Sensitivity Enhancement in WiBro Geo-location System by alendar


More Info
									                                        International Global Navigation Satellite Systems Society
                                                                         IGNSS Symposium 2009

                                                          Holiday Inn Surfers Paradise, Qld, Australia
                                                                               1 – 3 December, 2009

                       Sensitivity Enhancement
                    in WiBro Geo-location System
                                        Ji-won Park (1)
                   Department of Information and Communication Engineering
                              Chungnam National University, Korea
                 Tel: +82-42-821-7607, Fax: +82-42-824-6807,

                                     Seung-Hun Song (2)
                   Department of Information and Communication Engineering
                             Chungnam National University, Korea
             Tel: +82-42-821-7607, Fax: +82-42-824-6807,

                                     Tae-Kyung Sung (3)
                          Division of Electric and Computer Engineering
                              Chungnam National University, Korea
                 Tel: +82-42-821-5660, Fax: +82-42-824-6807,


           WiBro, a Korean mobile WiMAX, is a wireless broadband internet service
           based on IEEE (Institute of Electrical and Electronics Engineering) 802.16e.
           In WiBro system, preamble symbols are broadcasted by RAS (Radio Access
           Station) at the beginning of each frame. Using the correlator, TDoA (Time
           Difference of Arrival) measurements can be obtained from the preamble. For
           two-dimensional positioning, more than two RASs need to be detected.
           Since WiBro is a cell-based system, neighboring RASs are hard to detect. In
           order to increase the sensitivity of the receiver to detect the preamble signal
           from the neighboring RAS, long integration technique is often used.
           However, due to residual frequency in the carrier, length of the integration is
           limited. This paper presents a residual frequency estimator used in hybrid
           integration in WiBro network. WiBro channel model is taken into account
           based on ITU-R m.1225. By computer simulation, it is revealed that the
           residual frequency estimation error is less than 2 Hz in multi-path
           environment with regardless of user position.

           KEYWORDS: WiBro, hybrid integration, residual Doppler estimation


LBS (Location Based Service) are becoming essential to our life as the number of mobile
terminals is growing up. The LBS provides specific information according the location of the
subscriber. The quality of LBS service becomes better as the positioning accuracy is
improved. Several positioning methods are used in LBS to estimate the user position.
Examples include GPS, Assisted GPS, Cell-ID, Geo-location using cellular network, and
dead-reckoning systems.

WiBro is an internet service that provides high-speed data communication in mobile. It is
capable of providing seamless and stable data communication when moving within the
velocity of 60 Km/h. WiBro employs OFDMA (Orthogonal Frequency Division Multiple
Access) TDD (Time Division Duplex) technology, and symbol is used as a basic unit in
communication. Group of down-link and up-link symbols constitute a frame. A preamble
symbol is assigned to the first symbol of each frame, and orthogonal preambles are assigned
to RASs to distinguish each other.

Since all the RASs are synchronized by GPS clock, OWR (One-way Ranging) TDoA (Time
Difference of Arrival) measurements can be obtained at the receiver by measuring relative
delays between preambles from RASs. To carry out geo-location in the 2D plane, three or
more preambles should be detected at the receiver. However, WiBro is a cellular based
system and it is difficult to detect more than one preamble at the receiver. Thus, sensitivity of
the receiver needs to be enhanced to detect preambles from neighboring cell. Two integration
methods can be used to enhance sensitivity. One is the coherent integration and the other is
non-coherent integration. The former method is more effective, but integration loss may take
place if there is residual frequency in the carrier. The latter method does not suffer from
residual frequency, but the squaring loss becomes critical when the received signal has low

This paper presents a residual frequency estimator in order to enhance the integration gain in
WiBro network. The next section outlines the residual frequency estimator in WiBro network.
In section 3, based on ITU-R m.1225 channel model, performance of the frequency estimator
is analysed in pedestrian test environment.

