JP2009545755A - Weighted least squares positioning method using multipath channel statistics for non-line-of-sight mitigation - Google Patents

Abstract

The method mitigates NLOS conditions based on a weighted least squares (WLS) approach. In that case, the weight value is derived from the multipath component (MPC) of the received signal. The weighting methodology can be used in both linear and nonlinear least squares models, as well as different other NLOS mitigation schemes such as residue-based algorithms or maximum likelihood approaches.
[Selection] Figure 2

Description

Cross-reference of related applications

  This application includes: (a) U.S. Provisional Patent Application No. 60 / 821,378 filed on August 3, 2006; (b) U.S. Provisional Patent Application No. 60/822 filed on August 11, 2006; 127, (c) U.S. Provisional Patent Application No. 60 / 823,367, filed on August 23, 2006, (d) U.S. Patent Application No. 11/832, filed on Aug. 1, 2007. 547, (e) U.S. Patent Application No. 11 / 832,551 filed on August 1, 2007, and (f) U.S. Patent Application No. 11 / 832,558 filed on Aug. 1, 2007. Claim the priority of these applications. All of these applications are incorporated herein by reference. In the US designation, this application is a continuation of the above-mentioned US patent application Ser. No. 11 / 832,558.

Background of the Invention

1. The present invention relates to wireless positioning and communication technology. More specifically, the present invention relates to estimating mobile terminal location using a time of arrival (TOA) approach when a non-line-of-sight (NLOS) condition exists.

2. Discussion of Related Art Ultra-wideband (UWB) technology promises an accurate ranging and positioning system that can solve individual multipath components (MPC) because of its very wide bandwidth. Using UWB technology, the arrival time (TOA) of the received signal can be estimated with high accuracy when the first arrival path is correctly identified. Various systems have been disclosed that use UWB technology. Among these systems are the systems disclosed in the following papers. That is, (a) B. Alavi and K.K. By Pahlavan, Proc. IEEE Vehic. Technol. Conf. (VTC), vol. 4, Dallas, TX, Sep. 2005, pp. “Analysis of undetected direct path in time of arrival based on UWB interior geology” published in U.S. Pat. Guvenc, Z .; Sahinoglu, A.I. F. Morich and P.M. Orlik by Proc. IEEE Int. Works on Ultrawideband Networks (UWBNETS), Boston, MA, October 2005, pp. “Non-coherent TOA estimation in IR-UWB systems with differential signal forms” (invited paper), (c) D.C. Dardari, C.I. C. Chong, and M.M. Z. “Analysis of threshold-based Destination B” (Invited by the 14th European Signal Processing Conference (EUSIPCO 2006), Florence, Italy, September 2006). Dardari, C.I. C. Chong, and M.M. Z. Win by IEEE Intl. Conf. on Ultra-Wideband (ICUWB 2006), Waltham, MA, USA, September 2006, “Improved lower bounds on time of arrival estimation error in Uri Buri.

  One challenge of the positioning system is to successfully mitigate the NLOS effect. When the direct path between the anchor node (AN) and the mobile terminal is interrupted, the TOA of the signal to the AN is delayed, which introduces a positive bias. The NLOS TOA estimate has a negative impact on positioning accuracy. Thus, prior art cellular networks typically identify ANs in NLOS conditions and mitigate the effects of those ANs. For example, M.M. P. Wylie and J.W. By Holtzman, Proc. IEEE Int. Conf. Universal Personal Commun. , Cambridge, MA, Sept. 1996, pp. The paper “The non-line of light in mobile location estimation” published in 827-831 refers to comparing the standard deviation of ranging measurements with the NLOS signal discrimination threshold when the measurement noise variance is known. Teach. Similarly, J.M. Borras, P.M. Hatrack, and N.I. B. Mandyam, Proc. IEEE Vehicular Technol. Conf. (VTC), vol. 2, Ontario, Canada, May 1998, pp. The paper “Decision theoretic framework for NLOS identification” published in 1583-1587 discloses a decision-theoretic NLOS identification framework that uses various hypothesis tests for known and unknown probability density functions (PDFs) of TOA measurements.

  S. Gezici, H.C. Kobayashi, and H.K. V. Poor by Proc. IEEE Vehic. Technol. Conf. (VTC), vol. 4, Orlando, FL, Oct. 2003, pp. The paper “Non-parametric non-line-of-sight identification” published in 2544-2548 discloses a non-parametric NLOS identification approach that allows approximation of the PDF of TOA (ie distance) measurements. An appropriate distance metric is used between the known measurement noise distribution and the non-parameter estimated measurement distribution.

  The above NLOS identification techniques all assume that the NLOS base station (BS) TOA measurements change over time. This is reasonable for mobile terminals. In the case of a moving terminal, the dispersion of TOA measurement values is large. However, when the terminal is stationary (eg, in a wireless personal application network (WPAN) application), the distribution of NLOS measurements shows little deviation from the distribution in the LOS state. Thus, the multipath characteristics of the received signal provide useful insights for LOS / NLOS discrimination. For example, X. Diao and F.D. “A method distributing line of light (LOS) from non-line-of-sight (NLOS) in CDMA mobile communication” filed by Guo on March 29, 2003 and published on October 20, 2004. The European patent application publication EP 1,469,685 entitled: (1) the power ratio of the global maximum path to the local maximum path is greater than a given threshold; (2) the arrival time between the first path and the maximum path If the difference is less than a given time, then the received code division multiple access (CDMA) signal is disclosed to be LOS. Similarly, Rabbachin, I. et al. Oppermann, and B.I. By Denis, Proc. IEEE Int. Conf. Ultrawideband (ICUWB), Waltham, MA, Sept. The paper “ML time-of-arrival estimation based on low complexity UWB energy detection” published in 2006 states that the UWB NLOS identification may be performed by comparing the normalized strongest path to a fixed threshold. Disclose. In either scheme, parameters (eg threshold and time) need to be carefully selected.

  As an alternative to identifying the NLOS condition from the received multipath signal, information derived from the entire mobile network may be used to mitigate the NLOS condition. For example, P.I. C. Chen by Proc. IEEE Int. Conf. Wireless Commun. Networking (WCNC), vol. 1, New Orleans, LA, Sept. 1999, pp. The paper “A non-line-of-light error mitigation algorithm in location estimation” published at 316-320 discloses a residue-based algorithm for NLOS mitigation. This algorithm is based on three or more available BSs that use location estimates and remainders for different combinations of BSs. (When all nodes are LOS, 3 BSs are required to perform 2D (2-D) positioning and 4 BSs are required to perform 3D (3-D) positioning The location estimate with a smaller remainder is likely to represent the correct terminal location. Thus, the approach disclosed in the paper weights the different location estimates inversely with the corresponding remainder.

  Other NLOS mitigation techniques that use information derived from mobile networks are: Casas, A.M. Marco, J. et al. J. et al. Guerrero, and J.M. By Falco, Eurasip J. et al. Applied Sig. Processing, pp. 1-8, 2006, “Robust Estimator for non-line-of-light error mitigation in interior localization,” (b) Y. et al. T.A. Chan, W.H. Y. Tsui, H .; C. So, and P.I. C. By Ching, IEEE Trans. Vehic. Technol. , Vol. 55, no. 1, pp. 17-24, Jan. “Time-of-arrival based localization under NLOS conditions” published in 2006, (c) B.R. Li, A.I. G. Dempster, and C.I. Rizos, Proc. IEEE Position Location and Navigation Symposium (PLANS), San Diego, CA, Apr. “A database method to mitigate the NLOS error in mobile phone positioning” published in 2006, (d) X. Li by Proc. IST Mobile and Wireless Commun. “An iterative NLOS mitigation algorithm for location in sensor networks” published in Summit, Myconos, Greece, June 2006, (e) L. , Cong and W.W. Zhuang by Proc, IEEE INFOCOM, Hong Kong, Mar. 2004, pp. “Non-line-of-light error migrating in mobile location” published in 650-659, (f) J. Am. Riba and A.I. Urruela by Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP), vol. 2, quebec, canada, may 2004, pp. “A non-line-of-sight mitigation technique based on ML-detection” published on 153-156, (g) Venkate and R.K. M.M. By Buehler, Proc. IEEE IPSN, Nashville, Tennessee, Apr. “A linear programming approach to NLOS error mitigation in sensor networks” published in 2006, (h) C.I. L. Chen and K.K. T.A. Feng by Proc. IEEE Int. Conf. Wireless Networks, Commun. , Mobile Computing, Hawaii, USA, June 2005, pp. "An effective geometry-constrained location estimation algorithm for NLOS environment" published in 244-249, and (i) X. Wang, Z .; Wang, and B.W. O. Dea by IEEE Trans. Vehic. Technol. , Vol. 52, no. 1, pp. 112-116, Jan. It is disclosed in "A TOA based location algorithm reducing the errors due to non-line-of-sight (NLOS) propagation" published in 2003.

