Applications Of Higher-Order Cyclic Statistics In Mobile Localization Parameter Estimation

Mobile localization technology is a method to determine the geolocation of mobile terminal within a certain range by sending wireless signals. Applications of mobile localization service in mobile communication networks to provide service about location and direction will bring people many conveniences in daily life. However, under the complicated mobile communication circumstances, it still has many technical difficulties in accurate localization of mobile station, mainly including the affact of multipath propagation, complicated noise and interference, multiple access interference (MAI), and no light of sight (NLOS), etc. It leads to the increase of location error and the declination location accuracy. In this dissertation, by means of advanced signal processing tool, i.e., higher order statistics, especially higher order cyclic statistics (HOCS), combined with the knowledges of array signal processing, wireless communication and matrix theory, many key problems in location parameter estimation are thoroughly studied. According to the present localization approaches in mobile communication and further consideration of future development in mobile communication system, location parameter estimation problem is examined from three aspects, respectively, i.e., DOA location parameter estimation, TOA location parameter estimation, and joint DOA/TOA location parameter estimation. The innovation works and results are summed up as follow. 1. Higher-order cyclic statistics (HOCS) theory is applied in multipath environment of mobile communication, according to adapting to various multipath situation and noise/interference background, a series of HOCS-based multipath DOA estimation methods are proposed, such as the forward-backward linear prediction for DOA estimation method using higher-order cyclostationarity, the HOCS and linear prediction-based EVESPA algorithm (HOCS-LP-EVESPA), the HOCS and temporal smoothing based EVESPA algorithm (HOCS-TS-EVESPA). It is shown that HOCS based method possesses merits of both cyclic statistics and higher order statistics. It has more powerful signal detection capability and noise suppression. Furthermore, by defining the form of higher-order cyclic cumulant, the capability of array aperture extension can be obtained. For certain signals without second-order cyclostationarity, the higher order cyclic cumulants can be applied exploiting its higher-order cyclostationarity property. 2. When the received signal is corrupted by both multiplicative noise and additive noise, methods of one dimensional (1-D) DOA estimation and two dimensional (2-D) DOA estimation are proposed to effectively solve the problems of coexistence of multiplicative/additive noises in the presence of multiple coherent signal caused by multipath propagation. By exploiting third-order cyclic moment, the proposed method can effectively resolve directions of multiple coherent signals of interest (SOIs) even if the total number of impinging signals is no less than the number of sensors. The stationary additive and nonzero-mean multiplicative noises with any distribution can be suppressed no matter the noises are Gaussian and non-Gaussian, white or colored. Also, the cyclostationary interference with different cyclic frequency from the SOIs can be effectively removed. The temporal smoothing is applied to resolve 2-D DOAs of multiple coherent signals in multiplicative/ additive noises. The proposed temporal smoothing based method avoid multiple sub-planar array structure grouping, has less complexity, higher resolution and further array aperture extension. 3. Exploiting higher order cyclostationarity of communication signal, DOA algorithm is proposed using minimum-redundancy linear arrays (MRLA) based on fourth-order cyclic cumulants. Compared with previous fourth-order cyclic cumulants method using ULA proposed by S.Shamsunder, due to using virtualsensors to eliminate the redundancy information and extend array aperture, the proposed MRLA method is proved to provide better performance in terms of ability to detect and resolve a greater number of sources. More than 2M-2 sources can be estimated with M sensors. Its resolution capability is improved. Also, proposed method has more powerful signal detection capability and noise suppression by the use of HOCS. 4. In order to solve the problem of multiple correlative path fading amplitudes, spatial smoothing technology of array signal processing is extended to frequency domain signal processing. A novel TOA estimation method in multipath channel is proposed based on higher order statistics and frequency domain smoothing. The proposed method can solve the problem in the presence of Gaussian colored noise and when multiple path fading amplitudes are correlative with each other. 