基于射频信号及信道状态信息的被动式目标定位方法研究
发布时间：2018-11-27 10:28
【摘要】：以无线信号感知为中心的被动式目标定位,以其无需给待定位目标携带任何设备的优点,成为了许多应用中最受欢迎的技术之一,如入侵检测、老人健康监测等。与传统基于红外、视频、光学感知的被动式目标定位方法相比,基于无线信号的被动式目标定位优点在于：无线信号覆盖范围广,一般不存在死角问题,可以工作在有遮挡、非视距的场景下,对光线要求和遮挡物不敏感,可以全天候昼夜工作。近年来随着WiFi等无线设备在人们日常生活中的普及,使得基于无线信号的被动式目标定位成为技术主流。尽管基于无线信号的被动式目标定位在国际范围内取得了重要的进展,然而在实际应用中,现有方法还存在如下四个主要的问题。第一,为达到高精度的多目标定位,现有被动式目标定位算法需要采集大量目标对无线信号强度(received signal strength, RSS)造成的扰动改变值,导致了高能量消耗,制约了系统在实际环境下的可用性。第二,现有被动式目标定位方法,均假设先验获取的某类目标的RSS扰动先验知识数据可以适用于其他种类目标。然而,实际应用中不同种类目标的形状、散射面等属性不同,使得定位时得到的某类目标的RSS扰动数据与先验获得的其他类目标知识库失配,造成较大的定位误差。第三,仅依赖RSS信号的被动式定位技术尽管已经取得了重要的进展,但是由于RSS信号本身的不稳定性,其定位精度仍然十分有限。第四,现有基于通用设备的高精度被动式定位方法大都依赖指纹库,导致人力代价高的问题。因此,探寻真实场景下更为普适的被动式目标定位方法具有重大意义和价值。本文从目标对无线信号扰动的不同角度出发,分析了包括目标对信号幅度、相位、信道状态信息的影响规律,以低能量消耗、目标种类自适应、高精度、低代价为目标,提出了四种不同真实场景下具有可用性的被动式目标定位方法。本文工作的主要创新研究如下：(i)针对现有被动式目标定位算法需采集大量观测值导致高能量消耗的问题,本文提出一种基于压缩感知的低能耗高精度多目标被动式定位方法E-HIPAo根据多目标位置在物理空间中具有的固有稀疏性,将多目标定位问题建模为压缩感知稀疏恢复问题。利用压缩感知理论,通过收集少量RSS观测数据用于目标位置的估计,降低通信代价和能量消耗。提出自适应正交匹配追踪算法,在无需先验预知目标个数的情况下,精确估计出多目标的位置。大量真实实验结果表明,E-HIPA相比于现有的方法,能够在降低能量消耗的前提下保证高精度多目标定位,因此更适用于真实部署应用场景。(ii)针对实际应用中需要针对每一类目标建立先验知识库导致人力消耗巨大的问题,本文进一步提出一种可迁移压缩感知的多目标被动式定位方法TLCS。借鉴迁移学习思想构建迁移函数,在不同种类目标间进行指纹库迁移,保证了迁移后的不同种类目标的RSS干扰改变值在潜在特种空间具有分布相似性,使得不同种类目标共享相同一个先验数据库,从而避免了对每类目标多次人工勘测标定带来的部署开销和人力消耗。严谨的理论分析和大量真实实验表明,TLCS方法不仅保证了高定位精度,而且极大降低了多次学习先验指纹库的人力消耗代价。(iii)针对现有被动式目标定位方法在多径效应影响下无法精确定位目标的问题,本文提出基于多径信号空间谱分析的高精度被动式目标定位方法D-Watch。与传统方法均假设多径信号是不利因素而不同,D-Watch利用丰富的多径信号空间谱信息来识别目标的方位角,综合多个方位角可以定位目标位置。为了提高目标方位角估计精度,本文对对多径信号空间谱的相关矩阵进行特征值分解,利用信号子空间和噪声子空间的正交性,通过最优化理论对相位误差进行校正,使得可以获取精确的多径信号空间谱信息,达到了多经环境下高精度被动式定位目标的目的。大量真实实验表明,D-Watch能够利用细粒度的多径信号空间谱进行分米级精度定位,突破了传统方法无法再多径环境下精度定位目标的瓶颈。(iv)针对现有基于指纹的被动式目的定位方法,因面临场景变换需要重新采集指纹库而导致人力代价高的问题,本文提出一种基于信道状态模型的低成本高精度被动式目标定位方法LiFS。利用不同子载波的信道状态信息受到多径影响不同的特性,提取出未受多径影响的“干净”子载波,从而构建了精确的“信道状态-目标位置”关联模型。将所有链路的位置以及“干净”子载波的信道状体信息作为模型的输入,利用不同子载波上的关联模型来精确计算目标位置,避免了大量指纹库训练的代价问题。大量真实实验表明,LiFS在现有WiFi设备上,实现了低成本、高精度的被动式目标定位,因此更适用于真实场景下的定位需求。
[Abstract]:The passive target location, which is perceived as a center by the wireless signal, is one of the most popular technologies in many applications, such as the intrusion detection, the health monitoring of the elderly, and the like, without the advantage of carrying any equipment to the target to be located. Compared with the traditional passive target positioning method based on infrared, video and optical sensing, the passive target positioning method based on the wireless signal has the advantages that the wireless signal coverage range is wide, It is not sensitive to the light requirements and the shelter, and can work all the clock and night. In recent years, with the popularization of wireless devices such as WiFi in the daily life of people, the passive target positioning based on the wireless signal becomes the main technology. Although the passive target location based on the wireless signal has made important progress in the international range, in practical applications, there are four main problems in the prior art. First, in order to achieve the multi-target positioning with high accuracy, the present passive target positioning algorithm needs to acquire the disturbance change value caused by a large number of targets to the wireless signal strength (RSS), resulting in high energy consumption and the availability of the system in the actual environment. secondly, the prior passive target positioning method is used for assuming that the prior knowledge data of the RSS disturbance of a certain type of target acquired a priori can be applied to other kinds of targets. However, the shape, scattering surface and other properties of different kinds of objects in the practical application are different, so that the RSS disturbance data of some kind of target obtained at the time of positioning is not matched with the prior knowledge base of other classes, resulting in a large positioning error. Third, although the passive positioning technology relying on the RSS signal has made important progress, the positioning accuracy of the RSS signal is still very limited due to the instability of the RSS signal itself. Fourth, the existing high-precision passive positioning method based on general equipment mostly relies on the fingerprint library, resulting in high labor cost. Therefore, it is of great significance and value to find a more suitable passive target location method in real scene. In this paper, the influence of the target on the signal amplitude, phase and channel state information is analyzed from different angles of the target to the wireless signal disturbance. The target is low energy consumption, target type adaptation, high accuracy and low cost. A passive target location method with availability in four different real scenarios is presented. the main innovation of the work is as follows: (i) the problem of high energy consumption is caused by the acquisition of a large number of observation values for the existing passive target positioning algorithm, In this paper, a high-precision multi-target passive positioning method based on compression-sensing is proposed, which is based on the inherent sparsity of the multi-target location in the physical space, and the multi-target location problem is modeled as the compression-aware sparse recovery problem. By using the compression-aware theory, the communication cost and energy consumption are reduced by collecting a small amount of RSS observation data for estimation of the target position. A self-adaptive orthogonal matching tracking algorithm is proposed, and the position of the multi-target can be accurately estimated without a priori knowledge of the number of the targets. A large number of real-time experiments show that, compared with the existing method, E-HIPA can ensure high-precision multi-target positioning on the premise of reducing energy consumption, so that the E-HIPA is more suitable for real-deployment application scenarios. (ii) The need to establish a priori knowledge base for each class of objects in the actual application results in significant human consumption. In this paper, a multi-target passive location method TLCS is proposed. using the migration learning idea to construct a migration function, carrying out fingerprint library migration among different kinds of objects, ensuring that the RSS interference change value of different kinds of objects after the migration has the distribution similarity in the potential special space, so that different kinds of objects share the same prior database, so as to avoid the deployment cost and the manpower consumption caused by multiple manual survey and calibration of each class of targets. The rigorous theoretical analysis and a large number of real experiments show that the TLCS method not only ensures the high positioning accuracy, but also greatly reduces the labor consumption cost of the prior fingerprint library for many times. (iii) Aiming at the problem that the existing passive target positioning method cannot accurately position the target under the influence of the multi-diameter effect, the invention provides a high-precision passive target positioning method D-Watch based on the multi-path signal space spectrum analysis. Compared with the traditional method, the multi-path signal is assumed to be a disadvantage, and the D-Watch uses the abundant multi-path signal space spectrum information to identify the azimuth of the target, and the comprehensive multiple azimuth angles can be used for positioning the target position. In order to improve the accuracy of the target azimuth estimation, the correlation matrix of the multi-path signal space spectrum is decomposed, the orthogonality of the signal subspace and the noise subspace is utilized, the phase error is corrected by the optimization theory, so that the accurate multi-path signal space spectrum information can be obtained, and the purpose of the high-precision passive positioning target under the multi-channel environment is achieved. A large number of real experiments show that the D-Watch can use the fine-grained multi-path signal space spectrum to position the sub-meter-level precision, and breaks through the bottleneck that the traditional method can not accurately locate the target under the multi-path environment. (iv) Aiming at the problem of high labor cost due to the need to re-acquire the fingerprint library in the face of the scene transformation, a low-cost high-precision passive target positioning method based on the channel state model is proposed for the existing passive object positioning method based on the fingerprint. The channel state information of different sub-carriers is influenced by the multi-path to different characteristics, and the 鈥渃lean鈥，
本文编号：2360422
[Abstract]:The passive target location, which is perceived as a center by the wireless signal, is one of the most popular technologies in many applications, such as the intrusion detection, the health monitoring of the elderly, and the like, without the advantage of carrying any equipment to the target to be located. Compared with the traditional passive target positioning method based on infrared, video and optical sensing, the passive target positioning method based on the wireless signal has the advantages that the wireless signal coverage range is wide, It is not sensitive to the light requirements and the shelter, and can work all the clock and night. In recent years, with the popularization of wireless devices such as WiFi in the daily life of people, the passive target positioning based on the wireless signal becomes the main technology. Although the passive target location based on the wireless signal has made important progress in the international range, in practical applications, there are four main problems in the prior art. First, in order to achieve the multi-target positioning with high accuracy, the present passive target positioning algorithm needs to acquire the disturbance change value caused by a large number of targets to the wireless signal strength (RSS), resulting in high energy consumption and the availability of the system in the actual environment. secondly, the prior passive target positioning method is used for assuming that the prior knowledge data of the RSS disturbance of a certain type of target acquired a priori can be applied to other kinds of targets. However, the shape, scattering surface and other