This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

**Non-Line-of-Sight Identiﬁcation via Space-Frequency Correlation in MIMO-OFDM Systems: A Preliminary Study
**

Wenjie Xu

Electrical and Computer Engineering Michigan Technology University Houghton, Michigan 49931 Email: wenjiex@mtu.edu

**Seyed A. (Reza) Zekavat
**

Electrical and Computer Engineering Michigan Technology University Houghton, Michigan 49931 Email: rezaz@mtu.edu

Abstract—This paper presents a preliminary study on a nonline-of-sight (NLOS) identiﬁcation technique that exploits spacefrequency correlation features of multiple-output orthogonal frequency division multiplexing (OFDM) systems. Here, space correlation refers to the correlation across antenna elements and frequency correlation refers to the correlation across subcarriers. In this technique, a measure, subcarrier correlation difference (SCD), is introduced and used to identify NLOS conditions. SCD is deﬁned as the difference of cross-antenna subcarrier correlation and single antenna subcarrier correlation. The impact of the fading channel K-factor, spatial subchannel correlation and channel estimation error on the proposed techniques is investigated.

based on spatial correlation [7] and the method based on phase difference across antennas [8]. This paper investigates NLOS identiﬁcation problem of multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM) systems. OFDM is a bandwidthefﬁcient signaling scheme for wideband wireless digital communication [9], and it converts a frequency-selective fading channel into a set of parallel ﬂat-fading channels (referred to as subcarriers) to reduce inter-symbol interference. MIMO systems achieve high throughput and reliability by deploying multiple antennas at both receiver and transmitter [10]. MIMO and OFDM are the two key technologies for future high throughput communication. MIMO-OFDM has already been adopted as standard for wireless LAN 802.11n [11] and WiMAX 802.16e [12]. We propose a new NLOS identiﬁcation technique exploiting space-frequency channel correlation. The channel matrix of MIMO-OFDM systems is three dimensional, where two dimensions represent the MIMO spatial sub-channels, the third dimension represents the sub-carriers of each sub-channel. Thus, the channel correlation can be across antenna (space) and/or across subcarrier (frequency). It is depicted that the difference of cross-antenna subcarrier correlation and single antenna subcarrier correlation, which is called subcarrier correlation difference (SCD) can be used to identify NLOS. The identiﬁcation is based on the following observation: when the variation of spatial correlation across different multi-path component is minimal, SCD under LOS conditions would be much larger than that under NLOS condition. This technique only requires the knowledge of channel matrices of all subcarriers which are already available in most MIMO-OFDM systems. Therefore, it can be easily implemented on current systems. Section II describes the space-frequency channel correlation model of MIMO-OFDM systems. Section III introduces the metric SCD that is used for NLOS identiﬁcation. Section IV presents the performance of NLOS identiﬁcation, and Section V concludes the paper.

I. I NTRODUCTION Localization techniques have attracted a lot of research interests recently, because they have a wide range of military and civilian applications in wireless mobile networks, such as battleﬁeld command and control, patients monitor, and emergency 911 (E911) [1]. Localization techniques that are based on time-of-arrival (TOA) [2] and direction-of-arrival (DOA) [3] are very sensitive to the availability of line-ofsight (LOS) which is the direct path between the transmitter and the receiver. If LOS is not available, i.e., in non-LOS (NLOS) propagation conditions, the received signal travels longer distances compared to the LOS path. This results in wrong TOA and DOA estimations. Therefore, a large localization error is experienced. In order to reduce NLOS localization error, NLOS identiﬁcation is required. One NLOS identiﬁcation approach is based on the statistics of the estimated range. This range is Gaussian distributed with a small variance if LOS is available, and it is non-Gaussian or exponentially distributed with a large variance in NLOS situations [4]. This method takes a long processing time to obtain the range statistics. Many other proposed identiﬁcation techniques incorporate the distributions of channel parameters such as the received signal power, delay spread [5], and Rician K-factor [6]. Those techniques are proposed for single antenna systems. The potential of multiple receive antennas based NLOS identiﬁcation is also studied, such as the method

