Wireless Pers Commun (2015) 84:3143–3156

DOI 10.1007/s11277-015-2791-6

Adaptive NLOS Mitigation Location Algorithm

in Wireless Cellular Network

Shixun Wu1 • Dengyuan Xu1 • Honggang Wang1

Published online: 18 June 2015

 Springer Science+Business Media New York 2015

Abstract In a cellular network, a location scenario is considered in this paper, which the
home base station (BS) measures the range and angle while neighboring BSs measure the
ranges. Based on the range and angle measurements, two adaptive non-line-of-sight
(NLOS) identification and mitigation location algorithms with low complexity are pro-
posed. The main idea of our proposed algorithms is to search the optimal subset of range
and angle measurements by utilizing the hybrid lines of position algorithm and comparing
the normalized residual error to perform the NLOS identification and location. Compared
with the existing algorithms, the proposed algorithms have three advantages: one is dif-
ferent to the existing algorithms that require at least three BSs, the proposed algorithms can
be suitable for two BSs scenario. Second is the low complexity of our proposed algorithms
compared with the existing algorithms. The last is simulation results show that the pro-
posed algorithms outperform the existing algorithms in bad urban and urban environments.

Keywords Non line of sight (NLOS)  Adaptive  Cellular network  Location 

Algorithm

& Shixun Wu
[email protected]
Dengyuan Xu
[email protected]
Honggang Wang
[email protected]
1
Information Science and Engineering College, Chongqing Jiaotong University, Chongqing 400074,
China

123
3144 S. Wu et al.

1 Introduction

Due to the wide range applications of location-based services (LBSs), wireless location has
gained an increasing interest in wireless cellular network. The most popular means of
wireless location involve the measurement of time of arrival (TOA), time difference of
arrival (TDOA), received signal strength (RSS) and angle of arrival (AOA) [1–5] of radio
signals transmitted by the mobile station (MS) and received at a number of fixed base
stations (BSs) or vice versa.
The main challenge for accurate MS location estimation with the above measurements
is the presence of non-line-of-sight (NLOS) propagation. In NLOS environments, the work
in [6] proposes residual weighting (Rwgh) algorithm to alleviate the effect of NLOS error.
The main advantage of Rwgh algorithm is that no statistical prior information about NLOS
error is needed. However, it is required at least three BSs and computationally intensive
when the number of BSs is large. Moreover, for each combination of the range mea-
surements, the nonlinear least square (NLS) algorithm is employed to obtain the temporary
estimator of MS location. As we know, the NLS algorithm is an iterative approach,
selecting a good initial guess to avoid a local minimum is difficult and convergence to the
optimal solution can not be guaranteed. In terms of high computational complexity of the
Rwgh algorithm, iterative minimum residual (IMR) algorithm is presented in [7], which
has considerably lower computational complexity than the Rwgh algorithm and somewhat
better performance than Rwgh algorithm. However, the NLS algorithm is still employed to
solve each combination of the range measurements. In order to overcome the drawback of
NLS algorithm, the work in [8] employs line of position (LOP) algorithm to solve each
combination of the range measurements and proposes a low complexity localization
algorithm using minimum subset to further reduce the computational complexity of IMR
algorithm. However, it also assumes that there are at least three BSs involved in location.
In fact, due to hearability problem and the excessive signaling cost, the number of BSs is
limited.
Recently, there is a tendency to develop new location algorithms with limited BSs. In
[9, 10], three BSs location algorithms are proposed by utilizing nonlinear constrained
optimization. With the help of angle measurement, two BSs location algorithm is presented
in [11]. The work in [12] proposes the scattering distance-based algorithm with only one
BS using the multipath. Therefore, it is meaningful work to extend the idea of Rwgh and
IMR algorithm to adapt for limited BSs scenario. In fact, the angle measurement is a good
choice to assist localization in limited BSs scenario. With the help of angle measurement,
the work in [9] proposes the hybrid lines of position (HLOP) algorithm which has low
complexity and can avoid the disadvantage of the NLS algorithm. Based on the above two
considerations, two adaptive NLOS mitigation location algorithms are proposed in this
paper by comparing the normalized residual error and using the HLOP algorithm. The
proposed algorithms can effectively reject the measurement with large errors and select a
subset of the range and angle measurements to perform localization. Moreover, the
computational complexity with different algorithms is also analyzed. Simulation results
prove the effectiveness of our proposed algorithms.
The rest of paper is organized as follows. Section 2 introduces the system model.
Section 3 presents the proposed algorithms. Section 4 provides the simulation results and
Sect. 5 concludes the paper.