properties of different kinds of objects in the practical application are different, so that the RSS disturbance data of some kind of target obtained at the time of positioning is not matched with the prior knowledge base of other classes, resulting in a large positioning error. Third, although the passive positioning technology relying on the RSS signal has made important progress, the positioning accuracy of the RSS signal is still very limited due to the instability of the RSS signal itself. Fourth, the existing high-precision passive positioning method based on general equipment mostly relies on the fingerprint library, resulting in high labor cost. Therefore, it is of great significance and value to find a more suitable passive target location method in real scene. In this paper, the influence of the target on the signal amplitude, phase and channel state information is analyzed from different angles of the target to the wireless signal disturbance. The target is low energy consumption, target type adaptation, high accuracy and low cost. A passive target location method with availability in four different real scenarios is presented. the main innovation of the work is as follows: (i) the problem of high energy consumption is caused by the acquisition of a large number of observation values for the existing passive target positioning algorithm, In this paper, a high-precision multi-target passive positioning method based on compression-sensing is proposed, which is based on the inherent sparsity of the multi-target location in the physical space, and the multi-target location problem is modeled as the compression-aware sparse recovery problem. By using the compression-aware theory, the communication cost and energy consumption are reduced by collecting a small amount of RSS observation data for estimation of the target position. A self-adaptive orthogonal matching tracking algorithm is proposed, and the position of the multi-target can be accurately estimated without a priori knowledge of the number of the targets. A large number of real-time experiments show that, compared with the existing method, E-HIPA can ensure high-precision multi-target positioning on the premise of reducing energy consumption, so that the E-HIPA is more suitable for real-deployment application scenarios. (ii) The need to establish a priori knowledge base for each class of objects in the actual application results in significant human consumption. In this paper, a multi-target passive location method TLCS is proposed. using the migration learning idea to construct a migration function, carrying out fingerprint library migration among different kinds of objects, ensuring that the RSS interference change value of different kinds of objects after the migration has the distribution similarity in the potential special space, so that different kinds of objects share the same prior database, so as to avoid the deployment cost and the manpower consumption caused by multiple manual survey and calibration of each class of targets. The rigorous theoretical analysis and a large number of real experiments show that the TLCS method not only ensures the high positioning accuracy, but also greatly reduces the labor consumption cost of the prior fingerprint library for many times. (iii) Aiming at the problem that the existing passive target positioning method cannot accurately position the target under the influence of the multi-diameter effect, the invention provides a high-precision passive target positioning method D-Watch based on the multi-path signal space spectrum analysis. Compared with the traditional method, the multi-path signal is assumed to be a disadvantage, and the D-Watch uses the abundant multi-path signal space spectrum information to identify the azimuth of the target, and the comprehensive multiple azimuth angles can be used for positioning the target position. In order to improve the accuracy of the target azimuth estimation, the correlation matrix of the multi-path signal space spectrum is decomposed, the orthogonality of the signal subspace and the noise subspace is utilized, the phase error is corrected by the optimization theory, so that the accurate multi-path signal space spectrum information can be obtained, and the purpose of the high-precision passive positioning target under the multi-channel environment is achieved. A large number of real experiments show that the D-Watch can use the fine-grained multi-path signal space spectrum to position the sub-meter-level precision, and breaks through the bottleneck that the traditional method can not accurately locate the target under the multi-path environment. (iv) Aiming at the problem of high labor cost due to the need to re-acquire the fingerprint library in the face of the scene transformation, a low-cost high-precision passive target positioning method based on the channel state model is proposed for the existing passive object positioning method based on the fingerprint. The channel state information of different sub-carriers is influenced by the multi-path to different characteristics, and the 鈥渃lean鈥，
本文编号：2360422
本文链接：https://www.wllwen.com/shoufeilunwen/xxkjbs/2360422.html
