipath environments, we consider a direct path and indirect path, in which time and phase are delayed, and also white Gaussian noise is added. The location error depends strongly on both the noise level and phase difference under a low signal-to-noise ratio (SNR) regime, but only on the noise level under a high SNR regime. The phase difference effect can be minimized by matching it to the time delay difference at a ratio of 180 degrees per 1 chip time delay (Tc). At a relatively high SNR of 10 dB, a location error of less than 3 m is expected at any phase and time delay value of an indirect signal. At a low SNR regime, the location error range increases to 8.1 m at a 0.5 Tc, and to 7.3 m at a 1.5 Tc. However, if the correlation energy is accumulated for an 8-bit period, the location error can be reduced to 3.9 m and 2.5 m, respectively.

Keywords: RTLS, DSSS modem, location error, RFID reader, RFID tag.

Manuscript received Dec. 28, 2010; revised June 23, 2011; accepted July 1, 2011.

This work was supported by the IT R&D program of MKE/KEIT [2008-S-040-02, Real Time Location Service technology Development].

Seungil Myong (phone: +82 42 860 5360, email: msi@etri.re.kr) is with the Internet Research Laboratory, ETRI, Daejeon, Rep. of Korea.

Sanghyun Mo (email: shmo@etri.re.kr), Hoesung Yang (email: hsyang@etri.re.kr), and Heyungsub Lee (email: leehs@etri.re.kr) are with the IT Convergence Technology Research Laboratory, ETRI, Daejeon, Rep. of Korea.

Jongsub Cha (email: s1009@rndip.re.kr) is with the IP Strategy Activation Team, R&D Patent Center, Seoul, Rep of Korea.

Dongsun Seo (email: sdsphoto@mju.ac.kr) is with the Department of Electronics Engineering, Myongji University, Youngin, Rep of Korea.

doi:10.4218/etrij.11.1610.0023

I. Introduction

With the rapid growth of wireless communications technologies, location and tracking services for assets or people have been provided through the application of various wireless technologies such as direct-sequence spread spectrum (DSSS) and ultra-wideband (UWB). Among these, the active radio frequency identification (RFID)-based real-time locating system (RTLS) has received great attention in recent years, where the RTLS is proven to be an automated wireless system with real-time locating and tracking capabilities for certain assets or people. An RTLS can raise work productivity and make life easier. They are expected to advance further, along with recent services such as augmented reality and social networks. Various commercial applications using RFID technology that operate in frequencies of 135 kHz to 10 GHz have appeared on the market. The related ISO/IEC international standardization is currently in progress based on various technologies. RTLSs can be categorized into different systems based on their location accuracy ranges of within 200 m, 10 m, 3 m, and 10 cm [1], [2].

An IDTechEx report forecasts that the global RTLS market will grow from $203 million in 2009 to $2.73 billion in 2018 [3]. It is very challenging to find ways to shorten the time to market and stay ahead of the competition. There has been significant progress on location-based services (LBSs) and global positioning systems (GPSs), including RTLSs in real environments [4]-[6]. While the interest in RTLSs is increasing continuously, there is a lack of research and analysis. In [7], [8], active RFID-based RTLSs for outdoor localization were introduced. Due to their large size, high cost, and high power consumption, outdoor localization methods based on GPSs and cellular techniques are not suitable, but an active RFID system

Location Error Analysis of an Active RFID-Based RTLS in Multipath and AWGN Environments

Seungil Myong, Sanghyun Mo, Hoesung Yang, Jongsub Cha, Heyungsub Lee, and Dongsun Seo

(2)

is cost-effective and easily affixed to an object and has low power consumption and small size. Localization or positioning is the process of determining the physical position of an object.

For robust and accurate localization, the propagation delay of the radio signal arriving from the direct line-of-sight (LOS) is accurately estimated. However, in a real environment, the radio signal may additionally propagate over non-line-of-sight paths so we need to analyze multipath effects on the location accuracy [9], [10].