123
Adaptive NLOS Mitigation Location Algorithm in Wireless… 3145

2 System Model

There are N BSs available for the MS location, BS1 is the home BS which performs range
and angle measurements upon receiving the signal, while the neighboring BSi ; i ¼ 2; . . .; N
only measures the ranges. The measured ranges and angle can be described as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ri ¼ ðx  xi Þ2 þ ðy  yi Þ2 þ ei þ ni ¼ cti ; i ¼ 1; . . .; N
  ð1Þ
y  y1
b ¼ atan þ nb
x  x1
where c is the speed of light, ti is the measured TOA between MS and BSi, (x, y) is the
coordinate of MS, (xi, yi) is the coordinate of BSi. ei and ni are the NLOS error and
measurement noise, respectively. b is the angle measurement in home BS, atan is the
function of inverse tangent, nb is the error of angle measurement, which is the sum of angle
measurement noise and NLOS angle deviation. If BSi has a line-of-sight (LOS) path to the
MS, then ei = 0.
Equivalently, without angle measurement Eq. in (1), we can transform (1) into the
following vector form
r ¼ gðhÞ þ v ð2Þ

where r ¼ ½r1 ; r2 ; . . .rN T ; h ¼ ½x; yT ,

gðhÞ ¼ ½g1 ðhÞ; g2 ðhÞ; . . .; gN ðhÞT ;

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
gi ðhÞ ¼ ðx  xi Þ2 þ ðy  yi Þ2 ;
v ¼ ½e1 þ n1 ; e2 þ n2 ; . . .; eN þ nN T :

With the help of angle measurement in home BS, the MS location estimation h^ can be
obtained by HLOP algorithm [9], then we can compute the residual error as follow
^ ¼ ½r  gðhÞ
eðhÞ ^
^ T ½r  gðhÞ ð3Þ

3 Proposed Location Algorithms

In this section, we firstly present HLOP algorithm, and then explain Rwgh-HLOP algo-
rithm which is the combination of Rwgh algorithm and HLOP algorithm. In terms of high
computational complexity and incorrect combination of range and angle measurements
about Rwgh-HLOP algorithm, two location algorithms with low complexity are presented.
Finally, the computational complexity of different algorithms (Rwgh, IMR, MS-IMR,
Rwgh-HLOP and the proposed algorithms) are analyzed and compared.

3.1 HLOP Algorithm

If we ignore the effect of errors about range and angle measurement Eq. in (1), by doing
some mathematical manipulation, the nonlinear range and angle equations can be
expressed in linear matrix form

123
3146 S. Wu et al.

Ah ¼ b ð4Þ
where
2 3 2 2 3
x1  x2 y1  y2 r2  r12 þ K1  K2
6 .. .. 7 16 .. 7
6 . . 7 6 . 7
A¼6 7; b ¼ 6 7;
4 x1  xN y1  yN 5 2 4 r2  r 2 þ K1  KN 5
N 1
tanðbÞ 1 2ðx1 tanðbÞ þ y1 Þ
Ki ¼ x2i þ y2i and tan is the function of tangent. The MS location can be estimated with a
least squares approach

h ¼ ðAT AÞ1 Ab ð5Þ

3.2 Rwgh-HLOP Algorithm

For our location scenario, the Rwgh-HLOP algorithm which refers to Rwgh algorithm [6]
consists of the following steps.
Define two sets, H ¼ fr1 ; bg and Dmin ¼ fr2 ; . . .; rN g, where H is the measurement set
of home BS and Dmin is the measurement set of other BSs.
 
P N1
N1
1. 8 non-empty Set r  Dmin , do D ¼ H [ r to form M ¼ measurement
i¼1 i
combinations. Each combination is represented by a index set fSk ; k ¼ 1; . . .; M g.
2. For each combination, compute the intermediate estimate of MS h^k ; k ¼ 1; . . .; M and
^
the corresponding normalized residual eðh^k ; Sk Þ ¼ eðhk Þ
size of Sk 1 in Eq. (3) with HLOP
algorithm.
The final estimate of MS h is the weighted linear combination of the intermediate
estimates h^k
PM ^
eðh^k ; Sk ÞÞ1
hk ð
h ¼ Pk¼1 ð6Þ
M
eðh^k ; Sk ÞÞ1
k¼1 ð