5. Higher order cumulant is firstly applied in joint DOA/TOA estimation of multipath channel. A novel higher order cumulant-based TST method is proposed to joint estimation multipath DOA/TOA parameters. Further, a new higher-order cumulant-based algorithm is proposed to joint estimate three parameter of azimuth, elevation and propagation delay in a multipath channel using arrival signals at a planar antenna array. Both the proposed methods for joint parameter estimation can work in additive spatially and temporally colored Gaussian noise. Based on the obtained multiple location parameter information, three-dimensional space geolocation can be realized. The dissertation consists of ten chapters. Chapter one is introduction, in which the current application and the main methods of mobile positioning location in mobile communication network are elucidated. The data models of DOA parameter estimation, TOA parameter estimation and joint DOA/TOA parameter estimation are introduced, respectively. Their development history and status in quo are summarized. After discussing the key problems in the present mobile localization and parameter estimation, the research content of this dissertation are proposed.Chapter two is the fundamental knowledge involved in this dissertation. The knowledge of matrix algebra is introduced. The definitions and properties of higher-order cumulant, cyclic statistics and higher-order cyclic statistics are described. Chapter three examines the DOA parameter estimation methods for multipath signals based on higher-order cyclic statistics. Generally, wireless signals arrive at base station from mobile station through multiple routes, multipath propagation due to various reflectors are often encountered, which lead to the appearance of coherent signals and the estimation of directions of arrival is inaccurate, or even wrong, It greatly affects the veracity of DOA estimation. How to suppress the multipath interference is a difficult technical problem in position localization of mobile communication system. In this chapter, higher-order cyclic statistics are applied in multipath environment of mobile communication, a series of novel higher-order cyclic statistics-based DOA estimation of multipath signals are proposed, which adapt to various multipath situation and noise/interference background. The proposed methods consist of (1) De-correlating method for DOA estimation based on fourth-order cyclic cumulant; (2)Forward-backward linear prediction for DOA estimation using higher-order Cyclic Statistics; (3)Cyclic cumulant based DOA Estimation using linear operator without eigen-decomposition and spatial smoothing; (4)Higher-order cyclic statistics and linear prediction-based EVESPA algorithm (HOCS-LP-EVESPA); (5)Higher-order cyclic statistics and temporal smoothing-based EVESPA algorithm (HOCS-TS-EVESPA).The simulation results of all the algorithms are given in the chapter. In chapter four, 2-D DOA parameter estimation methods in multipath are proposed based on higher-order cyclic statistics (HOCS). In one-dimensional (1-D) DOA estimation methods, with a linear array, the resolved directions of arrival are only located in a plane. That is to say, the determined geolocation of mobile terminal using DOA-based methods with two base stations is only in a plane, andthe information of height cannot be obtained. The existing methods for multipath 2-D DOA estimation cannot effectively eliminate the influence of complicated noises and interference, and the estimated number of signals is limited by the size of sub-array, which is not applicable in mobile communication. In this chapter, an azimuth/elevation 2-D DOA estimation algorithm for multipath signal is proposed using higher-order cyclic cumulant to detect non-Gaussian coherent cyclostationary sources by designing a fourth-order cyclic cumulant matrix and using eigen vector (EV) approach. Further, another EVESPA-based azimuth/elevation 2-D DOA estimation algorithm for multipath signal is proposed based on cyclic cumulant and temporal smoothing. Chapter five examines the DOA estimation problem when the received signal is corrupted by both multiplicative noise and additive noise, which is often encountered in actual wireless communication systems. Firstly, a novel 2-D DOA estimation algorithm based on third-order cyclic moment is proposed to detect non-Gaussian cyclostationary sources in the presence of multiplicative/additive noises. Secondly, the problem in the presence of multiple coherent signals through multipath propagation and when the received signal is corrupted by both multiplicative noise and additive noise is studied. By exploiting cyclostationarity of the signals, a multipath direction finding algorithm is proposed to