2. WiBro receiver structure for geo-location system

2.1 WiBro Signal Structure

WiBro adopts OFDMA TDD and symbols are used as the fundamental transmission unit. A
group of down-link and up-link symbols constitute a frame as shown in Figure 1, and its
duration is 5 msec. The first symbol of a frame is used as a preamble. Mutually orthogonal
preamble code is assigned to RASs. All the RASs are synchronized by GPS timing and the
preamble symbol is transmitted every 5 msec by RASs simultaneously. From the receiver
correlator, TDoA measurement is obtained by detecting preamble. 2D plane geo-location
using TDoA requires at least 2 differential range measurements. Because WiBro is cell-based
system, however, neighboured RAS signals are hardly detected. Therefore, sensitivity of the
receiver should be enhanced by accumulating the preamble symbols.
                                                   Figure 1. Frame structure

2.2 Sensitivity enhancement with hybrid integration

Since identical preamble symbols are repeatedly transmitted every 5msec, coherent
integration method can be used with the preamble. Theoretically, SNR after coherent
integration is proportional to the number of integration. When residual frequency exists in the
signal, however, coherent loss occurs that is given by

                                                                           ⎡⎛                   2⎤
                                                                              sin(πf res NT ) ⎞ ⎥
                                              Coherent Loss[dB] = 10 log10 ⎢⎜                 ⎟
                                                                           ⎢⎜ πf res NT ⎟ ⎥                                          (1)
                                                                           ⎣                  ⎠ ⎥⎦

where f res is residual frequency, N is the number of coherent integration, T is the periodic
of the signal being integrated. For WiBro, T is 5msec, and coherent integration loss is
increased as the number of integration increase. To diminish the loss or enlarge the
integration time, residual frequency needs to be estimated accurately.

Non-coherent integration method integrates the signal after squaring, thus there is no loss
caused by residual frequency. However, squaring operation also influence the noise
characteristics and squaring loss occur as

                                                         ⎡        ⎡ SNR                                                   ⎤
                                                                                                                              2  ⎤
                                                         ⎢    π ⎢ − 4 ⎧⎛           SNR ⎞      ⎛ SNR ⎞ SNR     ⎛ SNR ⎞⎫ ⎥         ⎥
                                                         ⎢      ⋅ ⎢e     ⋅ ⎨⎜1 +       ⎟ ⋅ I0 ⎜     ⎟+    ⋅ I1⎜     ⎟⎬ − 1⎥      ⎥
                                                         ⎢    2
                                                                  ⎢        ⎩⎝       2 ⎠       ⎝ 4 ⎠    2      ⎝ 4 ⎠⎭ ⎥           ⎥
         Squaring Loss [dB] = 10 log10 [SNR ] − 10 log10 ⎢        ⎣                                                       ⎦      ⎥   (2)
                                                         ⎢                                                                      2⎥
                                                         ⎢             ⎡ − SNR                                                ⎤ ⎥
                                                                     π               ⎧⎛    SNR ⎞      ⎛ SNR ⎞ SNR     ⎛ SNR ⎞⎫⎥
                                                         ⎢ SNR + 2 − ⋅ ⎢e 4        ⋅ ⎨⎜1 +     ⎟ ⋅ I0 ⎜     ⎟+    ⋅ I1⎜     ⎟⎬ ⎥
                                                         ⎢           2 ⎢             ⎩⎝     2 ⎠       ⎝ 4 ⎠    2      ⎝ 4 ⎠ ⎭⎥ ⎥
                                                         ⎣             ⎢
                                                                       ⎣                                                      ⎥ ⎥
                                                                                                                              ⎦ ⎦

In the equation, SNR denotes the SNR before squaring, I 0 (•) and I1 (•) are 0th and 1th order
modified Bessel function of first kind each. As shown in Figure 2, squaring loss increase
exponentially as the signal SNR before squaring becomes small.
                                    Figure 2. Squaring loss

Hybrid integration scheme using both coherent and non-coherent integration is commonly
used in sensitivity enhancement. Figure 3 shows the structure of the hybrid integration. In
order to increase SNR, coherent integration is done first. Then, integrated signal is fed into
non-coherent integrator. To increase SNR in coherent integration, integration time should be
enlarged. This means that residual frequency should be reduced to maintain a certain coherent
loss. Therefore, in order to maximize sensitivity of a receiver, residual frequency estimation
with high accuracy is an important factor in sensitivity enhancement.

                             Figure 3. Hybrid integration structure

2.3 Residual frequency estimator

Using the consecutive preamble symbols, residual frequency in the received signal can be
estimated using FFT method. By finding the maximum peak in the FFT of successive
correlation profile, residual frequency is estimated. Input data to FFT is prepared as in Figure
4. Then, the FFT result is written as

                                    1 N −1
                                              − j 2π
                         X [k , m] = ∑ xi [m]e N                (k , i = 0,1, 2,L N − 1)     (3)
                                    N i =0

where xi [m] is the correlation profile of i th preamble symbol, N is the number of sample,
and m is the time index in the correlation.