  Because some prior art positioning algorithms assign equal confidence to each BS, these positioning algorithms do not take into account NLOS conditions. As a result, the presence of NLOS BS significantly degrades the positioning accuracy in these algorithms.

  The prior art further includes a number of weighted least square approaches for estimating the mobile terminal location. Typically, in these approaches, the weights for the received signal from each BS are derived from the measured variance (eg, MP Wylie et al., J. Borras et al., S. Gezici et al. See). These approaches rely on the fact that measurements for moving terminals exhibit a large variance in the NLOS state. However, such an approach does not reliably provide accurate information regarding NLOS BSs.

  Weighted least squares based on measured variance generally require a great deal of observation. Therefore, a large memory is needed to store the measured distance and delay. These measured distances and delays are necessary to estimate the location of the mobile terminal.

Summary of the Invention

The present invention provides an NLOS mitigation technique that suppresses NLOS fixed terminal ¹ (FT) based on UWB channel amplitude and delay statistics. Such statistics include, for example, the kurtosis of the received multipath component of the received signal, the average access delay, and the root mean square (RMS) delay spread. According to one embodiment of the present invention, the weighted least squares method uses weight values obtained from a likelihood function that distinguishes LOS states from NLOS states.

^One fixed terminal (FT) is a terminal that has not moved relative to the mobile terminal. For example, the FT includes a BS in a wireless or cellular communication network, an access point in a wireless computer network, and an anchor node in a sensor network.

  According to one embodiment of the present invention, the weighted least squares method of the present invention is based on conventional algorithms (eg, the residue-based weighted NLOS mitigation algorithm disclosed in the paper by PC Chen et al. Discussed in the previous paragraph). ) May be used to improve performance.

  The present invention is better understood upon consideration of the detailed description below in conjunction with the accompanying drawings.

Detailed Description of the Preferred Embodiment

ここで、Ｌはマルチパス成分（ＭＰＣ）の総数であり、γ_ｌ及びτ_ｌは、それぞれｌ番目のＭＰＣの振幅及び遅延である。受信信号のＴＯＡはτ_ｔｏａ＝τ_１によって与えられる（即ち、最初の到着パスに対する到着時刻）。ＬＯＳ仮説及びＮＬＯＳ仮説をそれぞれ表す仮説Ｈ_０及びＨ_１は、次式によって与えられる。

ここでｄはＦＴと移動端末との間の距離を表し、ｃは光の速度を表す。ＮＬＯＳ状態において、最初の到着パスが正しく識別されるときでも、ＴＯＡ推定値は依然として実際の距離と比較して大きな値を産出する。したがって、位置決め性能の劣化を避けるため、ＮＬＯＳ ＦＴが識別され、次にそれらＮＬＯＳ ＦＴの効果が軽減される。 According to one embodiment of the present invention, the channel impulse response (CIR) h (t) of the received signal is represented by the following equation.

Here, L is the total number of multipath components (MPC), and γ _l and τ _l are the amplitude and delay of the l-th MPC, respectively. The TOA of the received signal is given by τ _toa = τ ₁ (ie the arrival time for the first arrival path). Hypotheses H ₀ and H ₁ representing the LOS hypothesis and NLOS hypothesis, respectively, are given by:

Here, d represents the distance between the FT and the mobile terminal, and c represents the speed of light. In the NLOS state, even when the first arrival path is correctly identified, the TOA estimate still yields a large value compared to the actual distance. Therefore, to avoid positioning performance degradation, NLOS FTs are identified and then the effects of those NLOS FTs are reduced.

  The present invention provides a method for distinguishing LOS and NLOS conditions using received multipath component statistics. In one embodiment, statistical data that captures the amplitude and delay of the received signal (ie, kurtosis, average excess delay, and RMS delay spread) is used to distinguish between LOS and NLOS conditions.

  Kurtosis is the ratio of the fourth rank moment of a random variable to the square (ie, variance) of the second rank moment of the random variable. Since kurtosis characterizes how much data is taken, kurtosis also characterizes how strong the LOS state is in multipath CIR. A high kurtosis value for the CIR suggests that the received signal is likely coming from a LOS source.

ここで、μ_｜ｈ｜及びσ_｜ｈ｜は、それぞれＣＩＲの絶対値の平均及び標準偏差である。κの分布は、標本通信路実現を使用してＬＯＳ及びＮＬＯＳ状態の双方について取得され得る。例えば、ＩＥＥＥ ８０２．１５．４ａ通信路は、住居内ＬＯＳ及びＮＬＯＳ状態、事務所内ＬＯＳ及びＮＬＯＳ状態、屋外ＬＯＳ及びＮＬＯＳ状態、産業ＬＯＳ及びＮＬＯＳ状態に対応する８つの異なる通信路モデルＣＭ１〜ＣＭ８についてκのヒストグラムを提供する。ヒストグラムの各々は、次式によって与えられる対数正規ＰＤＦによってモデル化されてもよい。

ここで、μ_ｋはｐ（ｋ）の平均であり、σ_ｋは標準偏差である。このモデルは、ＩＥＥＥ ８０２．１５．４ａ通信路上における５％有意性レベルのコルモゴルフ・スミルノフ （ＫＳ）適合度仮説検定を使用して正当化される。 For a channel with CIR h (t), the kurtosis of | h (t) | is given by

Here, μ _{| h |} and σ _{| h |} are the average and standard deviation of absolute values of CIR, respectively. The distribution of κ can be obtained for both LOS and NLOS conditions using a sample channel realization. For example, the IEEE 802.15.4a channel is for eight different channel models CM1-CM8 corresponding to residential LOS and NLOS conditions, office LOS and NLOS conditions, outdoor LOS and NLOS conditions, industrial LOS and NLOS conditions. Provides a histogram of κ. Each of the histograms may be modeled by a log normal PDF given by:

Here, μ _k is an average of p (k), and σ _k is a standard deviation. This model is justified using the 5% significance level Kolmogolf Smirnov (KS) goodness-of-fit hypothesis test over the IEEE 802.15.4a channel.

ＲＭＳ遅延の広がりτ_ｒｍｓは次式によって与えられる。

The kurtosis provides information about the amplitude statistics of the received MPC, while the delay statistics for multipath components are provided by the average excess delay and the spread of the RMS delay. “Wireless Communications: Principles and Practice”, T.M. S. According to Rappaport (author), Indianapolis, IN: Prentice Hall, 2002, the average excess delay τ _m of the channel is given by:

The RMS delay spread τ _rms is given by:

  Similar to the kurtosis analysis described above, the histogram of mean excess delay and RMS delay spread for eight different channel models from IEEE 802.15.4a is based on the 5% significance level Kolmogolf Smirnov test, Justify the lognormal distribution of delays in the received signal.

（２）平均過剰遅延検定：

（３）ＲＭＳ遅延広がり検定：

If statistical a priori knowledge of κ, τ _m , and τ _rms is available under the LOS and NLOS conditions, a likelihood ratio test can be set up to distinguish the LOS and NLOS hypotheses. For example, P _los (x) and P _nlos (x) represent PDFs corresponding to the LOS and NLOS states, respectively, and κ, τ _m , and τ _rms are the kurtosis for the observed channel realization h (t), Given the average excess delay and RMS delay spread, each of the following three likelihood ratio tests can be used to identify the LOS / NLOS condition.
(1) Kurtosis test:

(2) Mean excess delay test:

(3) RMS delay spread test:

In each test, the LOS hypothesis (H ₀ ) is selected if the likelihood ratio is greater than 1, otherwise the NLOS hypothesis (H ₁ ) is selected. Taking all three parameters into account and deriving a combined PDF from the individual PDFs of these parameters, a test of the form:

ここで、

However, the combined PDF is difficult to derive. Assuming that the parameters κ, τ _m , and τ _rms are obtained independently, one approximation may be obtained.

here,

  This metric from the least squares algorithm may be used to weight the reliability of each FT.

  To improve positioning accuracy, NLOS condition identification may be used in many ways. For example, the NLOS FT may be excluded from the calculation of the mobile station location estimate. If the number of FTs available for location estimation is small, it may be difficult to exclude the NLOS FT. Furthermore, Venkatesh et al. (Described above) teach that the information in the NLOS FT can be used, particularly for high precision geometric dilution (GDOP) geometry, to provide better positioning accuracy. More specifically, when more than two LOS FTs are approximately placed along a line, calculating an estimate of the position including additional FTs off that line is Improve positioning accuracy even if FT is NLOS.

ここで、β_ｉは、ｉ番目のＦＴで受信された信号の信頼度を反映し、

は移動端末位置の推定値であり、ｘ_ｉ＝［ｘ_ｉ ｙ_ｉ］はｉ番目のＦＴの既知の位置であり、ｄ_ｉは移動端末とｉ番目のＦＴとの間の測定された距離である。ｄ_ｉの１つのモデルは、次式によって与えられる。
ｄ_ｉ＝ｒ_ｉ＋ｂ_ｉ＋ｎ_ｉ
ここで、ｒ_ｉは移動端末とｉ番目のＦＴとの間の実際の距離であり、ｎ_ｉ〜Ｎ（０，σ^２）は、分散σ^２を有する加法的白色ガウス雑音（ＡＷＧＮ）であり、ｂ_ｉは次式によって与えられる非負ＮＬＯＳバイアスである。

ここで、Ψ（μ_ψ）は平均μ_ψを有する指数分布を表す。 For N FTs, J. J. et al. Cafery and G.C. L. According to Stuber, IEEE Commun. Mag. , Vol. 36, no. 4, pp. 38-45, Apr. The article “Overview of radiolocation in CDMA cellular systems” published in 1998 teaches a weighted least squares solution.

Here, β _i reflects the reliability of the signal received at the i-th FT,

Is an estimate of the mobile terminal position, x _i = [x _i y _i ] is the known position of the i th FT, and d _i is the measured distance between the mobile terminal and the i th FT. is there. One model of d _i is given by:
d _i = r _i + b _i + n _i
Where r _i is the actual distance between the mobile terminal and the i-th FT, and n _{i to} N (0, σ ² ) are additive white Gaussian noise (AWGN) with variance σ ² . , B _i is a non-negative NLOS bias given by:

Here, Ψ (μ _ψ ) represents an exponential distribution having an average μ _ψ .

  In the above-mentioned Cafery et al., The inverse of the measured distance variance is used as a confidence metric for the i th FT. However, for stationary terminals, the dispersion of TOA measurements is not significantly different to allow the distinction of LOS FT from NLOS FT.

According to one embodiment of the present invention, performance superior to the Caffery et al. Approach is obtained using the following confidence metric.
β _i = log ₁₀ (1 + J (k, τ _m , τ _rms ))
This penalizes the NLOS FT, typically by assigning a weight between 0 and 1 to the FT. Such an approach may be referred to as soft weighted selection (SWS). The disadvantage of such an approach is that the dynamic range of the weight values is very large for the LOS node. A large dynamic range of weights unnecessarily favors some of the LOS measurements with respect to other LOS measurements, thereby degrading positioning accuracy in some cases.

ここで、ｋ_１及びｋ_２は、ＮＬＯＳ効果を抑圧するように適宜選択された２つの加重値であり、これにより、識別されたＮＬＯＳ ＦＴが、限定されたインパクトをＷＬＳ解の上に有する。ｋ_１＝０及びｋ_２＝１の場合、ＮＬＯＳ ＦＴの寄与は廃棄される。これは、本明細書で識別及び廃棄（ＩＡＤ）と呼ばれる手法である。ＩＡＤは誤識別の危険を冒す（即ち、ＬＯＳ ＦＴをＮＬＯＳ ＦＴと間違えるか、この反対）。こうして、或る場合には、移動端末の場所を推定するには、識別されたＬＯＳ ＦＴの数が不十分であるかも知れず、それによって、位置決め正確度を可能性として相当に劣化させる。 SWS performance may be improved by assigning fixed weights to LOS and NLOS measurements, ie, using hard weighted selection (HWS). In the HWS approach, β i can be set as follows:

Here, k ₁ and k ₂ are two weights appropriately selected to suppress the NLOS effect, so that the identified NLOS FT has a limited impact on the WLS solution. If k ₁ = 0 and k ₂ = 1, the NLOS FT contribution is discarded. This is a technique referred to herein as identification and discard (IAD). The IAD takes the risk of misidentification (ie, mistaken LOS FT for NLOS FT or vice versa). Thus, in some cases, the number of identified LOS FTs may be insufficient to estimate the location of the mobile terminal, thereby potentially significantly degrading positioning accuracy.

及び

ｒは、選択されたＦＴの索引である（即ち（ｘ_ｒ，ｙ_ｒ）は選択されたＦＴの位置ｘ_ｒである）。Ｖ．Ｄｉｚｄａｒｅｖｉｃ 及びＫ．ＷｉｔｒｉｓａｌによってＰｒｏｃ．Ｗｏｒｋｓｈｏｐ ｏｎ Ｐｏｓｉｔｉｏｎｉｎｇ，Ｎａｖｉｇａｔｉｏｎ，ａｎｄ Ｃｏｍｍｕｎ．（ＷＰＮＣ），Ｈａｎｎｏｖｅｒ，Ｇｅｒｍａｎｙ，Ｍａｒ．２００６，ｐｐ．１２９−１３８で発表された論文“Ｏｎ ｉｍｐａｃｔ ｏｆ ｔｏｐｏｌｏｇｙ ａｎｄ ｃｏｓｔ ｆｕｎｃｔｉｏｎ ｏｎ ＬＳＥ ｐｏｓｉｔｉｏｎ ｄｅｔｅｒｍｉｎａｔｉｏｎ ｉｎ ｗｉｒｅｌｅｓｓ ｎｅｔｗｏｒｋｓ”で検討されるように、選択されたＦＴ ｒに関する線形化は費用関数を最小にする。即ち、

Minimizing the equation in the nonlinear cost function described above requires a numerical search method, such as the steepest descent method or the Gauss-Newton method. This is computationally expensive and requires good initialization to avoid convergence to a local minimum in the cost function. (See, for example, “Mobile positioning used fitness networks: Possibilities: pos. base "on wireless network network measurements"). As an alternative, the cost function may be linearized with respect to the location of the selected FT using the method disclosed in Venkatesh above. In this method, the contribution of the selected FT to the terminal position x is separated from the other contributions to yield:
Ax = p
This has a least squares solution given by:
x = (A ^T A) ⁻¹ A ^T p
here,

as well as

r is the index of the selected FT (i.e. _(x r, _{y r)} is the position _{x r} of the selected FT). V. Dizdalevic and K.M. Witrisal, Proc. Works on Positioning, Navigation, and Commun. (WPNC), Hanover, Germany, Mar. 2006, pp. As discussed in the paper “On impact of topology and cost function on LSE position determination in wireless networks” published in 129-138, the linearization on the selected FT r minimizes the cost function. That is,