The ISO/IEC 24730-2 standard requires location data provided by an RTLS shall be within a 3 m radius of the actual location of the transmitter with at least 300 m open-field separation between the transmitter and receiver. However, location accuracy is not guaranteed in most real environments due to additive white Gaussian noise (AWGN) and multipath interference. In this sense, it is very important to analyze the effects of AWGN and multipath components on the location accuracy. Here, we study the object location errors of an active RFID-based RTLS under AWGN and multipath environments where direct and indirect paths exist. Through Monte-Carlo simulation, we examine the effects of signal-to-noise ratio (SNR), as well as the magnitude, time delay, and phase differences (PDs) between direct and indirect signal components. Our results will provide a guideline for the development of a commercial system applicable to practical environments. In section II, we introduce the signal model in a real environment where the physical parameters of ISO/IEC 24730-2 are considered. In section III, we examine the expected position error as a function of time delay and PDs between direct and indirect signals as well as AWGN. Finally, in section IV, we offer some concluding remarks.

II. ISO/IEC 24730-2 Physical Layer and Signal Modeling

In this section, we review the frame structure of the ISO/IEC 24730-2 standard and its physical layer parameters and describe a received signal model in a multipath environment.

1. Modulation and Demodulation Specifications of ISO/IEC 24730-2 Standard

Modem specifications are summarized in Table 1. ISO/IEC 24730-2 employs a DSSS modulation scheme using differential binary phase shift keying (DBPSK) data encoding and BPSK spreading. Data bits are spread using PN sequences of period 511 chips. The data bitrate is 59.7 kbps, and the chip rate is 30.521874 Mcps. The maximum carrier frequency offset is ±25 ppm. This corresponds to a ±61.04 kHz frequency offset from the center frequency at 2441.75 MHz [2].

Table 1. Baseband modem specifications.

Parameter Specification

Data modulation DBPSK

Data rate 59.7 kbps

Spreading DSSS

Spreading sequence 511

Chip rate 30.521875 Mcps ±25 ppm

Spreading code 0x1 CB

Channel bandwidth 61 MHz

Packet length 56,72,88,152

Sub-blink interval 125 ms ±16 ms (maximum)

Center frequency 2441.750 MHz

Frequency offset < ±25 ppm Maximum frequency drift < ±2 ppm

Phase accuracy < 0.5 radian

Phase noise < 15 degrees

BER < 10^–5

Chip rate is an important parameter for determining the time resolution of the locating systems shown in Table 1. Well- known spread spectrum (SS) technology indicates that an SS signal returns to the original signal by de-spreading with a matched sequence at the receiver to show the general maximum energy per bit. Therefore, the peak energy per bit provides time information of the arrived signal. This indicates that the spectral width of an SS signal is inversely proportional to the location accuracy; a wider spectrum improves location accuracy. The standard chip rate of 30.521875 Mcps shown in Table 1 gives 32.76 ns for the minimum time resolution unit.

This time resolution is equivalent to 9.6 m of location accuracy.

When there is only a single chip time difference (TD) (32.76 ns) between the two received tag signals, the difference in distance between the two readers is determined as 9.6 m. For a higher resolution, a locating system should have a higher chip rate, resulting in a wider SS (such as UWB).

Another way to improve the location accuracy is to increase the sampling rate at the receiver. The RTLS shown in Table 1 has about a 60 MHz spectral width, corresponding to twice the chip rate. This can reduce object position errors to half the number occurring for 9.6 m. A signal is generally sampled at 4 times or 8 times the chip rate at the receiver. This improves the resolutions to 2.4 m and 1.2 m, respectively. However, this method, which is based on a direct path signal, has an inherent limit in improvement due to multiple path interference signals reflected from the surroundings and the noise level of a practical location system. Signal modeling of a received signal,

(3)

which includes a reflective path and noise, is essential for investigating this limitation and examining the induced position errors.