By finding k that has a maximum magnitude of X [k , m] over a threshold, estimated
residual frequency is obtained that is given by
                                     k max
                           f res =         f FR , k max = arg max X [k , m]              (4)
                                      N                       k

where f FR is the frame rate. Because the frame duration is 5msec in WiBro system, f FR is
equal to 200 Hz. Therefore, if 200-point FFT is used with 1sec data, resolution of frequency
estimation becomes 1 Hz.

                       x0[m]             x1[m]           xN−1[m]

                         Figure 4. Residual frequency estimator using FFT

3. Simulation Results

3.1 Simulation Environment

Performance of the residual frequency estimation is analysed by computer simulation.
Channel model, based on the ITU-R M.1225 document, is used and only path loss and
multipath fading is considered. Channel environment is assumed as Outdoor to indoor and
pedestrian test environment. The system parameters of the WiBro system are shown in the
Table 1.

                 Parameter                                               value
Frequency allocated for WiBro                                       2.3GHz~2.4GHz
Cell coverage                                                            500 m
Bandwidth of the channel                                                10 MHz
System sampling frequency                                               20 MHz
TDD frame length                                                          5 ms
OFDMA symbol time                                                       115.2 us
FFT size per symbol                                                       1024
Noise Power Spectrum Density                                       -104 dBm /10 MHz

                               Table 1. System parameters for simulation

The cell structure is built up as shown in Figure 5.
                              Figure 5. Cell structure for simulation

Based on cell coverage, transmission power of the RAS is set as 14.3 dB as in Eq. (5), where
N 0 is the white Gaussian noise, PL is the path loss, Gcorr is the correlation gain, GMar is
the margin.

         PTX [dB] = N 0 + PL − Gcorr + GMar = −104 + 137.8 − 25.5 + 6 = 14.3[dB] (5)

Since every RAS are synchronized with GPS clock, each RAS’s residual frequency is
assumed same, and is set as 51 Hz. The FFT is used with 1 sec data length.

3.2 Performance of Residual Frequency Estimator

In no multipath environment, residual frequency estimation was performed near the centre of
a cell, and far at the edge of a cell. The coordinates near the centre was set as (10m, 0m) and
the coordinates far at the edge was set as (500m, 0). At each coordinates, the simulation was
repeated for 100 times and the estimated frequency histogram was shown as Figure 6.

                        (a)                                             (b)
        Figure 6. Histogram of an estimated residual frequency in no multipath environment.
                       (a) Near the centre at (10m, 0m), (b) at the edge at (500m, 0m).
When there is no multipath, estimated frequency is same regardless of the distance from the
centre. This indicates that residual frequency estimator is not influenced by SNR of a received
signal, or interference caused from other RAS signals.

Figure 7 is the histogram of the estimated frequency, when the residual frequency estimation
was performed at (500m, 0) in multipath environment.

         Figure 7. Histogram of an estimated residual frequency in multipath environment
                                       at the edge (500m, 0)

In the presence of multipath, additional frequency shift occurred in the residual frequency. In
the simulation, frequency shift error appeared be maximum 2Hz. This shows that multipath
fading has the effect of producing additional frequency error to the signal.


Sensitivity enhancement of the receiver is crucial for detecting more two RASs in WiBro geo-
location system. In this paper, residual frequency estimation was shown to enhance the
performance of the hybrid integration. It was shown in the simulation that multipath has the
effect of providing additional shift in the residual frequency. In conclusion, in multipath
environment the residual frequency estimator will have resolution error within 1Hz, and
frequency shift error with in 2Hz. With hybrid integration, after residual frequency estimation,
it will be able to detect three or more preamble in the most of the cell area.


Caffery Jr. JJ (1999) Wireless Location In CDMA Cellular Radio System, Kluwer Academic
      Publishers, 23-40pp
Holmes J (1991) Coherent Spread Spectrum Systems, Krieger Publishing Company
Strässle C, Megnet D, and Mathis H (2007) The Squaring-Loss Paradox, ION GNSS 20th
       International Technical Meeting of the Satellite Division, Fort Worth, TX, Sept.
Soliman S, Glazko S, and Agashe P, (1999) GPS Receiver Sensitivity Enhancement in Wireless
     Applications, Proceedings of the IEEE MTT-S International Tipical Symposium on
     Technologies for Wireless Applications
TTA (2005) Specifications for 2.3GHz band Portable Internet Service (PHY & MAC Layer).
     Available: Accessed 10 Sept 2009

To top