As in the previously discussed nonlinear model, the relative confidence of the i th FT may be characterized by weighting the i th term in the cost function by β _i . N-1 is constructed by N-1 diagonal matrix W = diag (β ₁ , β ₂ ,..., Β _N-1 ), and A _W = WA and p _W = WP are obtained, thereby obtaining a mobile terminal. The weighted location estimate x of is obtained using a linear model.
^{_{x = (A W T A W}} ) -1 A W T p W

  Thus, the least squares solution of the result to the mobile terminal location x uses the likelihood function obtained from the multipath component of the received signal to suppress the effect of the NLOS FT.

  FIG. 1A is a diagram illustrating a communication system in which arrival time (TOA) estimation and wireless positioning operations may be performed based on signals received at different FTs in an NLOS environment. As shown in FIG. 1, each of the FTs 10, 20, and 30 measures the TOA for each signal received from the mobile terminal 5. The TOA is forwarded to the central processing unit 35, and the location of the terminal 5 is estimated by triangulation. As an alternative, the terminal 5 may estimate its location using the measured value of the received signal at the FT.

  FIG. 1B illustrates a TOA estimation operation based on signals received at FTs 10, 20, and 30. FIG. Typically, each receiver locks on the strongest path to measure TOA. In FIG. 1B, the strongest path of each of FTs 10, 20, and 30 is denoted by reference numeral 9. From the identified strongest path, each receiver searches backward in time for the first arrival path. In the LOS state, the first arrival path (indicated by reference numeral 11 in FIG. 1B) corresponds to the shortest distance between the transmitter and the receiver. However, in the NLOS condition (ie, there is an obstruction between the transmitter and the receiver), the first arrival path indicated by reference numeral 7 in FIG. 1B is later than the first arrival path 11 of the LOS. Arrive. Thus, the NLOS arrival path introduces a positive bias into the TOA estimate even when the first arrival path is correctly identified. In addition, the receiver typically sets a threshold (indicated by reference numeral 8) that is used to qualify the first arrival path. When the NLOS first arrival path 7 has a signal strength that is less than the threshold, the estimated first arrival path (indicated by reference numeral 12) has a more delayed value.

ここで、ｄ_ｉはｉ番目のＦＴと移動端末との間の距離であり、ｘ_ｉは利用可能なＮ個のＦＴについてｉ番目のＦＴの場所である。推定された端末場所の平均二乗剰余誤差（「剰余」）は、次のように書くことができる。

In conventional systems, the TOA of the received signal is estimated at each FT using a ranging algorithm (eg, a threshold-based search technique using arbitrary thresholds). The TOA estimate is converted into distance estimates 31, 32, and 33 (see, eg, FIG. 1A). For example, an estimate of the location of the mobile terminal is provided by the least square method. The least square method selects the value of x that minimizes the sum of the squares of all the remainders as follows:

Here, d _i is the distance between the i th FT and the mobile terminal, and x _i is the location of the i th FT for the N available FTs. The estimated mean square residue error (“residue”) of the terminal location can be written as:

  In the LOS state, the remainder depends only on measurement noise and backward search error. The backward search error results from an incorrect identification of the first arrival path. Therefore, in the LOS state, an accurate estimate of the TOA corresponding to each FT is relatively easy to obtain. As a result, the estimated mobile terminal location is close to the actual mobile terminal location, and the residual error is generally small assuming that sufficient averaging reduces noise variance. However, when there is one or more NLOS FTs, the remainder is quite large due to the introduced NLOS bias. As mentioned above, NLOS bias is due to two reasons: (1) the delay between LOS TOA and NLOS TOA, and (2) the delay between the estimated NLOS TOA and the actual NLOS TOA. .

  Since the first type of bias (ie, the bias as a result of the difference between LOS TOA and NLOS TOA) cannot be handled directly in the backward search step, such bias is addressed in the triangulation step. The Basically, the LOS or NLOS information of the channel can be obtained from the multipath received signal (eg in the form of a likelihood weighted value). This information can be used in the triangulation step to reduce the effect of the NLOS FT.