978-1-4244-8179-8/10/$26.00 ©2010 IEEE

400

.q2 exp[−j2π∆f (n1 l=1 l χl L 2 p1 . 2. Next. 1.q1 . φ0 is the direction-of-arrival (DOA).q 2 2 l = E(|al | ).q (n). we investigate the ˆ expressions of subcarrier correlation using H p.q (n) is · exp[−j2π∆f (n1 − n2)τLOS ]+ L 2 p1 .q p1 . L where dt . Let the signal bandwidth be W and the total number of subcarriers is N .p2 . the ˆ estimated channel frequency response H p.q (n) in (4) with H p.q (n) (sub-channel from transmit antenna p to receive antenna q) instead of the ture one. nth subcarrier is 2 RLOS (p1 . When LOS is available. θ0 is the direction-ofdeparture (DOD). When LOS is not available. Replacing H p. "0 !0 hp. we ﬁnd out the channel response in the frequency domain.p2 .q2 )∗ }/ 2 . n2 ) = L 2 p1 . . Thus. that corresponds to RN LOS (p1 .q (t) = ap. q2 . respectively.q (n) + c∗ wp. the channel frequency response of nth subcarrier is obtained.q (n).p2 .q2 (n ))∗ ] = σ 2 δ(p − p )δ(q − q )δ(n − 1 2 1 2 1 2 1 1 x=0 Assuming uniform antenna array and uncorrelated scattering.q H (n) can be estimated using the pilot signal known at both the transmitter and the receiver [13]. n1 . E{ap1 . the multi-path channel impulse response (CIR) from pth (p = 1.q2 l=1 l χl − n2 )τl ] (6) (2) Next. ap.p2 . Usually. n1 .q exp(−j2πn∆f τLOS ) LOS L + l=1 ap. q2 . q2 . n = 0.q1 (n )(wp2 .e. 1. Note that τLOS in (5) can always be l normalized to 0 without loss of generality.q = E{HLOS1 (n1 )[HLOS2 (n2 )]∗ }/E{HLOS1 (n1 )[HLOS2 (n1 )]∗ } where wp.q (n)|2 ]/σ 2 = {|aLOS |2 exp[−j2π(p1 − p2 )dt sin θ0 − j2π(q1 − q2 )dr sin φ0 ] The maximum likelihood estimation of H p.q is the amplitude coefﬁcient of the LOS path.q1 . Let 2 be the power of lth tap and l p. 1 and shows that for the special case of parallel transmit and receive antenna arrays θ0 = −φ0 .q is a LOS l complex number and |ap. n2 ) Finally. Mt ) transmit antenna to q th (q = 1. Mr ) receive antenna is represented by a tappeddelay-line model.q (n) + wp. and τl is the associated TOA.q1 . . . N . p2 . S PACE .q p2 . which means that the average power of each tap is the same for any p and q. We consider a wideband MIMO channel with Mt and Mr transmit and receive antenna elements. The signal noise ratio (SNR) of 0 x=0 i. q1 receive antenna. and p χl 1 .q1 (al 2 .q δ(t − τLOS ) + LOS LOS l=1 ap. the subcarrier correlation is obtained by letting aLOS = 0 in (5). .p2 .q δ(t − τl ) l (1) where ap.q is a zero mean Gaussian process. is used to compute the LOS and NLOS subcarrier correlation given in (5) and (6).q (n) (7) p1 .q exp(−j2πn∆f τl ) l (3) Then. The received pilot signal corresponds to Y p. q1 . n2 ) SN R = E[|H p.q (n) c∗ wp. n2 ). q1 . .q (n) = H p.p2 .q1 . p2 . ap. which is also Gaussian. we present a MIMO-OFDM channel model and derive a general formula of the cross-space-crossfrequency correlation. The pilot signal is represented by a sequence {cn } and |cn |2 = 1.q |2 = |aLOS |2 . p. ..q2 is the spatial correlation coefﬁcient.II.q |2 LOS E(|ap. n1 . the frequency spacing is deﬁned as ∆f W/N Applying Fourier transform to (1).q p2 .q2 )∗ } = 0. we studies how does the channel estimation error impact the channel correlation. p2 .q is l the amplitude coefﬁcient of the lth multi-path component.q (n) = cn H p. the Rician K-factor LOS of the ﬁrst path is deﬁned as K |ap. In real OFDM systems. where n is the subcarrier identiﬁer. p. Assuming ap. the normalized channel correlation of pth transmit 1 th antenna. nth subcarrier and pth transmit 1 2 th antenna. χp1 .q2 = l p p E{al 1 .q HLOS (n) = ap.q (n) = c∗ Y p.FREQUENCY CHANNEL CORRELATION MODEL OF MIMO-OFDM SYSTEMS In this section.q (n). when l = k.q2 l χl l=1 L (8) exp[−j2π∆f (n1 − n2 )τl ]}/{ l=1 2 p1 . q2 receive antenna.q (n) 401 .q (n) is the additive Gaussian noise.q1 .q (n) n n (9) |aLOS |2 exp[−j2π(p1 − p2 )dt sin θ0 − j2π(q1 − q2 )dr sin φ0 ]} Let up. q1 . the above correlal k the received pilot signal is tion becomes RLOS (p1 . . dr are the transmit and receive antenna spacing normalized to the wavelength.q1 . τ1 = τLOS . identically n ˆ (5) distributed as wp. 2. The relationship of DOA and DOD of the direct path is sketched in Fig. . and (4) E[wp1 . δ(x) = . the ﬁrst tap contains the LOS ray and some NLOS rays. LOS τLOS is the time of arrival (TOA) of the LOS path.q1 (ap2 . . The DOA and DOD for LOS propagation.q2 + l χl ˆ H p.q |2 ) 1 = |aLOS | 2 1 2 Fig.