3.3 Proposed Location Algorithms

For the Rwgh-HLOP algorithm, there are two drawbacks. One is the number of combi-
nation M becomes very large as N increase, which leads to high computation complexity;
the other is the bad intermediate estimates from some combinations contained the NLOS
BSs, which will decrease the performance of the algorithm. Our motivation is devoted to
solve these two problems. The main idea is to iteratively reject the bad observation data (or
add the good observation data) one by one and search the optimal set of the observation
data. The final estimate of MS is obtained with Rwgh-HLOP algorithm by choosing one
observation data from the optimal set and combining with the set H.
The proposed location algorithms first perform data processing [13] with the range
measurements to reduce the effect of big NLOS error. The idea of rejecting the bad
observation data denoted as the proposed algorithm I is described as the following steps:

123
Adaptive NLOS Mitigation Location Algorithm in Wireless… 3147

1. Initialization
H ¼ fr1 ; bg is the measurement set of home BS and Dmin ¼ fr2 ;    ; rN g is the
measurement set of other BSs.
2. HLOP
Find HLOP solution h^ with the observation data D ¼ H [ Dmin and determine the
^ Set the normalized residual error emin ¼ eðhÞ=N.
residual error of the estimator eðhÞ. ^
3. Iteration  
N 1
For set Dmin, make ¼ N  1 Combinations ðDm ; m ¼ 1; . . .; N  1Þ. For
N 2
each set Dm, do Em ¼ H [ Dm . Compute temporary MS location estimation h^ðmÞ and
the residual error eðh^ðmÞ Þ using HLOP algorithm. Decide the minimum normalized
residual error e and the set D0min

½e; k ¼ minðeðh^ðmÞ Þ; m ¼ 1; . . .; N  1Þ
e ¼ e=ðN  1Þ ð7Þ
D0min ¼ Dk
0
emin and N [ 2, then N = N - 1, Dmin = Dmin, emin ¼ e, repeat 3; else return
If e\
Dmin.
8 element r 2 Dmin , do R ¼ H [ Dmin , and the final estimate is obtained following the
step 3 and 4 in Rwgh-HLOP algorithm. The flowchart of the proposed location
algorithm I is shown in Fig. 1.
The idea of adding the good observation data denoted as the proposed algorithm II is
presented as
1. Initialization
H ¼ fr1 ; bg is the measurement set of home BS, Dmin ¼ fr2 ; . . .; rN g is the
measurement set of other BSs and D ¼ fg is the empty set.
2. HLOP
For each element rm ; m ¼ 2; . . .; N in Dmin , do Em ¼ H [ rm . Find HLOP solution h^ðmÞ
with Em and determine the residual error of the estimate eðh^ðmÞ Þ. The minimum
residual error and the corresponding normalized residual error are

½e; k ¼ minfeðh^ðmÞ Þ; m ¼ 2; . . .; Ng
e ð8Þ
emin ¼
Size of Em
Do Dmin ¼ Dmin  rk ; D ¼ D [ rk
3. Iteration
For set Dmin , do Em ¼ H [ D [ rm ; rm 2 Dmin . Compute temporary MS location
estimation h^ðmÞ , the residual error eðh^ðmÞ Þ using HLOP algorithm. Decide the minimum
normalized residual error e

½e; k ¼ minðeðh^ðmÞ Þ; m ¼ 1; . . .; fSize of Ek g  1Þ

e ð9Þ
e ¼
fSize of Ek g  1
emin and Dmin is non-empty, then emin ¼ e, Dmin ¼ Dmin  rk , D ¼ D [ rk , repeat
If e\
3; else return D.

123
3148 S. Wu et al.

Fig. 1 The flowchart of the

proposed location algorithm I Start

Data Processing

Initialization
H ={r1 , β } Dmin = {r2 , , rN }

D = H ∪ Dmin
Compute ε min

For set Dmin , make ⎛ N − 1 ⎞ = N − 1 combination

⎜ ⎟
⎝ N − 2⎠
′
Dmin = Dmin ( Dm , m = 1, 2, , N − 1) . Do Em = H ∪ Dm , compute
ε min = ε θˆ( m ) , ε (θˆ( m ) ) and corresponding ′ ,ε
Dmin
N = N −1

No
ε < ε min

Yes

Yes
N >2

No

Return Dmin

Do H ∪ Dmin

Rwgh-HLOP Algorithm

End

8 element r 2 D, do R ¼ H [ r, and the final estimate is obtained following the step 3

and 4 in Rwgh-HLOP algorithm. The flowchart of the proposed location algorithm II is
shown in Fig. 2.