2. Signal Model

First, the signal modeling of a tag output is necessary to interpret a received signal at a reader. Figure 1 shows a block diagram for the transmitted signal from a tag. In Fig. 1, dn

indicates a data symbol of differential phase shift keying (DPSK) encoding, P is the power of the transmitted signal, cn is the PN code for spreading, H(f) is the transfer function of pulse-shaping filter, fc is the carrier frequency, and Δf is the offset of the carrier frequency. The transmitted signal s(t) of the RTLS system is given by

( ) 2 _m _{m N}/ ( _c) cos(2 ( _c ) )

m

s t P ^∞ c d_⎢_⎣ _⎥_⎦ h t mT π f f t

=−∞

=

∑

⋅ − + Δ ^{, (1)}

where N is the length of the PN code, m is m-th order of chips, h(t) is the impulse response of the pulse-shaping filter, cm is the PN code, dm is the DPSK symbol, and ⌊m/N⌋ is a minimum integer greater than m/N.

The received signal at a reader with non-selective frequency fading is given by

BP( ) ( ) 2 / ( )cos(2 ( ))

( )cos(2 ) ( ) sin(2 ), (2)

m m N c d c

m

I c Q c

r t t P c d h t mT f t

n t f t n t f t

α τ π ϕ

π π

∞

⎢ ⎥

⎣ ⎦

=−∞

= ⋅ − − +

+ − ⋅

∑

where α(t) is the envelop of the signal, φ(t) is the signal phase, τd is the propagation delay of the signal, and nI(t) and nQ(t) represent in-phase and quadrature-phase (I and Q) narrowband Gaussian noise processes.

Fig. 1. Block diagram for RTLS transmitted signal.

DPSK symbols d0, d1, d2,…

cn

H*(f)

2 cos(2 (P π f_c+ Δf t) ) s(t)

Fig. 2. Schematic diagram for de-spreading process of an RTLS signal.

rBP(t) 2 cos(2πf t_c) Z

2 sin(2πf t_c)

−

H^*(f) uI(t) t=nT^c yI,n

H^*(f) uQ(t) t=nT^c cn

yQ,n 1 N n=∑

1 N n=∑ ^Y^I

YQ

(·)²

Figure 2 shows a de-spreading process of the received signal rBP(t) at a reader. In the figure, H^*(t) is the transfer function of a matched filter with * denoting a conjugate operation.

The I and Q outputs of the matched filter, uI(t) and uQ(t), are given by

I BP LPF

'I /

( ) ( ) 2 cos(2 ) ( )

( ) cos(2 ( )) ( ) 1 ( )

2

c

m m N c d

m

u t r t f t h t

t P f t c d R t mT n t

π

α π ϕ ^∞ _⎢ _⎥ τ

⎣ ⎦

=−∞

= ⋅ ⊗ −

= Δ +

∑

− − +

(3) and

Q BP LPF

'Q /

( ) ( ) 2 sin(2 ) ( )

( ) sin(2 ( )) ( ) 1 ( ),

2

c

m m N c d

m

u t r t f t h t

t P f t c d R t mT n t

π

α π ϕ ^∞ _⎢_⎣ _⎥_⎦ τ

=−∞

= − ⋅ ⊗ −

= Δ +

∑

− − +

(4) where rBP(t) is demodulated by a local carrier frequency at a reader and is filtered by h(–t). The transfer function of the matched filter is denoted by h(–t). The convolution of the impulse response of the pulse-shaping filter h(t) and the matched filter h(–t) is R(t). Also, n'(t) is the convolution of the noise response and the matched filter asn t_I′( )=n t_I( )⊗ −h t( ),

Q( ) Q( ) ( )

n t′ =n t ⊗ − . h t

In (3) and (4), if there are no timing errors, α(t) and φ(t) become constant. We then get the following accumulated correlator output of the I and Q paths:

I I

1

I 1

( )

( ) cos(2 ) ,

N

c n n

N

c n

Y u nT c

PR d n fT n

α τ π ϕ

=

= Δ + +

∑

(5)

Q Q

1

Q 1

( )

( ) sin(2 ) .