q1 .q2 remains approximately the same for all multipath components. The distril butions of SSCDLOS and SSCDN LOS are shown in Fig. It is observed in Fig. such as p1 = 1. q1 ..p2 . deﬁned in (11) are nonzeros for different (n1 −n2 ). The sum of SCDs corresponds to N/2 SSCDLOS = | n1 −n2 =0 SCDLOS (n1 − n2 )| (15) 402 .q2 is deterl mined by the DOA distribution of the lth multipath component [14] and it has been demonstrated in [15].q ˆ p1 .p2 . there is a clear distinction between LOS and NLOS results. and (p1 − p2 )dt = (q1 − q2 )dr . q2 = 2. 2. and show that their difference can be employed in NLOS discrimination. φ0 .q1 . q1 . q2 . We assume that the orientations are uniformly distributed between [0. It is seen that when θ0 = −φ0 (see Fig.q1 p2 . al and bl (real and imaginary part of χp1 . q2 . SCDs of different subcarrier pairs can be added to obtain a better identiﬁcation performance. 3GPP broadband channel model [17] shows that the power angular spectrums for all multipath components are the same. deﬁned in (12) remains approximately 0 for all (n1 − n2 ). p2 = 2. q1 . i. p1 .p2 .7].q2 = al + jbl .q2 ∼ U [−0.q p2 . −0. 6dB is better than the performance of K = 0dB. the performance of K = 3. Additionally. the maximum TOA τmax = 1000 ns. q1 = q2 .q2 is illustrated in the next section. dt = dr . n1 . ∆f = W/N = 78. al and bl are both assumed to be uniformly distributed in the range [0. dt = dr = λ/2. The bandwidth W = 20 MHz. 2(b). as shown in Fig. The impact of Rician K factor Let p1 = 1. When p1 = p2 . i. SCD under LOS conditions.q1 . When the transmit and receive antenna arrays are parallel.e.125 kHz. χp1 .and using (9). Therefore. III. including DOD θ0 . which indicates that NLOS identiﬁcation performance is worse than that in Fig.q = E{HLOS1 (n1 )[HLOS2 (n2 )]∗ }/E{HLOS1 (n1 )[HLOS2 (n1 )]∗ } How (15) and (16) can be used to identify the NLOS scenarios p1 . q2 .e. which maintains a good NLOS identiﬁcation performance. but SCD under NLOS conditions. (12) becomes SCDN LOS (n1 − n2 ) ≈ 0 (14) Hence. θ0 and φ0 impact the phase of LOS part in (5) via the term exp[−j2π(p1 − p2 )dt sin θ0 − j2π(q1 − q2 )dr sin φ0 ].11 wireless LAN (WLAN) channel models [11]. q2 = 2. In addition. Now.e. 0. n2 ) The subcarrier correlation difference for NLOS is similarly deﬁned as SCDN LOS (n1 − n2 ) RN LOS (p1 . χp1 .3.7]. q1 = 1. The delay spread τrms = 150 ns. p2 .q1 . n2 ) − RN LOS (p1 .q1 . i.q2 l (13) This is because the spatial correlation χp1 . p2 = 2. 2π]. set p1 = p2 = 1.p2 . Parameters are selected based on 802. N UMERICAL SIMULATIONS AND DISCUSSIONS We set up simulations to plot the distributions of SSCDN LOS and SSCDLOS . n1 . such as p1 = p2 = 1. = {E{H (n1 )[H (n2 )]∗ } LOS LOS SSCDN LOS = | + σ 2 δ(p1 − p2 )δ(q1 − q2 )δ(n1 − n2 )}/ p1 .e.q ˆ p2 . n2 ) = − L 2 p1 . 2(b) that SSCD distributions of LOS and NLOS overlap. and the tap power is assumed to decay exponentially..p2 . 2π].q2 ≈ χp1 . −0.p2 . 3dB and 6dB.7] or [−0. Note that χp1 .p2 . n2 ) (11) − RLOS (p1 . 2 = exp(−τl /τrms ). q2 = 2 and let us study the single transmit antenna case where the inﬂuence of θ0 is which is seen that the channel estimation error impacts the correlation only when p1 = p2 . 2(a).7] + jU [−0. l Some parameters on which the correlation (5) depends have not been speciﬁed yet. The SSCDLOS distributions are generated for K = 0dB. −0.q1 ..e.q2 is a l p complex number and let χl 1 . Fig.q1 .3.p2 .p2 . i.q1 . q1 . i..q1 . p2 . the antenna array orientations of the transmitter and the receiver are arbitrary. n1 . This observation is helpful in identifying NLOS conditions.q2 l=1 l χl L 2 l=1 l exp[−j2π∆f (n1 − n2 )τl L 2 l=1 l − n2 )τl ] (12) p It is noted that χl 1 . the correlations becomes N/2 SCDN LOS (n1 − n2 )| (16) n1 −n2 =0 RLOS (p1 . Based on the measurements given in [17]. θ0 ∼ U [0. 1.q2 exp[−j2π∆f (n1 l=1 l χl L 2 p1 . q1 = 1.p2 . DOA φ0 and spatial p correlation coefﬁcient χl 1 . the number of subcarrier N = 256. the phase [−j2π(p1 − p2 )dt sin θ0 − j2π(q1 − q2 )dr sin φ0 ] = 0 and the correlation is independent of θ0 and φ0 . the subcarrier correlation difference for LOS is deﬁned as SCDLOS (n1 − n2 ) RLOS (p1 .q1 . the term θ0 vanishes in (5) and the correlation does not depend on θ0 . θ0 and φ0 are independent. In addition. −0. q2 = 2. p1 . n2 ) ˆ p1 . In Fig. S UBCARRIER CORRELATION DIFFERENCE Based on the study on the channel correlation in Section II.q {E{HLOS1 (n1 )[HLOS2 (n1 )]∗ } + σ 2 δ(p1 − p2 )δ(q1 − q2 )} (10) IV. Moreover.q2 . q1 .q1 . q1 = 1.q ˆ p2 . 2(b) shows that the performance is not sensitive to the value of K.3.3. n1 . A. WLAN Channel Model F which is suitable for the indoor or outdoor large open space and is used here to characterize the power delay proﬁle.q2 ) l are uniformly distributed in the range [−0.3.. 2(a)). q1 . p2 .q1 . q1 = 1. n1 .7].p2 . [16] that DOAs of different multipath components are similarly distributed for uniform and Gaussian scatterer distribution models.