3.4 Complexity Analysis

If the HLOP algorithm is adopted for each combination of the observation data, the
computational complexity among the different algorithms can be almostly ignored. Thus,
the computational complexity of different algorithms is mostly measured by the possible
combination of observation data. In order to compare the computational complexity of the

123
Adaptive NLOS Mitigation Location Algorithm in Wireless… 3149

Fig. 2 The flowchart of the

proposed location algorithm II Start

Data Processing

Initialization
H ={r1 , β }Dmin = {r2 , , rN } D = {}

Do Em = H ∪ rm , rm ∈ Dmin
Compute (8) and do
Dmin = Dmin − rk , D = D ∪ rk

For set Dmin , do Em = H ∪ D ∪ rm , rm ∈ Dmin

compute (9)
ε min = ε
Dmin = Dmin − rk
D = D ∪ rk No
ε < εmin

Yes

No
isempty Dmin

Yes

Return D

Do H ∪ r , r ∈ D

Rwgh-HLOP Algorithm

End

iteration algorithms, the worst case which achieves the maximum iteration is analyzed. For
different algorithms, the number of combination is shown in Table 1. From Table 1, we
can see that two proposed location algorithms have the same complexity, and lower
complexity in comparison with the existing algorithms, especially when the number of BSs
N is large.

123
3150 S. Wu et al.

Table 1 The number of combination in related algorithms

Related algorithm The number of combination
 
Rwgh P
N N
; N 3
i¼3 i
8  
IMR (maximum iteration) P
< N4 N i
þ 1; N  4
Ni1
: i¼0
1; N¼3
8   
MS-IMR (maximum iteration) <P N i i1
þ 1; N  4
i1 3
: i¼4
1; N¼3
 
Rwgh-HLOP P N 1
N1
; N 2
i¼1 i
8  
Proposed I (maximum iteration) P
< N1 N i
þ 1; N  3
Ni1
: i¼1
1; N¼2
8  
Proposed II (maximum iteration) P Ni
< N1
þ 1; N  3
1
: i¼1
1; N¼2

4 Simulation Results

In this section, simulation results are performed to illustrate the effective of the proposed
algorithms. The simulation parameters refer to [14], the LOS range error ni is modeled as
zero mean Gaussian distribution with the same standard deviation between 30 and 60 m.
The NLOS range error ei is the speed of light multiply the propagation delay between
direct path and other paths. The probability density function of the propagation delay is
modeled as an exponential distribution as follow
(
1 s s
e rms ; s [ 0
PðsÞ ¼ srms ð10Þ
0; otherwise
where s is the NLOS delay, srms is the root mean square (rms) delay spread, which is
modeled as srms ¼ T1 de q, where T1 is the median value of the rms delay spread at one
kilometer, d is the distance between the MS and BS, e is the path loss exponent with value
between 0.5 and 1, q is a lognormal variable (Such that 10 log q is a zero mean Gaussian
variable with standard deviation rq that lies between 4 and 6 dB). Typical parameters for
different environments are given in table 2.
The angle error nb is composed of the LOS angle error and NLOS angle error. The LOS
angle error can be modeled as zero mean Gaussian random variable with standard devi-
ation of approximately 3 and the NLOS angle error can be considered as a zero mean
Gaussian random variable with standard deviation 5–10 [14].
We assume a hexagonal layout with 1000 m of cell radius that is shown in Fig. 3. Four
pffiffiffi
BSs are deployed with coordinate BS1(0, 0), being the home BS, BS2 ð500 3; 1500Þ,
pffiffiffi pffiffiffi
BS3 ð1000 3; 0Þ and BS4 ð500 3; 1500Þ, the coordinate of MS is randomly generated
within OABCD. The performance criterion of the algorithm is chosen as the average
location error (ALE)

123
Adaptive NLOS Mitigation Location Algorithm in Wireless… 3151

Fig. 3 Geometry of circular y

scattering model

BS 2

O x
BS1 BS 3

C
D

BS 4

Table 2 Typical parameters in

Environments T1 (ls) e rq (dB)
different environments
Bad Urban 1.0 0.5 4
Urban 0.4 0.5 4
Suburban 0.3 0.5 4
Rural 0.1 0.5 4