  FIG. 2 is a flow diagram illustrating a weighted least squares (WLS) positioning algorithm 200 (or alternatively, a residue-based least squares positioning algorithm 210) according to one embodiment of the present invention. As shown in FIG. 2, the TOA estimate 100 is provided directly to the weighted LS algorithm 200 (or alternatively, the residue-based weighted LS algorithm 210) along with the likelihood function 150. Deriving weights with least squares positioning algorithm as discussed in the related application “Line-of-Sight (LOS) or non-LOS (NLOS) Identification Method Multipath Channel Statistics” incorporated by reference above. Likelihood function 150 may be used to distinguish between LOS FT and NLOS FT.

  The LOS weight obtained from the likelihood function may be further used to improve the performance of other algorithms. For example, the above-described P.I. C. The residue-based algorithm by Chen may be improved by assigning LOS weights to individual residues in the FT, as shown at 210. In this improved method, when calculating the residual error corresponding to different combinations of FTs, the error corresponding to each observation is further weighted by the LOS weight. By using the information in the MPC of the received signal to characterize the reliability of measurements from different FTs, the LOS weight according to the method of the present invention can be used with many other different positioning algorithms.

FIG. 3 is a diagram illustrating the WLS positioning algorithm 200 shown in FIG. 2 in more detail, according to one embodiment of the present invention. As shown in FIG. 3, the arrival times measured by FTs 201, 202, and 203 are passed to one of the least square algorithms of FIG. In these algorithms, distance measurements d _i between the mobile terminal and each FT are calculated. In one embodiment, the location of the mobile terminal is selected such that a weighted cost function (eg, the weighted cost function described above) is minimized. As mentioned above, the least squares model can be linear (thus producing a closed-form solution) or non-linear (ie a significant portion of the mobile terminal location needs to be searched if not all). May be.

  Unlike prior art received signal based NLOS mitigation methods, which typically require the recording of TOA (or distance) measurements over time, the method of the present invention does not require measurement time history. As long as the LOS / NLOS likelihood PDF is available, NLOS mitigation can be performed as small as a single channel realization from each FT. This is because TOA fluctuations are not taken into account. Instead, the NLOS information in the received multipath component is used.

  Furthermore, existing conventional algorithms favor mobile terminals. The NLOS measurement bias provides sufficient variation to distinguish NLOS measurements from LOS measurements. However, in the case of a stationary terminal, the NLOS bias may not show sufficient variation, thus making NLOS FT identification and mitigation difficult. However, the method of the present invention uses information embedded in the MPC of the received signal for NLOS mitigation and is therefore also effective for stationary terminals.

  The method of the present invention may also be used to improve positioning accuracy. When a sufficient number of FTs are available, the NLOS FT is discarded, preventing biasing of the mobile terminal's location estimate. The likelihood function of the LOS FT can be weighted in the LS positioning algorithm described above, giving less confidence to the NLOS measurement.

  The above detailed description is provided to illustrate specific embodiments of the present invention and is not intended to be limiting. Many modifications and variations are possible within the scope of the present invention. The invention is set forth in the appended claims.

FIG. 10 illustrates a communication system in which TOA estimation and wireless positioning operations may be performed based on signals received at different FTs in an NLOS environment. It is a figure which shows the TOA estimation operation | movement based on the signal received by FT10,20,30. 6 is a flow diagram illustrating a weighted least squares (WLS) positioning algorithm 200 (or alternatively, a residue-based least squares positioning algorithm 210), according to one embodiment of the present invention. FIG. 3 shows in more detail the WLS positioning algorithm 200 shown in FIG. 2 according to one embodiment of the invention.

Claims (1)

A mobile terminal positioning method comprising:
Receiving signals transmitted between the mobile terminal and a plurality of fixed terminals;
For each received signal, using statistical metrics to determine the likelihood that said received signal has traveled the line-of-sight path;
Determining the position of the mobile terminal based on the arrival time of the received signal using a positioning algorithm based in part on the likelihood of the line-of-sight path of the received signal.

Images