It is observed that the NLOS and LOS results have some overlapped components. B. K = 3 dB LOS.p2 . K = 0 dB LOS. K = 3dB. p2 = 2. However. 0.q1 .7] removed. −0. The impact of χp1 .3. The actual values of SSCDLOS may exceed 150 and those counts are included in the counts corresponding to 150. 2π] (b) Fig.7] + l jU [0.q2 ∼ U [−0.q1 . As K increases. and θ0 = −φ0 simulation assumptions of Fig.2500 NLOS LOS.3. χl 1 . the improvement is not very signiﬁcant. The distributions of SSCDLOS and SSCDN LOS when p1 = 1. −0.3.7] (b) χp1 . q1 = 1. Here. K = 6 dB Histogram count 1.7]. q2 = 2.5 x 10 4 2 2 NLOS LOS. the impact of the spatial correlation coefﬁcient is discussed. K = 3 dB LOS.7].q1 . which indicates a degraded identiﬁcation performance.q1 .q1 . the channel correlation is impacted by the channel estimation error.3. the overlapping area slightly decreases. K = 0 dB LOS. The mean and standard deviation (STD) of SSCD under LOS and 403 .p2 . K = 6 dB 2. q1 = 1. because the overlapped area decreases. 3.q1 . θ0 = −φ0 ). C.p2 . the performance in Fig. Mean and STD of SSCDLOS and SSCDN LOS when p1 = p 1. the impact of the channel estimation error is studied. and (a) χp1 . p2 = 2. The impact of channel estimation error As discussed in Section II. 3(b) l shows the results when they are both positive.p2 .q2 are negative. q2 = 2. It is found that compared to Fig. which means the performance slightly increases. 3(a). −0. K = 3 dB LOS. φ0 . 0. K = 3 dB LOS. 2 (K = 3dB.5 x 10 4 SSCD (a) NLOS LOS.q2 ∼ U [−0.p2 .p2 . and (a) θ0 = −φ0 (b) φ0 and θ0 are independent. φ0 is assumed to be uniformly distributed between [0.q2 ∼ U [0. 2π]. K = 6 dB 2. based on the 120 100 80 SSCD 60 mean for LOS mean for NLOS 40 20 0 0 5 10 15 20 25 30 SNR(dB) Fig.3.7] + l jU [−0.7] + jU [−0.3. φ0 ∼ U [0.5 1 1 0.7] + jU [−0. 3(a) shows the results when the real and imaginary part of χp1 . K = 6 dB Histogram count 1500 Histogram count 50 100 150 1.5 1000 1 500 0.5 0 0 SSCD 0 0 50 100 150 (a) 2. Fig. −0.q2 ∼ l U [−0. −0. The distributions of SSCDLOS and SSCDN LOS when p1 = p2 = 1. −0. 2π]. 3(b) improves. K = 0 dB LOS.3. 3(a). K = 0 dB LOS. 2.5 0.5 x 10 4 2000 2 NLOS LOS.3. θ0 ∼ U [0. q2 = 2.5 Histogram count 0 50 100 150 1. q1 = 1. 4.q2 l Next.5 0 0 SSCD 0 50 100 150 SSCD (b) Fig. χp1 . while Fig. The distributions of SSCDLOS and SSCDN LOS are depicted in Fig.