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1X T
ALE ¼ ðxi  x^i Þ2 þ ðyi  y^i Þ2 ð11Þ
T i¼1

where ð^xi ; y^i Þ is the estimate of MS position ðxi ; yi Þ and ALE is obtained by computing
T = 1000 independent location trials.
Different algorithms which are shown in Table 1 are compared to validate the proposed
algorithms. For our simulation scenario, IMR algorithm is equivalent with MS-IMR
algorithm. We only consider IMR algorithm. Moreover, only Bad Urban and Urban
environments in Table 2 are simulated. We first compare the performance with the number
of NLOS BSs. The simulation parameters are ni  Nð0; 45mÞ, nb  Nð0; 30 Þ if the home BS
experiences LOS propagation, while nb  Nð0; 100 þ 30 Þ if the home BS experiences
NLOS propagation. As shown in Figs. 4 and 5, the ALEs of different algorithms increase
with the number of NLOS BSs in Bad Urban and Urban environments, respectively. The
proposed two algorithms have the similar performance and are superior compared to other
algorithms. In fact, the proposed algorithm II is slightly better than the proposed algorithm
I when it exists LOS BSs, whereas the proposed algorithm I is slightly better than the
proposed algorithm II when there are no LOS BSs. With the assumption that all the BSs are
corrupted by NLOS propagation, Figs. 6 and 7 are shown to evaluate the effect of range
measurement error in Bad Urban and Urban environments, respectively. The simulation

123
3152 S. Wu et al.

Fig. 4 Performance comparison Bad Urban

with Number of NLOS BSs in 700
Bad Urban environment
Rwgh
600
IMR

Average Location error(ALE)

Rwgh-HLOP
500
proposed I
proposed II
400

300

200

100

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Number of NLOS BSs

Fig. 5 Performance comparison Urban

with number of NLOS BSs in 350
Urban environment
300 Rwgh
Average Location error(ALE)

IMR
250 Rwgh-HLOP
proposed I
200 proposed II

150

100

50

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Number of NLOS BSs

parameter is nb  Nð0; 100 þ 30 Þ. It is shown that the standard deviation of range mea-
surements have little effect on ALEs and the proposed two location algorithms have better
performance. With the assumption that all the BSs are corrupted by NLOS propagation,
Figs. 8 and 9 are shown to evaluate the effect of angle NLOS measurement error. The
simulation parameter is ni  Nð0; 45mÞ and the LOS angle measurement error is Nð0; 30 Þ.
As the NLOS angle measurement error increases, the performance of two proposed
location algorithms decreases, whereas other algorithms do not decrease. But the proposed
algorithms are still better than other algorithms.

123
Adaptive NLOS Mitigation Location Algorithm in Wireless… 3153

Fig. 6 Performance comparison Bad Urban

with the standard deviation of 800
range measurement in Bad Urban Rwgh
environment 750 IMR

Average Location error(ALE)

Rwgh-HLOP
700 proposed I
proposed II
650

600

550

500

450
30 35 40 45 50 55 60
The standard deviation of range measurements(m)

Fig. 7 Performance comparison Urban

with the standard deviation of 380
range measurement in Urban
environment 360
Rwgh
IMR
Average Location error(ALE)

340 Rwgh-HLOP
proposed I
320 proposed II

300

280

260

240

220
30 35 40 45 50 55 60
The standard deviation of range measurements

5 Conclusion

In this paper, we consider a location scenario that the home BS measures the range and
angle while the neighboring BSs measure the ranges. By using the HLOP algorithm and
comparing the normalized residual error, two adaptive NLOS mitigation algorithms with
low complexity are proposed. Simulation results demonstrate: (1) for different number of
NLOS BSs, the proposed location algorithms are superior to other algorithms in different
environments. (2) The standard deviations of range measurement do not affect the per-
formance of all location algorithms. (3) As the NLOS angle measurement error increases,
the performance of the proposed algorithms decreases, whereas other algorithms do not
affected.

123
3154 S. Wu et al.

Fig. 8 Performance comparison Bad Urban

with the standard deviation of
NLOS angle measurement in Bad 640
Urban environment 620

Average Location error(ALE)

600

580

560

540 Rwgh
IMR
520
Rwgh-HLOP
500 proposed I
480
proposed II

460

440
5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
The standard deviation of NLOS angle
measurements(Degree)

Fig. 9 Performance comparison Urban

with the standard deviation of 360
NLOS angle measurement in Rwgh
Urban environment 340 IMR
Rwgh-HLOP
Average Location error(ALE)

320 proposed I
proposed II
300

280

260

240

220

200
5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
The standard deviation of NLOS angle
measurements(Degree)

Acknowledgments This research was supported by the Science and Technology Research Project of
Chongqing Municipal Education Commission (KJ1400305), Foundation and Frontier Research Project of
Chongqing (cstc2013jcyjA30016), Natural Science Foudation of Chongqing (cstc2012jja40065).

References
1. Sayed, A. H., Tarighat, A., & Khajehnouri, N. (2005). Network-based wireless location: Challenges
faced in developing techniques for accurate wireless location information. IEEE Signal Processing
Magazine, 22(4), 24–40.

123
Adaptive NLOS Mitigation Location Algorithm in Wireless… 3155

2. Guvenc, I., & Chong, C. C. (2009). A survey on TOA based wireless localization and NLOS mitigation
techniques. IEEE Communications Survey & Tutorials, 11(3), 107–124.
3. Wu, S., Wang, S., Xu, K., & Wang, H. (2013). Hybrid TOA/AOA cooperative mobile localization in 4G
cellular networks. IEEK Transactions on Smart Processing and Computing, 2(2), 77–85.
4. Wu, S., Li, J., & Liu, S. (2013). Improved localization algorithms based on reference selection of linear
least squares in LOS and NLOS environments. Wireless Personal Communications, 68(1), 187–200.
5. Wu, S., Xu, D., & Liu, S. (2013). Weighted linear least square localization algorithms for received
signal strength. Wireless Personal Communications, 72(1), 747–757.
6. Chen, P. C. (1999). A non-line-of-sight error mitigation algorithm in location estimation. In IEEE
wireless communications and networking conference, Vol. 1, (pp. 316–320).
7. Li, X. (2008). An iterative NLOS mitigation algorithm for location estimation in sensor networks. In
15th IST mobile & wireless communications summit.
8. Fujita, T., & Ohtsuki, T. (2008). Low complexity localization algorithm based on NLOS node iden-
tification using minimum subset for NLOS environments. In IEEE global telecommunications con-
ference (pp. 1–5).
9. Venkatraman, S., & Caffery, Jr. J. (2004). Hybrid TOA/AOA techniques for mobile location in non-
line-of-sight environments. In IEEE wireless communications and networking conference, Vol. 1 (pp.
274–278).
10. Wu, S., Li, J., Liu, S., & Chong, J. W. (2012). Improved and extended range scale algorithm for wireless
cellular location. IEEE Communications Letters, 16(2), 196–198.
11. Tang, H., Park, Y. W., & Qiu, T. S. (2008). A TOA-AOA-based NLOS error mitigation method for
location estimation. EURASIP Journal on Advances in Signal Processing, 2008, Article ID 682528,
14 pp.
12. Zhaounia, M., Landolsi, M. A., & Bouallegue, R. (2009). A novel scattering distance-based mobile
positioning algorithm. In Global information infrastructure symposium (pp. 1–4).
13. Wu, S., Li, J., & Liu, S. (2011). Improved positioning algorithm based on two step least square in NLOS
environments. The journal of China University of Posts and Telecommunications, 18(5), 58–63.
14. Khajehnouri, N., & Sayed, A. H. (2003) A non-line-of-sight equalization scheme for wireless cellular
location. In IEEE international conference on acoustics, speech and signal processing, Vol. VI (pp.
549–552).

Shixun Wu was born in HuBei, China. He received the B.S. and M.S.
degrees in 2006 and 2009, respectively. He received the Ph.D. degree
in 2012 at the Department of Electrical and Computer Engineering,
Central China Normal University, China. His current research interests
include wireless communication and radio positioning.

123
3156 S. Wu et al.

Dengyuan Xu received his B.Sc. degree and M.Sc. degrees in Com-

puter Science from Southwest Petroleum University, China in 1995
and 2001 respectively. He received Ph.D. degree in Communication
and Information System from Southwest Jiaotong University China in
2005. From 1995 to 1998, he was an engineer in Information Centre at
Jinan Diesel Engine Factory. From 2005 to 2006, he was an engineer in
Network Development Department at Sichuan Telecom of China.
Currently he is a vice-professor at School of Information Science and
Engineering and he is dean of Communication Engineering Depart-
ment, Chongqing Jiaotong University. His research interests include
wireless ad-hoc and sensor networks, switch technology, bridge health
monitoring.

Honggang Wang was born in Shanxi, China, on 1st June, 1970. He

received his B.S. degree from Xidian University in 1992, his M.S.
degree from Taiyuan University of Science and Technology in 1997,
the Ph.D. degrees from China Academy of Railway Science in 2006.
His current research interests include hybrid dynamic system and train
traffic control.

123