N

c n n

N

c n

Y u nT c

PR d n fT n

α τ π ϕ

=

= Δ + +

∑

(6)

We assume that the energy is accumulated during N chips, that envelope α is normalized to 1, and that there are no frequency offset errors, which means that an automatic frequency control of a reader estimates the exact frequency of the received tag signal. The accumulated correlator output of the I and Q paths are then simplified by

I _c ( ) cos( ) I,

Y = N E Rτ ϕ +n (7)

Q _c ( )sin( ) Q,

Y = N E Rτ ϕ +n (8) where Ec is the energy per chip, N is the number of accumulated PN chips, and nI and nQ are the noises of the I and Q paths, respectively.

Now, consider the effects of an indirect path, where the

(4)

signal experiences an additional time delay of τd and an envelope suppression of β. The I and Q correlator outputs are then modified as follows:

I ( ) cos_c 1 _c ( _d) cos 2 I,

Y = N E Rτ ϕ β+ N E Rτ τ+ ϕ +n (9)

Q ( )sin_c 1 _c ( _d)sin 2 Q,

Y = N E Rτ ϕ β+ N E Rτ τ+ ϕ +n (10) where τ is the time delay of the direct path, R(t) is the overall response of the pulse shaping and matched filters, that is,R t( )=h t( ) * ( )h t− , φ₁ andφ are the phases of the direct ₂ and indirect paths, respectively, and τd is the difference in time delay between the two paths.

Finally, we obtain the following correlation energy Z:

2 2

I Q

2 2 2 2 2

2 1 2

2 2 2

1 2 I Q

2 2 2 2 2

2 2 2

I Q

( )

( ) ( )

2 ( ) ( ) cos cos 2 ( ) ( )sin sin

( ) ( )

2 ( ) ( ) cos ,

c c d

c d

c c d

c d

Z Y Y

N E R N E R

N E R R

N E R R n n

N E R N E R

N E R R n n

τ

τ β τ τ

β τ τ τ ϕ ϕ

τ β τ τ

β τ τ τ ϕ

= +

= + +

+ +

+ + + +

= + +

+ + Δ + +

(11)

where ∆φ is the PD (φ φ₂− or₁ φ φ₁− ). Equation (11) proves ₂ that correlation energy Z is affected mainly by the suppression ratio of the indirect signal, time delay, and PDs between two signals. To examine the effects in detail, we perform computer simulations of the correlation energy as a function of suppression, delay, and phase of the indirect path signal.

III. Simulation Results

Simulations were conducted for an analysis of the multipath effects on location accuracy. To examine the timing estimation error from the autocorrelation function (ACF) in a multipath environment, we assume a simple propagation model consisting of direct and indirect paths. Using (11), we examine the timing variation of the maximum ACF curve value, which represents the de-spread correlation energy of direct and indirect path signals under an AWGN environment.

Throughout this section, the following conditions are used for simulation. Amplitudes of the direct and indirect path signals are normalized to 1 and 1/ 2 , respectively. Thus, the signal strength of the indirect path is 3 dB lower than that of the corresponding direct path. The TD between two signals is varied from 0.5 chip time delay (Tc) to 1.5 Tc, where a chip interval of 1 Tc equals 32.76 ns. As a transmitted waveform, we use a square-root-raised-cosine pulse with a roll-off factor of 0.9. We assume that the receiver filter is matched to the transmitted waveform. The relative PD between the two paths

is varied from 0 degrees to 180 degrees with 45-degree steps.

The received signal is oversampled at 8 times the chip rate.

1. Comparison of ACF Curves Depending on TD of Two Paths (0.5 Tc to 1.5 Tc)

To analyze the impact of the TD between two signals, the SNR is fixed at 10 dB and the TD and PD are varied from 0.5 Tc to 1.5 Tc and from 0 degrees to 180 degrees, respectively.

Figures 3 to 5 show the autocorrelation functions at several TDs and PDs. For individual conditions, 500 simulations were performed and 100 cumulative results of the ACF given by (11) were plotted. When there is no indirect path signal, the ACF peaks are located at 0 Tc (ideal arrival time of the direct path signal), even if slight jitter is observed due to AWGN.

While the direct path signal is interfered with by an indirect path signal, the ACF peaks deviate from the ideal position of 0 Tc. Timing estimation based on those ACF peak positions may induce large location errors.

From the simulation results in Figs. 3 to 5, we can examine the ACF peak position dependence on TD and PD. If the TD is equal to 0.5 Tc, there exists only one ACF peak regardless of PD. This is understandable since the two signals are not resolvable for small TDs. However, the ACF peak location fluctuates significantly depending on the PD, which decides on constructive or destructive interaction between two signals.

The effects of PD on the ACF peak location will be discussed later in detail. At a TD of 1.0 Tc, the fluctuation of the ACF peak location becomes more severe at a lower value of PD, and is minimized at a PD of 180 degrees, as we can see in Fig. 4.

To examine the detailed trend, we need to check the distribution of the ACF peak, as we will discuss later. The variation of ACF peak position causes an error in the estimated arrival time of the direct path signal. As TD increases, the overlap between two signals is reduced further, resulting in negligible interaction between them. At a TD of 1.5 Tc, as shown in Fig. 5, there exist two ACF peaks regardless of PD, indicating that two signals become non-overlapped and distinguished. When the TD is greater than 1.5 Tc, the two signals are completely distinguished, and there is no phase interaction (that is, no effect from a change in PD), minimizing the effect of an indirect signal.

Next, the probability density function (PDF) of the ACF peak position is examined and illustrated in Figs. 6 through 8 at several values of TD and PD between the two paths, corresponding to Figs. 3 through 5. The PDF represents the distribution of arrival timing error for the direct path signal due to interference from the indirect path signal. SNR is fixed at 10 dB, and the PD increases from 0 degrees to 180 degrees at 45-degree intervals.

(5)

Fig. 3. Autocorrelation functions in multipath and AWGN environments, where TD = 0.5 Tc and SNR = 10 dB.

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 14,000

12,000 10,000 8,000 6,000

4,000

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 2.0

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

×10⁴

(a) PD = 0 degrees, accumulation bit = 1 bit

(b) PD = 90 degrees, accumulation bit = 1 bit 2,000

0

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 9,000

8,000 7,000 6,000 5,000 4,000

(c) PD = 180 degrees, accumulation bit = 1 bit 2,000

0 3,000

1,000

When two signals are overlapped (that is, TD is small), the phase relationship between the two signals affects the PDF significantly, as shown in Fig. 6. Note that, as PD approaches 90 degrees, the mean value of PDF nears 0 Tc. For a PD greater than 90 degrees, the trend is reversed and the PDF mean deviates from 0 Tc. This means that the position error can be minimized when PD is 90 degrees. Again, this is understandable since a TD of 0.5 Tc corresponds to a PD of 90 degrees. Under a matched condition, the effect of an indirect

Fig. 4. Autocorrelation functions in multipath and AWGN environments, where TD = 1.0 Tc and SNR = 10 dB.

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 14,000

12,000 10,000

8,000 6,000

4,000

–1.50 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 (a) PD = 0 degrees, accumulation bit = 1 bit

0

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 (c) PD = 180 degrees, accumulation bit = 1 bit 0

12,000

10,000

8,000

6,000

4,000

2,000

14,000

12,000

10,000 8,000 6,000

4,000 2,000

signal on the correlation function becomes symmetric near 0 Tc, resulting in a zero mean value of PDF. This indicates that, as PD increases from 0.5 Tc to 1.0 Tc, the overlap (that is, phase interaction) between the two signals is reduced, but the matched value of the PD moves from 90 degrees to 180 degrees to acquire the PDF mean at 0 Tc. Figure 7 shows the PDF of an arrival timing error at a TD of 1.0 Tc for various values of PD. Compared with Fig. 6, the amount of phase interaction, or accordingly the dependence of PD, is reduced

(6)

Fig. 5. Autocorrelation functions in multipath and AWGN environments, where TD = 1.5 Tc and SNR =10 dB.

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 14,000

12,000

10,000 8,000 6,000

4,000

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.00 2.5 (a) PD = 0 degrees, accumulation bit = 1 bit

0

–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 (c) PD = 180 degrees, accumulation bit = 1 bit 0

12,000

10,000

8,000

6,000

4,000

2,000

14,000

12,000

10,000 8,000 6,000

4,000 2000

due to the larger TD value. As expected, the mean value of PDF approaches 0 Tc when PD is 180 degrees. As PD decreases from the matched value (180 degrees), the position error increases; a smaller PD gives a larger error. For a TD greater than 1.0 Tc, the phase interaction becomes negligible, but the ACF given by (11) has double peaks, corresponding to both the direct and indirect signal positions. Depending on the noise contribution, a sub-peak from an indirect signal can be misinterpreted as the main peak. This can be clearly seen in

Fig. 6. PDF of timing error at TD = 0.5 Tc and SNR = 10 dB.

–15 –10 –5 0 5 10 15 20

Time error (1/8 Tc) 1.0

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

PDF

PD = 0 degrees PD = 45 degrees PD = 90 degrees PD = 135 degrees PD = 180 degrees Accumulation bit = 1 bit

–15 –10 –5 0 5 10 15 20

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

PDF

–15 –10 –5 0 5 10 15 20

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

PDF

Fig. 7, showing a PDF at a TD of 1.5 Tc. A negligible phase interaction induces a similar PDF regardless of the PD values.

(7)

Clear PDF peaks at 0 Tc ensure an accurate object location, but there is a significant probability that the ACF peak can be detected near the arrival time of the indirect path signal, as shown by the side peaks in Fig. 7. This means that a given SNR of 10 dB is not high enough to distinguish a direct path signal from an indirect path signal using a correlation method.

2. Comparison of Expected Values of Location Error vs.

SNR and PD

In this subsection, we analyze the expected value of location errors depending on the SNR and PD. The TD between two signals increases from 0.5 Tc to 1.5 Tc with 0.5 Tc steps.

Figures 9 through 11 show the results.

For a low SNR of 4 dB to 10 dB, the expected value of a location error is 3.3 m to 8.1 m. When the SNR is 10 dB to 18 dB, the expected value of a location error is 0.2 m to 2.8 m.

As discussed previously, when the TD is larger than 1.0 Tc, the expected value of a location error does not depend on the PD but strongly depends on the noise level, that is, SNR. When the TD is smaller than 1.0 Tc, the location error depends highly on the relative PD and SNR.

3. Comparison of Expected Value of Location Errors Based on Accumulation Bits

To reduce the amount of location errors caused by noise, the number of accumulation bits for obtaining correlation energy should be increased. We can obtain significant correlation energy even at a low SNR by increasing the number of accumulation bits for correlation; a smaller expected location error value is then expected. Simulations are conducted to analyze the expected value of the location error as a function of the accumulation bit. We chose TDs of 0.5 Tc and 1.5 Tc, and 4-bit accumulation and 8-bit accumulation. Under these conditions, we investigate changes in the expected values of location errors depending on the SNR and PD.

The simulation results in Figs. 12 to 15 show that when a TD is 0.5 Tc, SNR is 4 dB, and PD is 180 degrees, the expected value of a location error improves from 8.1 m (1-bit accumulation) to 3.9 m (8-bit accumulation). When a PD is 1.5 Tc and the other values are the same, the expected value of a location error improves from 7.3 m (1-bit accumulation) to 2.5 m (8-bit accumulation). If a TD is 0.5 Tc and the SNR is greater than 10 dB, there is no noticeable difference in the expected value of the location errors between 8-bit and 1-bit accumulations. At 8-bit accumulation, the expected value of a location error is less than 3 m even at a 4 dB SNR, except for a PD near 180 degrees. With a path difference of 1.5 Tc and an SNR greater than 8 dB, the expected value of a location error

Fig. 9. Expected value of location errors vs. SNR and PD, when TD = 0.5 Tc and accumulation bit = 1 bit.

18 0 14 16 10 12

6 8 4 10

8 6 4 2 0

SNR (dB) PD (degree)

Expectation value of location error (m)

50 100

150

Fig. 10. Expected value of location errors vs. SNR and PD when TD = 1.0 Tc and accumulation bit = 1 bit.

18 0 14 16

10 12 6 8

4 10 8 6 4 2 0

50 100

150

Fig. 11. Expected value of location errors vs. SNR and PD when TD = 1.5 Tc and accumulation bit = 1 bit.

18 0 14 16 10 12

6 8 4 10 8 6 4 2 0

50 100

150

(8)

Fig. 12. Expected value of location errors vs. SNR and PD when TD = 0.5 Tc and accumulation bit = 4 bits.

18 0 14 16

10 12 6 8

4 10

8 6 4 2 0

50 100

150

0 16 18

12 14 8 10 4 6

10 8 6 4 2 0

50 100

150

0 16 18

12 14 8 10 4 6

10 8 6 4 2 0

50 100

150

0 16 18 12 14 8 10 4 6

10 8 6 4 2 0

50 100

150

can be less than 1 m regardless of the PD.

IV. Conclusion

In this paper, we addressed the effects of indirect path interference signals and AWGN on object location accuracy, where we considered an RTLS reader complying with the ISO/IEC 24730-2 physical layer. To analyze the location errors occurring in multipath environments, we considered direct and indirect path components with time and phase delay differences and AWGN. The power of the indirect component was assumed to be half that of the direct component. Monte- Carlo simulation showed that the location errors depend strongly on both the AWGN level and phase difference when the time delay difference is small. For a large time difference, location errors are more sensitive to the AWGN level than a phase difference. The effects of the phase difference on position error can be minimized by matching the phase and time delay differences at a matching ratio of 180 degrees per 1 Tc. At an SNR between 10 dB to 18 dB, the expectation value of a location error was 0.2 m to 2.8 m, depending on the phase and time differences between the direct and indirect components. For a low SNR (4 dB to 10 dB), the expected value of a location error increased to 3.3 m to 8.1 m. The large location error at a low SNR can be improved by increasing the number of accumulation bits. When the TD is 0.5 Tc, SNR is 4 dB, and PD is 180 degrees, the expected value of a location error improved from 8.1 m to 3.9 m through an 8-bit accumulation period. When the PD is 1.5 Tc, the expected value of a location error improved from 7.3 m to 2.5 m. If the SNR is greater than 10 dB, there is no noticeable difference in the expected value of location errors between 8-bit and 1-bit accumulations.

(9)

References

[1] J.S. Cha et al., “Technology Trend of Active RFID-Based Real-Time Locating Systems,” Electron. Telecommun. Trends, vol. 24, no. 5.

Oct. 2009, pp. 87-97 (in Korean).

[2] ISO/IEC 24730-2, Information Technology – Real Time Locating Systems (RTLS) – Part 2: 2, 4 GHz Air Interface Protocol, 2006.

[3] IDTechEx, “Real Time Locating System (RTLS) 2008-2018,”

Apr. 2008.

[4] L. Cong and W. Zhuang, “Nonlinear of Sight Error Migration in Mobile Location,” IEEE Trans Wireless Commun, vol. 4, no. 2, Mar. 2005, pp. 560-573.

[5] L.J. Gartin, “The Shaping Correlator,” Novel Multipath Mitigation Technique Applicable to GALILEO BOC(1,1) Modulation Waveforms in High Volume Markets,” GNSS, 2005.

[6] A.J. van Dierendonck, P. Fenton, and T. Ford, “Theory and Performance of Narrow Correlator Spacing in a GPS Receiver,” J.

Inst. Navigation, vol. 39, no. 3, Fall 1992.

[7] S. Behera and C. Maity, “Active RFID Tag in Real Time Location System,” IEEE SSD, July 2008, pp. 1-7.

[8] X. Huang, R. Janaswamy, and A. Ganz, “Scout: Outdoor Localization Using Active RFID Technology,” BROADNETS, Oct. 2006, pp. 1-10.

[9] S.-Y. Lee and J.-T. Park, “NLOS Error Mitigation in a Location Estimation of Object based on RTLS Using Kalman Filter,”

SICE-ICASE, 2006, pp. 2942-2949.

[10] M. Marx, R. Kokozinski, and H.C. Muller, “Time Synchronization for Real Time Localizatino Systems with Multi Path Mitigation,” IEEE IMWS, Sept. 2009, pp. 1-4.

Seungil Myong received the BS and MS in electronics engineering from Myongji University, Rep. of Korea, in 1997 and 1999, respectively. He received the PhD in electrical engineering from Myongji University in 2010.

Since 2000, he has been a senior member of the research staff at ETRI, Rep. of Korea. His current research interests include optical CDMA, optical communications, RFID systems, RTLS, and modem designs for communication systems.

Sanghyun Mo received the BS and MS in electronic and electrical engineering from Pohang University of Science and Technology (POSTECH), Rep. of Korea, in 2006 and 2008, respectively. Since February 2008, he has been a researcher at ETRI, Rep. of Korea. His research interests include RFID systems, mobile communication systems, and modem designs for communication systems.

Hoesung Yang received the BS in electronic engineering from Gwangwoon University, Seoul, Rep. of Korea, in 2003, the MS in radio engineering from Seoul National University in 2005. In 2005, he joined ETRI, Daejeon, Rep.

of Korea, as a senior member of the research staff in RFID Basic Technology Team. His research interests include structure of RF systems and wireless communications.

Jongsub Cha received his BS in information and telecommunication engineering from Korea Aerospace University, Rep. of Korea, in 2002, and the MS and PhD in electrical engineering from Korea Advanced Institute of Science and Technology, Rep. of Korea, in 2004 and 2007, respectively. His research interests include MIMO, STC, and RFID.

Heyungsub Lee received the BS, MS, and PhD in electronics engineering from Chungnam National University, Rep. of Korea, in 1985, 1994, and 2002, respectively. Since 1990, he has been with ETRI. He is currently working as the team leader of the RFID Basic Technology Research Team. His research interests include RFID systems, premise network systems, telecommunication protocol, and digital signal processing

Dongsun Seo received the BS and MS in electronic engineering from Yonsei University, Korea, in 1980 and 1985, respectively, and the PhD in electrical engineering (optoelectronics) from the University of New Mexico in 1989.

From 1980 to 1986, he was with the Agency for Defense Development as a research engineer.

From 1986 to 1990, he was a research assistant and later a member of the research staff at the Center for High Technology Materials, the University of New Mexico. In 1990, he joined the faculty of Myongji University, Rep. of Korea, where he is currently a professor in the Department of Electronics Engineering. During 1994 and 1995, he was a visiting research fellow at the Photonics Research Laboratory, University of Melbourne, Australia. From 2002 to 2004 and 2010, he was with Purdue University as a visiting research professor in the School of Electrical and Computer Engineering. His current research interests are in the areas of ultrafast optics, optical CDMA, microwave photonics, and sensors.