It is found that as K increases. no. AES12. the channel estimation error does not signiﬁcantly degrade the identiﬁcation performance. In this work. the antenna spacing and transmit and receive antenna index are subject to equation (p1 −p2 )dt = (q1 −q2 )dr .3gpp. no.q2 and the channel l estimation error are minimal in the proposed technique.p2 . Zhong and D.p2 . IEEE Transactions on.” Signal Processing. the polarity of χp1 . Zekavat. pp. p clustering arrivals may be considered. [3] X. [14] A. vol. may. Wu. Ye Li. 6.” IEEE Std 802. 2009. Buehrer. Hwang. where the deﬁnition of SNR is given in (8) and the vertical line corresponds to the plus and minus one STD around the mean of SSCD. dec 2001. Then. [11] E. pp. Apr.NLOS conditions for different error variance levels are shown in Fig. CONCLUSION This work proposes a metric called spatial correlation difference (SCD) for MIMO-OFDM systems to implement NLOS identiﬁcation. Then hypothesis testing can be used to identify NLOS: ﬁrstly determine the SSCD distributions under LOS and NLOS conditions. 187 –194. In general.htm [18] S.16-2004). 404 . e. February 1998. the performance mildly improves. 49. When the receive and transmit antenna arrays are in parallel.” in the ninth IEEE international symposium on personal. Abdi and M. Lastly. [6] F. May 2004. The transmit and receive antennas should be selected according to the antenna array orientations. IEEE Transactions on. Amundson and X. no. any antenna pair can be selected and the SSCD distributions would be similar to those in Fig. [7] W. It is also observed that the positive p χl 1 .p2 . VTC2007-Spring. “OFDM and its wireless applications: A survey. Wylie and J. 461–465. Therefore. pp. 4.. 2. Xu and S. “A survey on localization for mobile wireless sensor networks.” in 2nd International Workshop on Mobile Entity Localization and Tracking in GPS-less Environments (MELT). Rep. [10] L. “Position-location solutions by taylor-series estimation.-H.” Microwaves. In the next step.” Vehicular Technology. C. pp. and L. on universal personal communications. These results conﬁrm that the proposed technique is a good candidate for NLOS identiﬁcation that can be employed in MIMO-OFDM systems.q1 . “TOA and AOD statistics for down link Gaussian scatterer distribution model. Inf. 1. 4-7 2009. spatial subchannel correlation and channel estimation error is minimal. Fundamentals of statistical signal processing.q2 (whose real l and imaginary part are negative) as shown in Fig.e. May 2003. “IEEE P802. Sep 2009. [13] M. 827–831. [17] (2003. “The non-line of sight problem in mobile location estimation.q1 . this metric can be used to identify NLOS conditions. indoor and mobile radio communications. Sep) Spatial channel model for multiple input multiple output (MIMO) simulations.. 3. Giunta. then ﬁnd the threshold SSCDth by using a NeymanPearson detector or Bayes detector [18]. Kong. 3071–3075. vol.q2 would be determined from DOA distributions.p2 . 3. no. DSP/SPE 2009. 5. “Angle and time of arrival statistics for circular and elliptical scattering models. Jacobsen and et al. We also develop a rule on how to select the transmit and receive antennas for better performance. vol. Toscano. In this case. G. 2(a).. [2] W. the impact of antenna deployment is discussed. 4. Antennas and Propagation.11 Wireless LANs. Morelli and U. Discussions and future work In the simulation results. 58. “Spatially correlated multi-user channels: LOS vs. it is depicted that the impact of Rician K factor. [5] S. Available: http://ftp. 3. 