detect multiple non-Gaussian cyclostationary coherent signals in the multiplicative/additive noises environments. Thirdly, based on real mobile communication channel characteristics, in which the multipath reflection coefficient changes more rapidly than their DOAs, the temporal smoothing is applied to resolve 2-D DOAs of coherent signals. The 2-D multipath DOA estimation algorithm using a planar array is provided exploiting higher-order cyclostationarity and temporal smoothing technology. The simulation results indicate that the proposed method can effectively resolve directions of multiple coherent signals of interest (SOIs) even if the total number of impinging signals is no less than the number of sensors. The stationary additive and nonzero-mean multiplicative noises with any distribution can be suppressed no matter the noises are Gaussian and non-Gaussian, white or colored. Also, thecyclostationary interference with different cyclic frequency from the SOIs can be effectively removed. In chapter six, DOA estimation method with extended array aperture using special-shaped arrays is examined. In most actual communications, the overall number of signals impinging on the array is often greater than the number of sensors. However, the array aperture in conventional uniform linear array is limited. In this chapter, a DOA estimation algorithm is proposed to detect non-Gaussian cyclostationary sources using minimum-redundancy linear arrays (MRLA) based on fourth-order cyclic cumulants. Simulation results show that our method can effectively suppress additive stationary noise and Gaussian noise in environments where the spatial characteristics of noise are unknown, even when the noise shares the same cycle frequency as SOIs. Moreover, compared with the designed fourth-order cyclic cumulants method using ULA, the MRLA method is proved to provide better performance in terms of ability to detect and resolve a greater number of sources. More than 2M-2 sources can be estimated with M sensors. In chapter seven, the DOA estimation of CDMA system is studied in MAI environment. Most conventional DOA estimation algorithms proposed so far are not applicable for a direct-sequence code-division multiple access (DS-CDMA) system. In order to improve the DOA resolution in CDMA system and effectively suppress MAI, this chapter develops an effective high resolution DOA estimation method for multiple mobile subscribers in the synchronous DS-CDMA systems over frequency-selective fading channels, which is mainly based on a de-correlating multiuser detector and a cumulant-based aperture-extended MUSIC algorithm. The simulation results of the proposed algorithm are given in the chapter. In chapter eight, TOA estimation method for multipath coherent signals is studied using higher order statistics. In the previous methods proposed so far, an assumption is given that the fading amplitude of each path is statistically independent with each other. However, in some actual surroundings, the pathfading amplitudes are correlative with each other, which lead to rank-deficiency of constructed frequency domain covariance matrix in spectrum estimation, and performance of TOA estimation is seriously descended. In the dissertation, spatial smoothing technology of array signal processing is extended to frequency domain signal processing. A novel TOA estimation method in multipath channel is proposed based on higher order statistics and frequency domain smoothing. The proposed method can solve the problem in the presence of Gaussian colored noise and when multiple path fading amplitudes are correlative with each other. The simulation results of the proposed algorithm are given in the chapter. In chapter nine, the problem of joint estimation of DOA/TOA parameter is discussed. In order to suppress Gaussian colored noise and improve the parameter resolution, higher order cumulant is firstly applied in joint DOA/TOA estimation of multipath channel. A novel higher order cumulant-based TST method is proposed to joint estimation mulipath DOA/TOA parameters, which has high resolution. Based on this, another higher-order cumulant-based subspace algorithm for multipath signals arrival at a planar antenna array is proposed to joint estimate three parameter of azimuth, elevation and propagation delay. The proposed method for joint parameter estimation can work in additive unknown spatially and temporally colored Gaussian noise in a wireless channel. Based on the obtained multiple location parameter information, three-dimensional space geolocation can be realized. The simulation results of both proposed algorithms are given in the chapter. In chapter ten, a brief summary of the dissertation is given. The suggestion for future researches related to the location parameter estimation is put forward.