4.e. IEEE Transactions on. Li.” Aerospace and Electronic Systems. Xu. Foy. 20. 2007. θ0 is independent of φ0 . no. we ﬁnd that the channel estimation error could degrade the performance as shown in Fig. pp. “A space-time correlation model for multielement antenna systems in mobile fading channels. It is observed that the distance of the LOS mean and NLOS mean slightly increases as the SNR increases. 2007. April 2007. [8] W. vol. 2 and Fig. “Dynamic LOS/NLOS statistical discrimination of wireless mobile channels. In the future. Dec. the LOS and NLOS SSCD distributions are shown in Fig.q1 . we assume that the power of taps exponentially decays and χp1 . vol. IEEE Transactions on. “A comparison of pilot-aided channel estimation methods for OFDM systems. Simulations using WLAN channel model conﬁrms that this metric can secure a high identiﬁcation performance. vol.11-03/940r4. march 1976. no.16-2009 (Revision of IEEE Std 802. R EFERENCES [1] I. 11. Sep. [12] “IEEE standard for local and metropolitan area networks part 16: Air interface for broadband wireless access systems. 3. it is observed that the inﬂuence of K. which provides a basis for distinguishing between LOS and NLOS. IEEE 802. Holtzman. pp. 17. IEEE Journal on. 2609 –2617. [9] T. Kaveh.q1 . vol. no. 3065 –3073. pp. volume 2: detection theory. 1673 –1694. may 2009.” IEEE Trans. The larger the distance is. pp. may 2009. NLOS. M. Thus. Venkatesh and R. 2002. Yang. no. the impact of θ0 is removed.” Selected Areas in Communications.” in Vehicular Technology Conference. D. IET. pp. “Diversity and multiplexing: a fundamental tradeoff in multiple-antenna channels. V.” High Throughput Task Group. This technique can also be used for single antenna transmitters.p2 . Vegni. and G. Sel. Li.p2 .” IEEE J. other power delay proﬁles. Tech.” IET Communications. pp. 12. i. 2009. [15] R. Theory. 1998. which is shown in Fig. 1829 –1840. θ0 = −φ0 . 5. Benedetto. 308 –313.q2 (whose real and imaginary part are positive) has a better performance than the negative χp1 .q2 is also studied.” in 5th IEEE international conf. 1120–1130. 8. The impact of Rician K factor and spatial correlation p coefﬁcient χl 1 . Reed. [Online]. we have observed that SSCD under LOS is probably larger than the SSCD under NLOS.q1 . 1996.org/specs/htmlinfo/25996.g. ﬁnally compare the obtained SSCD with SSCDth and make a decision between NLOS and LOS. G. in the 2nd round of review. IEEE 65th. Prentice hall. A. 49. and χl 1 . Areas Commun. Next. It is observed that the difference of crossantenna and single antenna subcarrier channel correlation under LOS conditions is much higher compared to NLOS conditions. C1 –2004. i. Mengali. Ertel and J. IEEE 13th. and the SSCD distributions are shown in Fig. the better we can identify NLOS from LOS. “A selective model to suppress NLOS signals in angle-of-arrival (AOA) location estimation. 2(b). Koutsoukos. C.q2 takes the value from a l reference. N. nov 1999. S. vol. Tse.q1 . pp. When the receive and transmit antenna array orientations are independent. we plan to use more realistic channel parameters to evaluate the NLOS identiﬁcation performance based on SSCD.” Wireless Communications. In addition. [4] M. Sep.. “Non-line-of-sight identiﬁcation in wireless localization via phase difference statistics across two antenna elements. “Non-line-of-sight identiﬁcation in ultrawideband systems based on received signal statistics. [16] S. Kay.” in Digital Signal Processing Workshop and 5th IEEE Signal Processing Education Workshop.

- Gsm&Spread Spectrum Reviewer
- OPNET2004b.pdf
- ultra wide ban
- Cortes Mimo Stbc Matlab
- Error Rate Analysis for Bit-Loaded Coded OFDM
- Interference Cancellation
- CS425_ Computer Networks_ Lecture 03
- uwb
- 05426055 Traffic-based Study of Femtocell Access Policy Impacts on HSPA Service Quality.pdf
- Multisite Field Trial for LTE and Advanced Concepts
- UMTS Network Design Using xWizard_including HSPA
- FastForward Paper
- mimo
- Matlab, Simulink Building a Direct Sequence Spread Spectrum Model
- Chirp Pulse Compression in Non–Coherent Impulse–Radio Ultra–Wideband Detection Without Waveform Signature Estimation
- syllabus
- C11_ASSEC05_mimo
- 2-Wcdma Ran Fundamental Issue1.0
- IJAIEM-2015-02-28-65
- 03 Cma5000 Xta Wander
- adcqb
- Indoor Geolocation in the Absence of Direct Path
- 01 DCOM Basics
- Distributed Scheduling in MIMO Empowered Cognitive Radio Ad Hoc
- EFL GoodOne
- wimax lte
- OFDM-MIMO
- 05 Transmission Hierarchy
- IJCNS_2013061814270587
- Basics of Spread Spectrum

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd