WO2015192696A1 - 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法 - Google Patents

分布式网络中基于gossip算法的多目标DOA估计系统及估计方法 Download PDF

Info

Publication number
WO2015192696A1
WO2015192696A1 PCT/CN2015/078172 CN2015078172W WO2015192696A1 WO 2015192696 A1 WO2015192696 A1 WO 2015192696A1 CN 2015078172 W CN2015078172 W CN 2015078172W WO 2015192696 A1 WO2015192696 A1 WO 2015192696A1
Authority
WO
WIPO (PCT)
Prior art keywords
iteration
receiving node
value
gossip
vector
Prior art date
Application number
PCT/CN2015/078172
Other languages
English (en)
French (fr)
Inventor
谢宁
张莉
王晖
林晓辉
曾捷
Original Assignee
深圳大学
谢宁
张莉
王晖
林晓辉
曾捷
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳大学, 谢宁, 张莉, 王晖, 林晓辉, 曾捷 filed Critical 深圳大学
Publication of WO2015192696A1 publication Critical patent/WO2015192696A1/zh

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W64/00Locating users or terminals or network equipment for network management purposes, e.g. mobility management
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16ZINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
    • G16Z99/00Subject matter not provided for in other main groups of this subclass

Definitions

  • the invention relates to a multi-objective DOA estimation system and an estimation method based on gossip algorithm in a distributed network.
  • the DOA estimation of applying the gossip algorithm to a distributed network is mainly to share information between nodes to obtain a DOA estimate for each node.
  • the original DOA algorithm such as the CAPON algorithm for distributed DOA estimation
  • matrix inversion is not easy to achieve, this can cause a lot of interference in a multi-target scenario. Therefore, it is of great significance to seek a DOA estimation algorithm that does not require matrix inversion and combine the gossip algorithm for DOA estimation in distributed networks.
  • the technical problem to be solved by the present invention is to propose a multi-objective DOA estimation system and estimation method based on gossip algorithm in a distributed network, and can realize information sharing and DOA value in a distributed network without a matrix inversion method. estimate.
  • the present invention is implemented as follows:
  • a multi-objective DOA estimation method based on gossip algorithm in a distributed network comprising the following steps:
  • the gossip iteration is performed on the ⁇ th signal data vector according to the ⁇ th iteration rule, and after each iteration, it is determined whether the ⁇ th signal data vector is equal to before the iteration, and if so, recording and accumulating Equal number of times, otherwise the number of equals is zeroed; wherein, the initial value of ⁇ is 1, and the ⁇ signal data vector is constructed by using the initial values of the receiving nodes in the ⁇ th iteration period;
  • the iteration of the ⁇ th iteration period is completed, and the ⁇ th signal data vector at this time is stored, and each of the ⁇ th signal data vectors is received.
  • the signal of the node is taken as the initial value of each receiving node in the ⁇ +1 iteration period;
  • the above steps are executed cyclically.
  • the value of ⁇ is increased by 1 from the value of ⁇ in the previous cycle.
  • the calculation formula of the estimated value of DOA is used according to the iterative result of each iteration period.
  • the DOA estimate is calculated; the preset value is the number of iteration cycles required to calculate the DOA estimate based on the calculation formula of the DOA estimate.
  • a multi-objective DOA estimation system based on gossip algorithm in distributed network comprising:
  • a loop iteration module configured to perform a gossip iteration on the ⁇ th signal data vector according to the ⁇ th iteration rule in the ⁇ th iteration period, and after each iteration, determine whether the ⁇ th signal data vector is equal to before the iteration, if Yes, then record and accumulate the same number of times, otherwise return the equal number of times; wherein, the initial value of ⁇ is 1, the ⁇ signal data vector is constructed by using the initial values of the receiving nodes in the ⁇ iteration period; When the equal number of times reaches the preset number, and the value of ⁇ does not reach the preset value, the iteration of the ⁇ th iteration period is completed, and the ⁇ th signal data vector at this time is stored, and the signals of the receiving nodes in the ⁇ th signal data vector are obtained. As the initial value of each receiving node in the ⁇ +1 iteration cycle; cyclically performing the above steps, each time the cycle, the value of ⁇ is increased by 1 from the value of ⁇ in the
  • a DOA estimation value calculation module configured to calculate a DOA estimation value by using a calculation formula of the DOA estimation value according to an iteration result of each iteration period when the value of ⁇ reaches a preset value; the preset value is based on the DOA estimation value
  • the calculation formula calculates the number of iteration cycles required for the DOA estimate.
  • the present invention calculates the number of iteration cycles required for calculating the DOA estimation value according to the calculation formula of the DOA estimation value, and uses the gossip algorithm to continuously update the signal component of the DOA estimation to an ideal for information sharing. State, after all iterations have been performed, a good estimate of the DOA can be obtained.
  • the invention does not require a matrix inversion method, and can realize information sharing and DOA value estimation in a distributed network.
  • FIG. 1 is a schematic flowchart of a multi-target DOA estimation method based on a gossip algorithm in a distributed network according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of a composition of a multi-target DOA estimation system based on a gossip algorithm in a distributed network according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of values obtained in each step of the multi-objective DOA estimation method based on the gossip algorithm in the distributed network provided by the embodiment of the present invention.
  • the invention takes a wireless communication distributed system composed of three transmitting array elements and three receiving array elements as an example, and illustrates a process of implementing DOA estimation by using a gossip iterative algorithm combined with an AV algorithm.
  • the gossip method is a more effective method, and for the inverse operation of the autocorrelation matrix of the distributed signal, the present invention can obtain the optimal weight without the inverse of the matrix.
  • the AV algorithm finally implements DOA estimation in a distributed network.
  • FIG. 1 is a schematic flowchart of a multi-objective DOA estimation method based on the gossip algorithm in a distributed network according to an embodiment of the present invention
  • FIG. 2 is a diagram showing a multi-target DOA estimation based on a gossip algorithm in a distributed network according to an embodiment of the present invention. Schematic diagram of system composition.
  • the random gossip algorithm can be used to solve the distributed convex problem, given the initial scalar value of a random N-node network and the ith node.
  • the purpose of the random gossip algorithm is to achieve an average of all targets by using only local information and local processing and an iterative mechanism.
  • g(t) [g 1 (t),...,g N (t)] T (0.1) (Note, “(0.1)” indicates the number of the formula, not part of the formula, subsequent formulas The same reason.)
  • each node runs an independent Poisson clock.
  • the node randomly selects a neighboring j-node with probability p i,j and communicates with it.
  • the general vector representation of the Gossip algorithm is
  • T is an N-dimensional vector whose i-th element is 1.
  • U(t) is a double backward random matrix and the network is connected, it can ensure that all nodes in the network can converge to the mean g ave .
  • the most important task is to define the initial vector g(0) for all nodes.
  • d k (0) is the initial distance between the target and the origin at time 0.
  • the i-th transmitting node continuous time transmit waveform is represented as x i (t)e j2 ⁇ ft , where f is the carrier frequency and all transmitting nodes use the same carrier frequency, and x i (t) is the periodic narrow-band signal with T p .
  • the received signal of the kth target can be expressed as
  • the latter assumption is based on the far-field assumption that the distance between network nodes is much smaller than the distance between the nodes and the target. Therefore, since the nodes are close together, it can be considered that all receiving nodes see the same surface of the target.
  • the signal received by the lth receiver is expressed as follows
  • ⁇ l (t) represents an independent and identical distribution with a mean of 0 and a Gaussian noise with a variance of ⁇ 2 .
  • the sampled signal can be regarded as the synchronization signal of the information reflected by the first target, and since the transmitted waveform is a narrowband signal, the delay in the transmitted waveform x i (t) can be ignored, only Consider the delay of the phase part. Therefore, the received baseband signal of the lth receiving end can be approximated as
  • ⁇ lm [ ⁇ l ((m-1)T+0T s ),... ⁇ l ((m-1)T+(L-1)T s )] T (0.13)
  • the transmit waveform of each transmit antenna is independent, therefore, relative In terms of i ⁇ i' It is negligible.
  • the conventional CAPON algorithm produces a beamforming vector w that can suppress noise, while the interference and noise are suppressed, while the desired signal remains undistorted.
  • w can be expressed as follows:
  • the LS method is applied to the beamforming output by w * , and the estimation of the target reflection coefficient can be easily obtained based on the assumption of the third point as follows:
  • auxiliary vector (AV) technique In the traditional capon algorithm, in order to obtain the optimal weight in the formula (0.18), matrix inversion operation is required, but in distributed signal processing, matrix inversion is not easy to implement. Therefore, another method that does not require matrix inversion can be used to obtain an optimal weight vector, namely the auxiliary vector (AV) technique.
  • the traditional AV algorithm is mainly applied to the space-time filtering of the antenna array, and can be directly applied to the DOA estimation problem.
  • the optimal weight vector based on AV technology can be expressed as
  • G( ⁇ ) is chosen to be able to process data with v r ( ⁇ )
  • the amplitude of the cross-correlation function with the AV processed data G( ⁇ ) H Z is maximized.
  • v r ( ⁇ ) is orthogonal to each other, so that The largest AVG( ⁇ ) can extract the perturbation component of most beamforming outputs.
  • the optimal AVG( ⁇ ) can be obtained according to the following formula.
  • a single AV G( ⁇ ) usually represents a degree of freedom, and if you need to increase the resolution, you can use multiple auxiliary vectors.
  • the beam synthesis weight vector can be expressed as follows
  • the estimated value of the target reflection coefficient becomes
  • the target reflection coefficient is estimated as
  • each receiving node i in the WSN has two initial values in each given time slot.
  • N r vector Expressed as the initial value of the N r node
  • Each node jointly transmits a signal, and at the same time, each node receives a signal and constructs an initial signal based on the received signal, the initial signal being represented as Where i is the serial number of the node and ⁇ is the angle;
  • represents the wavelength of the transmitted signal
  • z i (l-1) represents the received signal of the i-th node at the l-1th sampling point
  • L represents the number of sampling points
  • x m is the transmitted waveform.
  • Multi-objective DOA estimation method based on gossip algorithm in distributed network
  • the gossip estimation method introduced above can be considered as an extension of distributed delay and beam sum, and is only used to receive spatial signals.
  • the main disadvantage is the use of R EE instead of R in equation (0.18). Assuming that there are multiple targets in the system or that the length L of the waveform is not long enough, its performance will be seriously degraded.
  • an iterative random gossip algorithm (IR-Gossip) (0.21) using AV technology is proposed.
  • each receiving node i in the WSN has an initial value for a given time interval
  • the node obtains a steady state by some iterations t 2 -t 1 In the third loop, you can get the new initial value of t 2 ,
  • the node reaches a steady state by a certain number of iterations t 3 -t 2 get
  • the node reaches a steady state by a certain number of iterations t 3 -t 2
  • the fourth cycle get the initial value of the new t 3
  • the node reaches a steady state by a certain number of iterations t 4 -t 3 Can get
  • each node i has an initial value at t 4
  • the receiving node reaches a steady state by a certain number of iterations t 6 -t 5 Can get
  • ⁇ 6 ( ⁇ ) can be obtained directly without the need for a cyclic operation.
  • the DOA estimation method includes the following steps:
  • Step S1 performing a gossip iteration on the ⁇ th signal data vector according to the ⁇ th iteration rule in the ⁇ th iteration period, and determining, after each iteration, whether the ⁇ th signal data vector is equal to before the iteration, and if so, Recording and accumulating the number of equal times, otherwise returning the number of equals to zero; wherein the initial value of ⁇ is 1, and the ⁇ signal data vector is constructed by using the initial values of the receiving nodes in the ⁇ th iteration period;
  • Step S2 When the equal number of times reaches the preset number of times, and the value of ⁇ does not reach the preset value, the iteration of the ⁇ th iteration period is completed, and the ⁇ th signal data vector at this time is stored, and the ⁇ th signal data vector is obtained.
  • the signals of the receiving nodes in the middle are used as initial values of the receiving nodes in the ⁇ +1 iteration period;
  • Step S3 cyclically executing the above steps.
  • the value of ⁇ is increased by 1 from the value of ⁇ in the previous cycle.
  • the DOA estimation value is used according to the iterative result of each iteration period.
  • the calculation formula calculates a DOA estimate; the preset value is the number of iteration cycles required to calculate the DOA estimate based on the calculation formula of the DOA estimate.
  • the preset value is 6.
  • Each receiving node receives the initial signal z i (l-1) and according to Construct the initial value of each node in the first iteration cycle
  • represents the wavelength of the transmitted signal
  • the approximate distance z i (l-1) of the target representing the angle ⁇ and the ith receiving node represents the received signal of the i-th node at the l-1th sampling point
  • L represents the number of sampling points
  • [ ] * represents a conjugate operation
  • N r G-dimensional vectors Forming a first signal data vector; wherein N r is the number of receiving nodes, and G is a discretization precision of the angular space;
  • Each receiving node Stored in N r GL dimension vector Forming a second signal data vector
  • Each receiving node Stored in N r GL dimension vector Forming a third signal data vector
  • the fifth iteration rule Iterating the fifth signal data vector wherein Represents the gossip update matrix, An identity matrix representing N r GL, where e GLi represents a GL dimension vector in which the i-th GL-GL-1 element to the i-th GL element are 1 and the other elements are 0;
  • each node produces a rough DOA estimate.
  • the sixth iteration rule Iterating the sixth signal data vector wherein Represents the gossip update matrix, An identity matrix representing N r G, where e Gi represents a G-dimensional vector in which the i-th G-G-1th element to the i-th element are 1 and the other elements are 0;
  • each node produces a rough DOA estimate.
  • each node gets an accurate DOA estimate based on the formula (0.48).
  • FIG. 3 is a schematic diagram of values obtained in each step of the DOA estimation method based on the gossip algorithm in the distributed wireless sensor network according to an embodiment of the present invention. Starting from the fifth cycle (t>t 4 ), the IR-Gossip algorithm begins to generate valid DOA estimates.
  • the present invention also provides a multi-objective DOA estimation system based on the gossip algorithm in a distributed network.
  • the system includes a loop iteration module 1 and a DOA estimate calculation module 2.
  • the loop iteration module 1 is configured to perform gossip iteration on the ⁇ th signal data vector according to the ⁇ th iteration rule in the ⁇ th iteration period, and after each iteration, determine whether the ⁇ th signal data vector is equal to before the iteration.
  • the initial value of ⁇ is 1, and the ⁇ signal data vector is constructed by using the initial values of the receiving nodes in the ⁇ iteration period;
  • the phase When the number of times reaches the preset number and the value of ⁇ does not reach the preset value, the iteration of the ⁇ th iteration period is completed, and the ⁇ th signal data vector at this time is stored, and the signals of the receiving nodes in the ⁇ th signal data vector are obtained.
  • the initial value of each receiving node in the ⁇ +1 iteration cycle the above steps are executed cyclically, and the value of ⁇ is increased by one from the value of ⁇ in the previous cycle.
  • the DOA estimation value calculation module 2 is configured to calculate a DOA estimation value by using a calculation formula of the DOA estimation value according to an iteration result of each iteration period when the value of ⁇ reaches a preset value; the preset value is according to the DOA estimation value.
  • the calculation formula calculates the number of iteration cycles required for the DOA estimate.
  • the working principle and working process of the system can refer to the above DOA estimation method, and will not be described again.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Optical Communication System (AREA)
  • Variable-Direction Aerials And Aerial Arrays (AREA)

Abstract

本发明涉及分布式网络中基于gossip算法的多目标DOA估计方法。本发明按照根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数,利用gossip算法对实现DOA估计的信号组成部分不断更新至一个信息完全共享的理想状态,执行完所有迭代周期后,可以得到DOA的良好估计。本发明不需要矩阵求逆的方法,即可实现分布式网络中信息共享及DOA值的估计。

Description

分布式网络中基于gossip算法的多目标DOA估计系统及估计方法 技术领域
本发明涉及一种分布式网络中基于gossip算法的多目标DOA估计系统及估计方法。
背景技术
在分布式网络中,由于不存在fusion center收集所有接收信号并进行处理,因此无法采用传统的DOA估计算法对目标进行参数估计。即使存在中心节点对信号进行处理,也需要耗费大量的传输代价,且系统的稳定性依赖于中心节点的稳定性。分布式网络中,已有算法一般将整个系统划分为多个子系统并在多个子系统中进行信号传递并进行参数估计,然而这对网络结构有一定的要求,且要求每个子系统中存在一个中心节点,算法的稳定性依然不高。
由于在传感器网络内部实现信息共享时不需要特定的路线,也不需要预先设置中心节点以避免出现由于中心节点崩溃使得整个网络崩溃的问题,在不稳定的传感器网络中算法性能也很稳定,gossip算法在近几年颇受关注,在计算机科学、控制、信号处理和信息理论领域都有gossip算法的应用。将gossip算法应用到分布式网络的DOA估计主要是在节点之间共享信息以得到每个节点的DOA估计。然而,利用原始的DOA算法如CAPON算法进行分布式的DOA估计时必须利用噪声相关信号替代接收信号的自相关矩阵。由于矩阵求逆不易实现,这在多目标的场景下会产生很多干扰。因此寻求不需要矩阵求逆的DOA估计算法并结合gossip算法进行分布式网络中的DOA估计意义重大。
发明内容
本发明所要解决的技术问题是:提出一种分布式网络中基于gossip算法的多目标DOA估计系统及估计方法,不需要矩阵求逆的方法,即可实现分布式网络中信息共享及DOA值的估计。本发明是这样实现的:
一种分布式网络中基于gossip算法的多目标DOA估计方法,包括如下步骤:
在第φ迭代周期中,根据第φ迭代规则对所述第φ信号数据向量进行gossip迭代,每次迭代后,判断所述第φ信号数据向量与迭代前是否相等,如果是,则记录并累加相等次数,否则将相等次数归零;其中,φ的初始值为1,所述第φ信号数据向量利用第φ迭代周期中各接收节点的初始值构建得到;
当所述相等次数达到预设次数,且φ的值未达到预设值时,完成第φ迭代周期的迭代,并储存此时的第φ信号数据向量,并将第φ信号数据向量中各接收节点的信号作为第φ+1迭代周期中各接收节点的初始值;
循环执行上述各步骤,每次循环时,φ的值比上一循环中φ的值增加1,当φ的值达到预设值时,根据各迭代周期的迭代结果,利用DOA估计值的计算公式计算DOA估计值;所述预设值为根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数。
一种分布式网络中基于gossip算法的多目标DOA估计系统,包括:
循环迭代模块,用于在第φ迭代周期中,根据第φ迭代规则对所述第φ信号数据向量进行gossip迭代,每次迭代后,判断所述第φ信号数据向量与迭代前是否相等,如果是,则记录并累加相等次数,否则将相等次数归零;其中,φ的初始值为1,所述第φ信号数据向量利用第φ迭代周期中各接收节点的初始值构建得到;当所述相等次数达到预设次数,且φ的值未达到预设值时,完成第φ迭代周期的迭代,并储存此时的第φ信号数据向量,并将第φ信号数据向量中各接收节点的信号作为第φ+1迭代周期中各接收节点的初始值;循环执行上述各步骤,每次循环时,φ的值比上一循环中φ的值增加1;
DOA估计值计算模块,用于当φ的值达到预设值时,根据各迭代周期的迭代结果,利用DOA估计值的计算公式计算DOA估计值;所述预设值为根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数。
与现有技术相比,本发明按照根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数,利用gossip算法对实现DOA估计的信号组成部分不断更新至一个信息完全共享的理想状态,执行完所有迭代周期后,可以得到DOA的良好估计。本发明不需要矩阵求逆的方法,即可实现分布式网络中信息共享及DOA值的估计。
附图说明
图1:本发明实施例提供的分布式网络中基于gossip算法的多目标DOA估计方法流程示意图;
图2:本发明实施例提供的分布式网络中基于gossip算法的多目标DOA估计系统组成示意图;
图3:本发明实施例提供的分布式网络中基于gossip算法的多目标DOA估计方法流程中各步骤得到的值示意图。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本发明进行进一步详细说明。应当理解,此处所描述的具体实施例仅用于解释本发明,并不用于限定本发明。
本发明以由三个发射阵元、三个接受阵元组成的无线通信分布式系统为例,说明采用gossip迭代算法结合AV算法实现DOA估计的过程。对于分布式的无线通信系统信息共享的实现,gossip方法是比较有效的方法,且对于分布式信号求自相关矩阵的逆运算,本发明采用不需要求矩阵逆运算也能得到最优权值的AV算法,最终实现分布式网络中的DOA估计。图1示出了本发明实施例提供的分布式网络中基于gossip算法的多目标DOA估计方法流程示意图;图2示出了本发明实施例提供的分布式网络中基于gossip算法的多目标DOA估计系统组成示意图。
首先对现有技术中的gossip算法进行详细说明,以便更清楚地阐述本发明的具体实施方案。
经典的随机gossip算法:
随机的gossip算法可以用来解决分布式的凸问题,假设给定一个随机的N节点网络和第i个节点的初始标量值。随机gossip算法的目的在于通过仅使用局部信息和局部处理和一种迭代机制来实现所有目标端达到一个均值。假设
g(t)=[g1(t),...,gN(t)]T(0.1)(注,“(0.1)”表示该公式的编号,并不是该公式的一部分,后续各公式同理。)
表示第t次迭代后的每个节点的值组成的向量。第t次迭代过程中,每个节点运行一个独立的泊松时钟,当第i个节点的时钟响起时,该节点以概率pi,j随机选择一个邻近的j节点并与之通信。所有两两节点之间的概率pi,j可以组成一个N×N的概率矩阵p。如果第i个节点与第j个节点之间能够通信,则pi,j>0,否则pi,j=0。每次迭代,节点i和j交换它们的局部信息并将它们的当前局部信息更新为gi(t)=gj(t)=(gi(t-1)+gj(t-1))/2,除了这些活跃的节点,网络中其他节点保持它们上一次迭代后得到的信息不变。Gossip算法的一般向量表达形式为
g(t)=U(t)g(t-1)    (0.2)
其中U(t)是每个时间段独立选择的随机N×N的更新矩阵,第t次迭代过程中对于2个通信节点i和j的更新矩阵为
Figure PCTCN2015078172-appb-000001
其中ei=[0,...,0,1,0,...,0]T为第i个元素为1的N维向量。当U(t)是双重倒向随机矩阵且网络联通时,能够确保网络中的所有节点能够收敛到均值gave。注意,在gossip算法中,最重要的任务是定义所有节点的初始向量g(0)。
无线传感器网络的信号模型:
假设无线传感器网络(WSN)中有Mt个发射节点和Nr个接收节点,且它们均匀分布在一个半径为r的小区域内。为简单起见,假设目标和节点处在同一个平面且无杂波干扰。并且假设已知节点的位置信息且相位完全同步,
Figure PCTCN2015078172-appb-000002
Figure PCTCN2015078172-appb-000003
分别表示极坐标中第i个发射节点和第j个接收节点的坐标信息。假设系统中有K个节点,且第k个节点方位角为θk且以固定的速度vk移动。目标的距离为dk(t)=dk(0)-vkt,其中dk(0)是目标在0时刻与原点之间初始距离。在远场假设下,
Figure PCTCN2015078172-appb-000004
因此第i个发射、接收节点与目标之间的距离
Figure PCTCN2015078172-appb-000005
可以表示如下
Figure PCTCN2015078172-appb-000006
其中,
Figure PCTCN2015078172-appb-000007
假设第i个发射节点连续时间发射波形表示为xi(t)ej2πft,其中f为载波频率且所有发射节点使用相同的载波频率,xi(t)为以Tp为周期窄带信号。
第k个目标端的接收信号可以表示为
Figure PCTCN2015078172-appb-000008
k,k=1,...,K}是第k个目标的反射系数复幅度,且对于所有接收节点都是一致的。后者的假设是基于远场假设,即网络节点之间的距离远远小于节点与目标之间的距离
Figure PCTCN2015078172-appb-000009
因此,由于节点之间相隔较近,可以视为所有接收节点看到目标的同一表面。
由于目标反射,第l个接收端接收到的信号表示如下
Figure PCTCN2015078172-appb-000010
其中εl(t)表示独立同分布,均值为0,方差为σ2的高斯噪声。
对于目标分布在一个小区域内,采样信号可以看成是第一个目标反射回来的信息的同步信号,且由于发射波形是窄带信号,可以忽略发射波形xi(t)中的延时,只需要考虑相位部分的延时即可。因此,第l个接收端的接收基带信号可以近似表示为
Figure PCTCN2015078172-appb-000011
其中λ是发射信号波长,fk=2vkf/c是第k个目标产生的多普勒平移,
Figure PCTCN2015078172-appb-000012
Figure PCTCN2015078172-appb-000013
假设L为波形的长度,lTs,l=0,...,L-1表示脉冲内的时间,T表示脉冲重复间隔,接收端在第m个脉冲上的采样信号表示为:
Figure PCTCN2015078172-appb-000014
其中:
Figure PCTCN2015078172-appb-000015
Figure PCTCN2015078172-appb-000016
εlm=[εl((m-1)T+0Ts),...εl((m-1)T+(L-1)Ts)]T  (0.13)
X=[x(0Ts),...,x((L-1)Ts)]T(L×Mt)  (0.14)
在此,作如下两种假设:
目标移动非常缓慢,因此,一个脉冲内的多普勒频移可以忽略不计,即对于k=1,...,K有fkTp>>1,其中Tp为脉冲持续时间。
每个发射天线的发射波形是独立的,因此,相对
Figure PCTCN2015078172-appb-000017
来说,i≠i′时
Figure PCTCN2015078172-appb-000018
是可以忽略不计的。
传统的集中式DOA估计:
假设目标是固定的,因此只需要考虑一个脉冲内的数据,因此第l个节点的接收信号简化表示如下:
Figure PCTCN2015078172-appb-000019
将Nr个接收节点的信号放在一个矩阵里
Figure PCTCN2015078172-appb-000020
其中
Figure PCTCN2015078172-appb-000021
传统的CAPON算法产生可以抑制噪声的波束合成向量w,干扰和噪声被抑制的同时,期望的信号保持不失真。特别地,w可以表示如下:
Figure PCTCN2015078172-appb-000022
其中R=ZZH,公式(0.17)的解可以表示如下:
Figure PCTCN2015078172-appb-000023
通过w*将LS方法应用到波束合成输出,基于假设第三点可以很容易得到目标反射系数的估计为如下所示:
Figure PCTCN2015078172-appb-000024
其中Rx=XTX*
传统的Auxiliary Vector(AV)技术:
传统的capon算法中,为了获得公式(0.18)中的最优权值,需要进行矩阵求逆的操作,但是在分布式信号处理中,矩阵求逆是不容易实现的。因此可采用另外一种不需要矩阵求逆的方法来获得最优权值向量,即auxiliary vector(AV)技术。传统的AV算法主要适用于天线阵列的空时滤波,可以直接运用到DOA估计问题上来。
首先,不失一般性,假设vr(θ)是归一化的,即
Figure PCTCN2015078172-appb-000025
此时,考虑任意一个与vr(θ)相互正交的固定辅助向量G(θ)
G(θ)Hvr(θ)=0
G(θ)HG(θ)=1  (0.20)
基于AV技术的最优权值向量可以表示为
wAV(θ)=vr(θ)-μ(θ)G(θ)  (0.21)
使得输出的波束合成权值向量WAV(θ)最小的复标量μ(θ)的值为
Figure PCTCN2015078172-appb-000026
对于AV技术,G(θ)的选择原则为能够使得vr(θ)处理数据
Figure PCTCN2015078172-appb-000027
和AV处理数据G(θ)HZ的互相关函数的幅度最大化。同时需要满足(0.20)的条件
Figure PCTCN2015078172-appb-000028
s.t.G(θ)Hvr(θ)=0and G(θ)HG(θ)=1      (0.23)
对于该准则的物理直观的解释,可以说与vr(θ)相互正交,使得
Figure PCTCN2015078172-appb-000029
最大的AVG(θ)可以提取出大部分波束合成输出的扰动成分,最优的AVG(θ)可以根据下式获得
Figure PCTCN2015078172-appb-000030
单个的AV G(θ)通常表示一个自由度,如果需要提高分辨率,可以采用多个auxiliary向量。假设有P个相互正交的AV G1(θ),G2(θ),...,GP(θ)构成的集合,且它们均与vr(θ)相互正交,从而,整体的波束合成权值向量可以表示如下
Figure PCTCN2015078172-appb-000031
其中
Figure PCTCN2015078172-appb-000032
Figure PCTCN2015078172-appb-000033
注意,为了简化起见,只关注公式(0.22)(0.24)单个AVG(θ)技术,但是可以直接扩展到多个AV G(θ)技术。
分布式网络中基于gossip算法的单目标DOA估计方法:
假设只有一个目标,WSN的信号模型可以简化为
Figure PCTCN2015078172-appb-000034
假设εi(i=1,...,Nr)为零均值功率谱密度为
Figure PCTCN2015078172-appb-000035
的空间不相关与目标也不相关的噪声。从而可以得到
Figure PCTCN2015078172-appb-000036
R=RSS+REE  (0.30)
其中RSS=β1vr1)vT1)XT1vr1)vT1)XT)H,利用矩阵求逆原理,公式(0.18)的最优解表示为
Figure PCTCN2015078172-appb-000037
目标反射系数的估计值变为
Figure PCTCN2015078172-appb-000038
假设将角度空间以间隔Δθ均匀离散化θG=[θ1,...,θG],意味着每个接收节点在一次gossip算法开始之前需要计算G个角度估计。公式(0.32)的分子和分母可以表示为
Figure PCTCN2015078172-appb-000039
Figure PCTCN2015078172-appb-000040
目标反射系数的估计为
Figure PCTCN2015078172-appb-000041
假设WSN中每个接收节点i在每个给定的时隙内有两个初始值
Figure PCTCN2015078172-appb-000042
Figure PCTCN2015078172-appb-000043
假设每个接收节点已知所有发射节点的位置信息
Figure PCTCN2015078172-appb-000044
和发射波形xm。噪声方差
Figure PCTCN2015078172-appb-000045
可以估计出来。用Nr维向量
Figure PCTCN2015078172-appb-000046
表示为Nr节点的初始值
Figure PCTCN2015078172-appb-000047
类似的,所有的
Figure PCTCN2015078172-appb-000048
(i=1,...,Nr)存放在一个Nr维向量
Figure PCTCN2015078172-appb-000049
中。可以轻易得到
Figure PCTCN2015078172-appb-000050
其中1表示全1向量。从而本节算法的目的就是要寻找分布式系统中的平均的
Figure PCTCN2015078172-appb-000051
Figure PCTCN2015078172-appb-000052
值。假设第t次迭代的
Figure PCTCN2015078172-appb-000053
分别表示为向量
Figure PCTCN2015078172-appb-000054
Gossip DOA估计方法在第t次迭代的估计结果的一般表达式为
Figure PCTCN2015078172-appb-000055
Figure PCTCN2015078172-appb-000056
Figure PCTCN2015078172-appb-000057
Figure PCTCN2015078172-appb-000058
表示的是第i个接收节点的估计输出θg(g=1,...,G)。注意,每次迭代,一对随机节点的G个格点信息相互交换。从而可以重新定义新的更新矩阵:
Figure PCTCN2015078172-appb-000059
其中
Figure PCTCN2015078172-appb-000060
是N1维单位阵,
Figure PCTCN2015078172-appb-000061
是从第(iN2-N2+1)个到iN2元素等于1其他元素等于0的N1维向量。Gossip DOA估计算法的表达式可以重新表示为
Figure PCTCN2015078172-appb-000062
Figure PCTCN2015078172-appb-000063
Figure PCTCN2015078172-appb-000064
综上所述,可总结出分布式网络中基于gossip算法的单目标DOA估计方法的基本技术思想如下:
各节点共同发射信号,同时,各节点接收信号,并根据接收到的信号构建初始信 号,所述初始信号表示为
Figure PCTCN2015078172-appb-000065
其中,i为节点的序号,θ为角度;
将所有
Figure PCTCN2015078172-appb-000066
存放在Nr维向量
Figure PCTCN2015078172-appb-000067
中,据此构建第一信号数据向量,将所有
Figure PCTCN2015078172-appb-000068
存放在Nr维向量
Figure PCTCN2015078172-appb-000069
中,据此构建第二信号数据向量,其中,i=1,...,Nr,Nr为节点个数;
根据
Figure PCTCN2015078172-appb-000070
对第一信号数据向量进行迭代,每次迭代后,判断所述第一信号数据向量是否与迭代前相等,如果相等,则记录并累加相应的相等次数,否则将相应相等次数归零,当相等次数达到预设次数时,停止迭代并存储此时的第一信号数据向量
Figure PCTCN2015078172-appb-000071
根据
Figure PCTCN2015078172-appb-000072
对第二信号数据向量进行迭代,每次迭代后,判断所述第二信号数据向量是否与迭代前相等,如果相等,则记录并累加相应相等次数,否则将相应相等次数归零,当相等次数达到预设次数时,停止迭代并存储此时的第二信号数据向量
Figure PCTCN2015078172-appb-000073
其中,t为迭代次数;
根据停止迭代后存储的第一信号数据向量
Figure PCTCN2015078172-appb-000074
及第二信号数据向量
Figure PCTCN2015078172-appb-000075
利用公式
Figure PCTCN2015078172-appb-000076
计算DOA估计值,其中,
Figure PCTCN2015078172-appb-000077
为DOA估计值。
其中,
Figure PCTCN2015078172-appb-000078
其中,λ表示发射信号的波长,
Figure PCTCN2015078172-appb-000079
表示信号从发射节点经角度为θ的目标反射到达第i接受节点的近似距离,
Figure PCTCN2015078172-appb-000080
为所有发射节点的位置信息,zi(l-1)表示第i节点在第l-1采样点的接收信号,L表示采样点数目,xm为发射波形。
Figure PCTCN2015078172-appb-000081
其中,
Figure PCTCN2015078172-appb-000082
是N1维单位阵,
Figure PCTCN2015078172-appb-000083
是从第(iN2-N2+1)个到iN2元素等于1其他元素等于0的N1维向量,N1为NrG,N2为G。
分布式网络中基于gossip算法的多目标DOA估计方法:
前面介绍的gossip估计方法可以认为是分布式时延和波束和的一个扩展,只是用来接收空间信号。但是,最主要的缺点是公式(0.18)中利用REE代替R。假设系统中有多个目标或者是波形的长度L不够长,其性能将严重退化。为了解决这个问题,这里提出了一种利用AV技术的迭代的随机gossip算法(IR-Gossip)(0.21)。
将公式(0.22)代入公式(0.21),可以得到
Figure PCTCN2015078172-appb-000084
假设
Figure PCTCN2015078172-appb-000085
则将(0.24)代入(0.46)并作简化处理,得
Figure PCTCN2015078172-appb-000086
则IR-Gossip算法目标反射系数的估计值为
Figure PCTCN2015078172-appb-000087
其中
Figure PCTCN2015078172-appb-000088
Figure PCTCN2015078172-appb-000089
Figure PCTCN2015078172-appb-000090
如果假设
Figure PCTCN2015078172-appb-000091
γ6(θ)=vT(θ)Rxv*(θ),(0.49)(0.50)(0.51)可以变化为:
Figure PCTCN2015078172-appb-000092
Figure PCTCN2015078172-appb-000093
Figure PCTCN2015078172-appb-000094
由于
Figure PCTCN2015078172-appb-000095
假设WSN中每个接收节点i对于一个给定的时间间隔内有初始值
Figure PCTCN2015078172-appb-000096
将所有的
Figure PCTCN2015078172-appb-000097
i=1,...,Nr存在一个Nr维向量
Figure PCTCN2015078172-appb-000098
中且对于第t次迭代的一般形式由下式给出
Figure PCTCN2015078172-appb-000099
通过一定次数t1的迭代,
Figure PCTCN2015078172-appb-000100
达到一个稳定状态
Figure PCTCN2015078172-appb-000101
此时为下一次循环定义一个初始值
Figure PCTCN2015078172-appb-000102
将所有的
Figure PCTCN2015078172-appb-000103
i=1,...,Nr存在一个Nr维向量
Figure PCTCN2015078172-appb-000104
中,第t次迭代的一般形式由下式给出
Figure PCTCN2015078172-appb-000105
通过一定次数t2-t1迭代,
Figure PCTCN2015078172-appb-000106
达到一个稳定的状态
Figure PCTCN2015078172-appb-000107
可以得到
Figure PCTCN2015078172-appb-000108
从式(0.55)到(0.60)可以看出,为了得到γ1(θ),gossip算法需要两个顺序循环。第一个循环得到
Figure PCTCN2015078172-appb-000109
第二个循环得到
Figure PCTCN2015078172-appb-000110
因此,需要设定一个门限CT,确定每个 节点当前的状态是否不再改变,即
Figure PCTCN2015078172-appb-000111
当计数器变量C>CT,该节点将进入下一个循环。本发明中定义CT=ρNr
其中ρ是按照经验设定的。注意ρ较小时算法可以较快收敛,ρ较大时,提出的IR-Gossip算法收敛较慢,但是一般都能达到稳定的状态。
类似地,由于
Figure PCTCN2015078172-appb-000112
需要三个gossip循环才能得到γ2(θ)。第一个循环得到
Figure PCTCN2015078172-appb-000113
第二个循环,得到t1的一个新的初始值
Figure PCTCN2015078172-appb-000114
通过一些迭代次数t2-t1,节点获得稳定状态
Figure PCTCN2015078172-appb-000115
在第三个循环,可以得到t2的新的初始值,
Figure PCTCN2015078172-appb-000116
通过一定迭代次数t3-t2,节点达到稳定状态
Figure PCTCN2015078172-appb-000117
得到
Figure PCTCN2015078172-appb-000118
由于
Figure PCTCN2015078172-appb-000119
为了获得γ3(θ)需要四次gossip循环。第一个循环,得到
Figure PCTCN2015078172-appb-000120
第二个循环得到
Figure PCTCN2015078172-appb-000121
第三个循环得到新的t2的初始值
Figure PCTCN2015078172-appb-000122
通过一定迭代次数t3-t2,节点达到稳定状态
Figure PCTCN2015078172-appb-000123
第四个循环,得到新的t3的初始值
Figure PCTCN2015078172-appb-000124
通过一定迭代次数t4-t3,节点达到稳定状态
Figure PCTCN2015078172-appb-000125
可以得到
Figure PCTCN2015078172-appb-000126
由于
Figure PCTCN2015078172-appb-000127
假设每个节点i在t4时有初始值
Figure PCTCN2015078172-appb-000128
通过一些迭代t5-t4,节点达到一个稳定状态
Figure PCTCN2015078172-appb-000129
随后可以得到
Figure PCTCN2015078172-appb-000130
由于
Figure PCTCN2015078172-appb-000131
需要两个gossip顺序循环来得到γ5(θ)。第一个循环,可以得到t4的新的初始值
Figure PCTCN2015078172-appb-000132
经过一定迭代次数t5-t4,节点达到一个稳定状态
Figure PCTCN2015078172-appb-000133
第二个循环可以得到t5的新的初始值
Figure PCTCN2015078172-appb-000134
通过一定迭代次数t6-t5,接收节点达到稳定状态
Figure PCTCN2015078172-appb-000135
可以得到
Figure PCTCN2015078172-appb-000136
由于
Figure PCTCN2015078172-appb-000137
可以直接获得γ6(θ),不需要循环操作。
综上所述,可归纳总结出本发明提供的分布式网络中基于gossip算法的多目标DOA估计方法的基本技术思想。如图1所示,该DOA估计方法包括如下步骤:
步骤S1:在第φ迭代周期中,根据第φ迭代规则对所述第φ信号数据向量进行gossip迭代,每次迭代后,判断所述第φ信号数据向量与迭代前是否相等,如果是,则记录并累加相等次数,否则将相等次数归零;其中,φ的初始值为1,所述第φ信号数据向量利用第φ迭代周期中各接收节点的初始值构建得到;
步骤S2:当所述相等次数达到预设次数,且φ的值未达到预设值时,完成第φ迭代周期的迭代,并储存此时的第φ信号数据向量,并将第φ信号数据向量中各接收节点的信号作为第φ+1迭代周期中各接收节点的初始值;
步骤S3:循环执行上述各步骤,每次循环时,φ的值比上一循环中φ的值增加1,当φ的值达到预设值时,根据各迭代周期的迭代结果,利用DOA估计值的计算公式计算DOA估计值;所述预设值为根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数。
进一步地,所述预设值为6。
以下是对预设值为6时,上述基本技术思想的细节表达:
各接收节点接收初始信号zi(l-1),并根据
Figure PCTCN2015078172-appb-000138
构建第一迭代周期中各节点的初始值
Figure PCTCN2015078172-appb-000139
其中,λ表示发射信号的波长,
Figure PCTCN2015078172-appb-000140
表示角度为θ的目标与第i接收节点的近似距离zi(l-1)表示第i节点在第l-1采样点的接收信号,L表示采样点数目,[ ]*表示共轭操作;
将第一迭代周期中各接收节点的初始值
Figure PCTCN2015078172-appb-000141
存储在NrG维向量
Figure PCTCN2015078172-appb-000142
中,形成第 一信号数据向量;其中,Nr为接收节点个数,G为角度空间的离散化精度;
根据第一迭代规则
Figure PCTCN2015078172-appb-000143
对第一信号数据向量进行迭代;其中,
Figure PCTCN2015078172-appb-000144
表示gossip更新矩阵,
Figure PCTCN2015078172-appb-000145
表示NrG的单位矩阵,eGi表示第iG-G-1个元素至第iG个元素为1而其他元素都为0的一个G维向量,t表示迭代次数;
每次迭代后,判断是否
Figure PCTCN2015078172-appb-000146
如果是,则记录并累加相等次数C,否则,将相等次数C归零;其中,
Figure PCTCN2015078172-appb-000147
Figure PCTCN2015078172-appb-000148
分别为第i个接收节点在第t和第t-1次迭代得到的值,θ为目标的角度,C为计数器变量;
当相等次数CT达到预设次数CT时,储存当前每个接收节点的信号
Figure PCTCN2015078172-appb-000149
并将其作为第二迭代周期中各接收节点的初始值;其中:CT=ρNr,ρ是预设的常数;
Figure PCTCN2015078172-appb-000150
其中,t1表示实现第一迭代周期信息共享所耗费的迭代次数,Nr为接收节点个数,λ表示发射信号的波长,角度为θ的目标与第i接收节点的近似距离,zj(l-1)表示第i节点在采样点l-1的接收信号,L表示采样点数目;
如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
Figure PCTCN2015078172-appb-000151
并进入本次迭代周期内的下一次gossip循环。
根据
Figure PCTCN2015078172-appb-000152
计算第二迭代周期中各接收节点的新的初始值
Figure PCTCN2015078172-appb-000153
利用第二迭代周期中各接收节点的新的初始值
Figure PCTCN2015078172-appb-000154
构成新的L维初始向量:
Figure PCTCN2015078172-appb-000155
将各接收节点的
Figure PCTCN2015078172-appb-000156
存储在NrGL维向量
Figure PCTCN2015078172-appb-000157
中,形成第二信号数据向量;
根据第二迭代规则
Figure PCTCN2015078172-appb-000158
对第二信号数据向量进行迭代;其中,
Figure PCTCN2015078172-appb-000159
表示gossip更新矩阵,
Figure PCTCN2015078172-appb-000160
表示NrG(L+1)的单位矩阵,eG(L+1)i表示第iG(L+1)-G(L+1)-1个元素至第iG(L+1)个元素为1而其他元素都为0的一个G(L+1)维向量;
每次迭代后,判断是否
Figure PCTCN2015078172-appb-000161
如果是,则记录并累加相等次数C,否则,将相等次数C归零;其中,
Figure PCTCN2015078172-appb-000162
Figure PCTCN2015078172-appb-000163
分别为第i个接收节点在第t和第t-1次迭代得到的值;
当相等次数C达到预设次数CT时,根据
Figure PCTCN2015078172-appb-000164
计算γ1(θ),并储存当前每个接收节点的输出
Figure PCTCN2015078172-appb-000165
并将其作为第三迭代周期中各接收节点的初始值;其中:
Figure PCTCN2015078172-appb-000166
t2为实现第二迭代周期信息共享所耗费的迭代次数;
如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
Figure PCTCN2015078172-appb-000167
并进入本次迭代周期内的下一次gossip循环。
根据
Figure PCTCN2015078172-appb-000168
得到
Figure PCTCN2015078172-appb-000169
根据
Figure PCTCN2015078172-appb-000170
得到第三迭代周期中各接收节点的新的初始值
Figure PCTCN2015078172-appb-000171
根据第三迭代周期中各接收节点的新的初始值
Figure PCTCN2015078172-appb-000172
构成新的L维初始向量:
Figure PCTCN2015078172-appb-000173
将各接收节点的
Figure PCTCN2015078172-appb-000174
存储在NrGL维向量
Figure PCTCN2015078172-appb-000175
中,形成第三信号数据向量;
根据第三迭代规则
Figure PCTCN2015078172-appb-000176
对第三信号数据向量进行迭代;其中,
Figure PCTCN2015078172-appb-000177
表示gossip更新矩阵,
Figure PCTCN2015078172-appb-000178
表示NrG(L+1)的单位矩阵,eG(L+1)i表示第iG(L+1)-G(L+1)-1个元素至第iG(L+1)个元素为1而其他元素都为0的一个G(L+1)维向量;
每次迭代后,判断是否
Figure PCTCN2015078172-appb-000179
如果是,则记录并累加相等次数C,否则,将相等次数C归零;
当相等次数C达到预设次数CT时,根据
Figure PCTCN2015078172-appb-000180
计算γ2(θ),并存储当前每个接收节点的信号
Figure PCTCN2015078172-appb-000181
并将其作为第四迭代周期中各接收节点的初始值;其中,
Figure PCTCN2015078172-appb-000182
t3为实现第三迭代周期信息共享所耗费的迭代次数;
如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
Figure PCTCN2015078172-appb-000183
并进入本次迭代周期内的下一次gossip循环。
根据
Figure PCTCN2015078172-appb-000184
计算第四迭代周期中各接收节点的新的初始值
Figure PCTCN2015078172-appb-000185
将第四迭代周期中各接收节点的新的初始值
Figure PCTCN2015078172-appb-000186
存储在NrG维向量
Figure PCTCN2015078172-appb-000187
中,形成第四信号数据向量;
根据第四迭代规则
Figure PCTCN2015078172-appb-000188
对所述第四信号数据向量进行迭代;其中,
Figure PCTCN2015078172-appb-000189
表示gossip更新矩阵,
Figure PCTCN2015078172-appb-000190
表示NrG的单位矩阵,eGi表示第iG-G-1个元素至第iG个元素为1而其他元素都为0的一个G维向量,t表示迭代次数;
每次迭代后,判断是否
Figure PCTCN2015078172-appb-000191
如果是,则记录并累加相等次数C,否则,将相等次数C归零;
当C>CT=ρNr时,根据
Figure PCTCN2015078172-appb-000192
得到γ3(θ),并存储当前每个接收节点的γ1(θ),γ2(θ),γ3(θ),并将存储的当前每个接收节点的γ1(θ),γ2(θ),γ3(θ)作为第五迭代周期中各接收节点的初始值;其中:
Figure PCTCN2015078172-appb-000193
由第二迭代周期结束时获得;
Figure PCTCN2015078172-appb-000194
由第三迭代周期结束时获得;
Figure PCTCN2015078172-appb-000195
由第四迭代周期结束时获得,t4为实现第四迭代周期信息共享所耗费的迭代次数;
如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
Figure PCTCN2015078172-appb-000196
并进入本次迭代周期内的下一次gossip循环。
根据
Figure PCTCN2015078172-appb-000197
计算
Figure PCTCN2015078172-appb-000198
根据
Figure PCTCN2015078172-appb-000199
计算
Figure PCTCN2015078172-appb-000200
根据
Figure PCTCN2015078172-appb-000201
计算γ6(θ);其中,L为采样点的个数,Mt为发射节点个数,xm为第m发射节点的发射信号,Rx为发射信号的自相关矩阵,
Figure PCTCN2015078172-appb-000202
λ表示发射信号的波长,
Figure PCTCN2015078172-appb-000203
表示第m发射节点与角度为θ的目标之间的近似距离,
Figure PCTCN2015078172-appb-000204
表示角度为θ的目标与第i接收节点的近似距离,[ ]*表示共轭操作,[ ]T表示转置操作;
利用
Figure PCTCN2015078172-appb-000205
和γ6(θ),构成新的L维初始向量:
Figure PCTCN2015078172-appb-000206
其中:
Figure PCTCN2015078172-appb-000207
Figure PCTCN2015078172-appb-000208
Figure PCTCN2015078172-appb-000209
Figure PCTCN2015078172-appb-000210
存储在NrGL维向量中,形成第五信号数据向量;
根据第五迭代规则
Figure PCTCN2015078172-appb-000212
对所述第五信号数据向量进行迭代;其中,
Figure PCTCN2015078172-appb-000213
表示gossip更新矩阵,
Figure PCTCN2015078172-appb-000214
表示NrGL的单位矩阵,eGLi表示第iGL-GL-1个元素至第iGL个元素为1而其他元素都为0的一个GL维向量;
每次迭代后,判断是否
Figure PCTCN2015078172-appb-000215
如果是,则记录并累加相等次数C,否则,将相等次数C归零;
当C>CT=ρNr时,根据
Figure PCTCN2015078172-appb-000216
计算γ4(θ);并存储当前每个接收节点的信号
Figure PCTCN2015078172-appb-000217
并将其作为第六迭代周期中各接收节点的初始值,t5为实现第五迭代周期信息共享所耗费的迭代次数;
如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
Figure PCTCN2015078172-appb-000218
并进入本次迭代周期内的下一次gossip循环。该迭代周期结束时,每个节点产生一个粗略的DOA估计值。
根据公式
Figure PCTCN2015078172-appb-000219
得到
Figure PCTCN2015078172-appb-000220
根据
Figure PCTCN2015078172-appb-000221
构成新的L维向量:
Figure PCTCN2015078172-appb-000222
其中:
Figure PCTCN2015078172-appb-000223
Figure PCTCN2015078172-appb-000224
Figure PCTCN2015078172-appb-000225
存储在NrG维的向量
Figure PCTCN2015078172-appb-000226
中,形成第六信号数据向量;
根据第六迭代规则
Figure PCTCN2015078172-appb-000227
对所述第六信号数据向量进行迭代;其中,
Figure PCTCN2015078172-appb-000228
表示gossip更新矩阵,
Figure PCTCN2015078172-appb-000229
表示NrG的单位矩阵,eGi表示第iG-G-1个元素至第iG个元素为1而其他元素都为0的一个G维向量;
每次迭代后,判断是否
Figure PCTCN2015078172-appb-000230
如果是,则记录并累加相等次数C,否则,将相等次数C归零;
当C>CT=ρNr时,根据
Figure PCTCN2015078172-appb-000231
得到γ5(θ);并计算DOA估计值;计算公式为:
Figure PCTCN2015078172-appb-000232
其中:
Figure PCTCN2015078172-appb-000233
Figure PCTCN2015078172-appb-000234
Figure PCTCN2015078172-appb-000235
其中,
Figure PCTCN2015078172-appb-000236
如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
Figure PCTCN2015078172-appb-000237
并进入本次迭代周期内的下一次gossip循环。
这个循环的最开始,每个节点产生一个粗略的DOA估计,
Figure PCTCN2015078172-appb-000238
该迭代周期结束后,每个节点根据公式(0.48)得到一个准确的DOA估计值。
IR-Gossip算法需要6个循环来实现分布式信号中的AV技术。图3是本发明实施例提供的分布式无线传感器网络中基于gossip算法的DOA估计方法流程中各步骤得到的值示意图。从第五个循环开始(t>t4),IR-Gossip算法开始产生有效的DOA估计值。
根据本发明所提供的分布式网络中基于gossip算法的多目标DOA估计方法,本发明还提供了一种分布式网络中基于gossip算法的多目标DOA估计系统。根据图2所示,该系统包括循环迭代模块1及DOA估计值计算模块2。
其中,循环迭代模块1用于在第φ迭代周期中,根据第φ迭代规则对所述第φ信号数据向量进行gossip迭代,每次迭代后,判断所述第φ信号数据向量与迭代前是否相等,如果是,则记录并累加相等次数,否则将相等次数归零;其中,φ的初始值为1,所述第φ信号数据向量利用第φ迭代周期中各接收节点的初始值构建得到;当所述相 等次数达到预设次数,且φ的值未达到预设值时,完成第φ迭代周期的迭代,并储存此时的第φ信号数据向量,并将第φ信号数据向量中各接收节点的信号作为第φ+1迭代周期中各接收节点的初始值;循环执行上述各步骤,每次循环时,φ的值比上一循环中φ的值增加1。
DOA估计值计算模块2用于当φ的值达到预设值时,根据各迭代周期的迭代结果,利用DOA估计值的计算公式计算DOA估计值;所述预设值为根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数。
该系统的工作原理及工作过程可参照上述DOA估计方法,再次不再赘述。
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等,均应包含在本发明的保护范围之内。

Claims (10)

  1. 一种分布式网络中基于gossip算法的多目标DOA估计方法,其特征在于,包括如下步骤:
    在第φ迭代周期中,根据第φ迭代规则对所述第φ信号数据向量进行gossip迭代,每次迭代后,判断所述第φ信号数据向量与迭代前是否相等,如果是,则记录并累加相等次数,否则将相等次数归零;其中,φ的初始值为1,所述第φ信号数据向量利用第φ迭代周期中各接收节点的初始值构建得到;
    当所述相等次数达到预设次数,且φ的值未达到预设值时,完成第φ迭代周期的迭代,并储存此时的第φ信号数据向量,并将第φ信号数据向量中各接收节点的信号作为第φ+1迭代周期中各接收节点的初始值;
    循环执行上述各步骤,每次循环时,φ的值比上一循环中φ的值增加1,当φ的值达到预设值时,根据各迭代周期的迭代结果,利用DOA估计值的计算公式计算DOA估计值;所述预设值为根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数。
  2. 如权利要求1所述的多目标DOA估计方法,其特征在于,所述预设值为6。
  3. 如权利要求2所述的多目标DOA估计方法,其特征在于,当φ=1时,所述D0A估计方法包括如下步骤:
    各接收节点接收初始信号zi(l-1),并根据
    Figure PCTCN2015078172-appb-100001
    构建第一迭代周期中各节点的初始值
    Figure PCTCN2015078172-appb-100002
    其中,λ表示发射信号的波长,
    Figure PCTCN2015078172-appb-100003
    表示角度为θ的目标与第i接收节点的近似距离zi(l-1)表示第i节点在第l-1采样点的接收信号,L表示采样点数目,[]*表示共轭操作;
    将第一迭代周期中各接收节点的初始值
    Figure PCTCN2015078172-appb-100004
    存储在NrG维向量
    Figure PCTCN2015078172-appb-100005
    中,形成第一信号数据向量;其中,Nr为接收节点个数,G为角度空间的离散化精度;
    根据第一迭代规则
    Figure PCTCN2015078172-appb-100006
    对第一信号数据向量进行迭代;其中,
    Figure PCTCN2015078172-appb-100007
    表示gossip更新矩阵,
    Figure PCTCN2015078172-appb-100008
    表示NrG的单位矩阵,eGi表示第iG-G-1个元素至第iG个元素为1而其他元素都为0的一个G维向量,t表示迭代次数;
    每次迭代后,判断是否
    Figure PCTCN2015078172-appb-100009
    如果是,则记录并累加相等次数C,否则,将相等次数C归零;其中,
    Figure PCTCN2015078172-appb-100010
    Figure PCTCN2015078172-appb-100011
    分别为第i个接收节点在第t和第t-1次迭代得到的值,θ为目标的角度,C为计数器变量;
    当相等次数CT达到预设次数CT时,储存当前每个接收节点的信号
    Figure PCTCN2015078172-appb-100012
    并将其作为第二迭代周期中各接收节点的初始值;其中:CT=ρNr,ρ是预设的常数;
    Figure PCTCN2015078172-appb-100013
    其中,t1表示实现第一迭代周期信息共 享所耗费的迭代次数,Nr为接收节点个数,λ表示发射信号的波长,角度为θ的目标与第i接收节点的近似距离,zi(l-1)表示第i节点在采样点l-1的接收信号,L表示采样点数目;
    如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
    Figure PCTCN2015078172-appb-100014
    并进入本次迭代周期内的下一次gossip循环。
  4. 如权利要求3所述的多目标DOA估计方法,其特征在于,当φ=2时,所述DOA估计方法包括如下步骤:
    根据
    Figure PCTCN2015078172-appb-100015
    计算第二迭代周期中各接收节点的新的初始值
    Figure PCTCN2015078172-appb-100016
    利用第二迭代周期中各接收节点的新的初始值
    Figure PCTCN2015078172-appb-100017
    构成新的L维初始向量:
    Figure PCTCN2015078172-appb-100018
    将各接收节点的
    Figure PCTCN2015078172-appb-100019
    存储在NrGL维向量
    Figure PCTCN2015078172-appb-100020
    中,形成第二信号数据向量;
    根据第二迭代规则
    Figure PCTCN2015078172-appb-100021
    对第二信号数据向量进行迭代;其中,
    Figure PCTCN2015078172-appb-100022
    表示gossip更新矩阵,
    Figure PCTCN2015078172-appb-100023
    表示NrG(L+1)的单位矩阵,eG(L+1)i表示第iG(L+1)-G(L+1)-1个元素至第iG(L+1)个元素为1而其他元素都为0的一个G(L+1)维向量;
    每次迭代后,判断是否
    Figure PCTCN2015078172-appb-100024
    如果是,则记录并累加相等次数C,否则,将相等次数C归零;其中,
    Figure PCTCN2015078172-appb-100025
    Figure PCTCN2015078172-appb-100026
    分别为第i个接收节点在第t和第t-1次迭代得到的值;
    当相等次数C达到预设次数CT时,根据
    Figure PCTCN2015078172-appb-100027
    计算γ1(θ),并储存当前每个接收节点的输出
    Figure PCTCN2015078172-appb-100028
    并将其作为第三迭代周期中各接收节点的初始值;其中:
    Figure PCTCN2015078172-appb-100029
    t2为实现第二迭代周期信息共享所耗费的迭代次数;
    如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
    Figure PCTCN2015078172-appb-100030
    并进入本次迭代周期内的下一次gossip循环。
  5. 如权利要求4所述的多目标DOA估计方法,其特征在于,当φ=3时,所述DOA估计方法包括如下步骤:
    根据
    Figure PCTCN2015078172-appb-100031
    得到
    Figure PCTCN2015078172-appb-100032
    根据
    Figure PCTCN2015078172-appb-100033
    得到第三迭代周期中各接收节点的新的初始值
    Figure PCTCN2015078172-appb-100034
    根据第三迭代周期中各接收节点的新的初始值
    Figure PCTCN2015078172-appb-100035
    构成新的L维初始向量:
    Figure PCTCN2015078172-appb-100036
    将各接收节点的
    Figure PCTCN2015078172-appb-100037
    存储在NrGL维向量
    Figure PCTCN2015078172-appb-100038
    中,形成第三信号数据向量;
    根据第三迭代规则
    Figure PCTCN2015078172-appb-100039
    对第三信号数据向量进行迭代;其中,
    Figure PCTCN2015078172-appb-100040
    表示gossip更新矩阵,
    Figure PCTCN2015078172-appb-100041
    表示NrG(L+1)的单位矩阵,eG(L+1)i表示第iG(L+1)-G(L+1)-1个元素至第iG(L+1)个元素为1而其他元素都为0的一个G(L+1)维向量;
    每次迭代后,判断是否
    Figure PCTCN2015078172-appb-100042
    如果是,则记录并累加相等次数C,否则,将相等次数C归零;
    当相等次数C达到预设次数CT时,根据
    Figure PCTCN2015078172-appb-100043
    计算γ2(θ),并存储当前每个接收节点的信号
    Figure PCTCN2015078172-appb-100044
    并将其作为第四迭代周期中各接收节点的初始值;其中,
    Figure PCTCN2015078172-appb-100045
    t3为实现第三迭代周期信息共享所耗费的迭代次数;
    如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
    Figure PCTCN2015078172-appb-100046
    并进入本次迭代周期内的下一次gossip循环。
  6. 如权利要求5所述的多目标DOA估计方法,其特征在于,当φ=4时,所述DOA估计方法包括如下步骤:
    根据
    Figure PCTCN2015078172-appb-100047
    计算第四迭代周期中各接收节点的新的初始值
    Figure PCTCN2015078172-appb-100048
    将第四迭代周期中各接收节点的新的初始值
    Figure PCTCN2015078172-appb-100049
    存储在NrG维向量
    Figure PCTCN2015078172-appb-100050
    中,形成第四信号数据向量;
    根据第四迭代规则
    Figure PCTCN2015078172-appb-100051
    对所述第四信号数据向量进行迭代;其中,
    Figure PCTCN2015078172-appb-100052
    表示gossip更新矩阵,
    Figure PCTCN2015078172-appb-100053
    表示NrG的单位矩阵,eGi表示第iG-G-1个元素至第iG个元素为1而其他元素都为0的一个G维向量,t表示迭代次数;
    每次迭代后,判断是否
    Figure PCTCN2015078172-appb-100054
    如果是,则记录并累加相等次数C,否则,将相等次数C归零;
    当C>CT=ρNr时,根据
    Figure PCTCN2015078172-appb-100055
    得到γ3(θ),并存储当前每个接收节点的γ1(θ),γ2(θ),γ3(θ),并将存储的当前每个接收节点的γ1(θ),γ2(θ),γ3(θ)作为第五迭代周期中各接收节点的初始值;其中:
    Figure PCTCN2015078172-appb-100056
    由第二迭代周期结束时获得;
    Figure PCTCN2015078172-appb-100057
    由第三迭代周期结束时获得;
    Figure PCTCN2015078172-appb-100058
    由第四迭代周期结束时获得,t4为实现第四迭代周期信息共享所耗费的迭代次数;
    如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
    Figure PCTCN2015078172-appb-100059
    并进入本次迭代周期内的下一次gossip循环。
  7. 如权利要求6所述的多目标DOA估计方法,其特征在于,当φ=5时,所述DOA估计方法的包括如下步骤:
    根据
    Figure PCTCN2015078172-appb-100060
    计算
    Figure PCTCN2015078172-appb-100061
    根据
    Figure PCTCN2015078172-appb-100062
    计算
    Figure PCTCN2015078172-appb-100063
    根据
    Figure PCTCN2015078172-appb-100064
    计算γ6(θ);其中,L为采样点的个数,Mt为发射节点个数,xm为第m发射节点的发射信号,Rx为发射信号的自相关矩阵,
    Figure PCTCN2015078172-appb-100065
    λ表示发射信号的波长,
    Figure PCTCN2015078172-appb-100066
    表示第m发射节点与角度为θ的目标之间的近似距离,
    Figure PCTCN2015078172-appb-100067
    表示角度为θ的目标与第i接收节点的近似距离,[]*表示共轭操作,[]T表示转置操作;
    利用
    Figure PCTCN2015078172-appb-100068
    和γ6(θ),构成新的L维初始向量:
    Figure PCTCN2015078172-appb-100069
    Figure PCTCN2015078172-appb-100070
    其中:
    Figure PCTCN2015078172-appb-100071
    Figure PCTCN2015078172-appb-100072
    Figure PCTCN2015078172-appb-100073
    Figure PCTCN2015078172-appb-100074
    存储在NrGL维向量
    Figure PCTCN2015078172-appb-100075
    中,形成第五信号数据向量;
    根据第五迭代规则
    Figure PCTCN2015078172-appb-100076
    对所述第五信号数据向量进行迭代;其中,
    Figure PCTCN2015078172-appb-100077
    表示gossip更新矩阵,
    Figure PCTCN2015078172-appb-100078
    表示NrGL的单位矩阵,eGLi表示第iGL-GL-1个元素至第iGL个元素为1而其他元素都为0的一个GL维向量;
    每次迭代后,判断是否
    Figure PCTCN2015078172-appb-100079
    如果是,则记录并累加相等次数C,否则,将相等次数C归零;
    当C>CT=ρNr时,根据
    Figure PCTCN2015078172-appb-100080
    计算γ4(θ);并存储当前每个接收节点的信号
    Figure PCTCN2015078172-appb-100081
    并将其作为第六迭代周期中各接收节点的初始值,t5为实现第五迭代周期信息共享所耗费的迭代次数;
    如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
    Figure PCTCN2015078172-appb-100082
    并进入本次迭代周期内的下一次gossip循环。
  8. 如权利要求7所述的多目标DOA估计方法,其特征在于,当φ=6时, 所述DOA估计方法包括如下步骤:
    根据公式
    Figure PCTCN2015078172-appb-100083
    得到
    Figure PCTCN2015078172-appb-100084
    根据
    Figure PCTCN2015078172-appb-100085
    构成新的L维向量:
    Figure PCTCN2015078172-appb-100086
    其中:
    Figure PCTCN2015078172-appb-100087
    Figure PCTCN2015078172-appb-100088
    Figure PCTCN2015078172-appb-100089
    存储在NrG维的向量
    Figure PCTCN2015078172-appb-100090
    中,形成第六信号数据向量;
    根据第六迭代规则
    Figure PCTCN2015078172-appb-100091
    对所述第六信号数据向量进行迭代;其中,
    Figure PCTCN2015078172-appb-100092
    表示gossip更新矩阵,
    Figure PCTCN2015078172-appb-100093
    表示NrG的单位矩阵,eGi表示第iG-G-1个元素至第iG个元素为1而其他元素都为0的一个G维向量;
    每次迭代后,判断是否
    Figure PCTCN2015078172-appb-100094
    如果是,则记录并累加相等次数C,否则,将相等次数C归零;
    当C>CT=ρNr时,根据
    Figure PCTCN2015078172-appb-100095
    得到γ5(θ);并计算DOA估计值;计算公式为:
    Figure PCTCN2015078172-appb-100096
    其中:
    Figure PCTCN2015078172-appb-100097
    Figure PCTCN2015078172-appb-100098
    Figure PCTCN2015078172-appb-100099
    其中,
    Figure PCTCN2015078172-appb-100100
    如果相等次数未达到预设的次数CT,则记录当前迭代次数内的
    Figure PCTCN2015078172-appb-100101
    并进入本次迭代周期内的下一次gossip循环。
  9. 一种分布式网络中基于gossip算法的多目标DOA估计系统,其特征在于,包括:
    循环迭代模块,用于在第φ迭代周期中,根据第φ迭代规则对所述第φ信号数据向量进行gossip迭代,每次迭代后,判断所述第φ信号数据向量与迭代前是否相等,如果是,则记录并累加相等次数,否则将相等次数归零;其中,φ的初始值为1,所述第φ信号数据向量利用第φ迭代周期中各接收节点的初 始值构建得到;当所述相等次数达到预设次数,且φ的值未达到预设值时,完成第φ迭代周期的迭代,并储存此时的第φ信号数据向量,并将第φ信号数据向量中各接收节点的信号作为第φ+1迭代周期中各接收节点的初始值;循环执行上述各步骤,每次循环时,φ的值比上一循环中φ的值增加1;
    DOA估计值计算模块,用于当φ的值达到预设值时,根据各迭代周期的迭代结果,利用DOA估计值的计算公式计算DOA估计值;所述预设值为根据所述DOA估计值的计算公式计算DOA估计值所需要的迭代周期数。
  10. 如权利要求9所述的分布式网络中基于gossip算法的多目标DOA估计系统,其特征在于,所述预设值为6。
PCT/CN2015/078172 2014-06-20 2015-05-03 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法 WO2015192696A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201410281232.9 2014-06-20
CN201410281232.9A CN104023397B (zh) 2014-06-20 2014-06-20 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法

Publications (1)

Publication Number Publication Date
WO2015192696A1 true WO2015192696A1 (zh) 2015-12-23

Family

ID=51439911

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/078172 WO2015192696A1 (zh) 2014-06-20 2015-05-03 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法

Country Status (2)

Country Link
CN (1) CN104023397B (zh)
WO (1) WO2015192696A1 (zh)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109143154A (zh) * 2018-07-24 2019-01-04 南京航空航天大学 一种应用于l型阵列的信号二维doa与频率联合估计方法
CN109188345A (zh) * 2018-08-27 2019-01-11 电子科技大学 基于去预延迟空时结构的相干信号源doa估计方法
CN110061770A (zh) * 2019-03-27 2019-07-26 浙江理工大学 基于随机分组的无线传感器网络分布式波束形成方法
CN111181526A (zh) * 2020-01-03 2020-05-19 广东工业大学 一种用于信号处理的滤波方法
CN113569195A (zh) * 2021-07-19 2021-10-29 江南大学 一种用于分布式网络系统的预条件方法
US11196498B2 (en) * 2018-03-08 2021-12-07 Tsinghua University High-precision time synchronization method
CN115051924A (zh) * 2022-06-08 2022-09-13 上海佰贝网络工程技术有限公司 基于数据广播的Gossip算法中通信方式协调方法、装置、设备、系统及介质
CN115119142A (zh) * 2022-06-22 2022-09-27 杭州电子科技大学 一种基于传感器网络的分布式直接定位方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104023397B (zh) * 2014-06-20 2017-06-13 深圳大学 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102866383A (zh) * 2012-10-22 2013-01-09 重庆大学 一种基于空域自适应滤波的波达方向估计方法
CN103760519A (zh) * 2014-01-24 2014-04-30 深圳大学 高分辨率doa估计方法及系统
CN104023397A (zh) * 2014-06-20 2014-09-03 深圳大学 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006067857A1 (ja) * 2004-12-24 2006-06-29 Fujitsu Limited 到来方向推定装置及びプログラム

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102866383A (zh) * 2012-10-22 2013-01-09 重庆大学 一种基于空域自适应滤波的波达方向估计方法
CN103760519A (zh) * 2014-01-24 2014-04-30 深圳大学 高分辨率doa估计方法及系统
CN104023397A (zh) * 2014-06-20 2014-09-03 深圳大学 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
BOYD, STEPHEN ET AL.: "Randomized Gossip Algorithms", IEEE TRANSACTIOINS ON INFORMATION THEORY, vol. 52, no. 6, 30 June 2006 (2006-06-30), XP055244232 *
GUO, JUNYING ET AL.: "DOA estimating base on distributed sub space method", JOURNAL OF GUILIN UNIVERSITY OF ELECTRONIC TECHNOLOGY, vol. 33, no. 2, 30 April 2013 (2013-04-30) *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11196498B2 (en) * 2018-03-08 2021-12-07 Tsinghua University High-precision time synchronization method
CN109143154A (zh) * 2018-07-24 2019-01-04 南京航空航天大学 一种应用于l型阵列的信号二维doa与频率联合估计方法
CN109188345A (zh) * 2018-08-27 2019-01-11 电子科技大学 基于去预延迟空时结构的相干信号源doa估计方法
CN109188345B (zh) * 2018-08-27 2023-03-10 电子科技大学 基于去预延迟空时结构的相干信号源doa估计方法
CN110061770A (zh) * 2019-03-27 2019-07-26 浙江理工大学 基于随机分组的无线传感器网络分布式波束形成方法
CN111181526A (zh) * 2020-01-03 2020-05-19 广东工业大学 一种用于信号处理的滤波方法
CN111181526B (zh) * 2020-01-03 2023-03-17 广东工业大学 一种用于信号处理的滤波方法
CN113569195A (zh) * 2021-07-19 2021-10-29 江南大学 一种用于分布式网络系统的预条件方法
CN113569195B (zh) * 2021-07-19 2023-12-01 江南大学 一种用于分布式网络系统的预条件方法
CN115051924A (zh) * 2022-06-08 2022-09-13 上海佰贝网络工程技术有限公司 基于数据广播的Gossip算法中通信方式协调方法、装置、设备、系统及介质
CN115051924B (zh) * 2022-06-08 2023-11-21 上海佰贝网络工程技术有限公司 基于数据广播的Gossip算法中通信方式协调方法、装置、设备、系统及介质
CN115119142A (zh) * 2022-06-22 2022-09-27 杭州电子科技大学 一种基于传感器网络的分布式直接定位方法

Also Published As

Publication number Publication date
CN104023397A (zh) 2014-09-03
CN104023397B (zh) 2017-06-13

Similar Documents

Publication Publication Date Title
WO2015192696A1 (zh) 分布式网络中基于gossip算法的多目标DOA估计系统及估计方法
Einemo et al. Weighted least squares algorithm for target localization in distributed MIMO radar
Tirer et al. High resolution direct position determination of radio frequency sources
Gholami et al. TDOA based positioning in the presence of unknown clock skew
Yang et al. Phase synchronization for coherent MIMO radar: Algorithms and their analysis
CN108512787B (zh) 大规模mimo系统的超精细信道估计方法
CN107592671B (zh) 一种网络化多智能体主动变拓扑的自主协同定位方法
Noroozi et al. Efficient closed-form solution for moving target localization in MIMO radars with minimum number of antennas
CN110809240A (zh) 一种基于WiFi多维参数特征的室内目标被动跟踪方法
Lu et al. Novel energy-based localization technique for multiple sources
CN106656882B (zh) 一种信号合成方法及系统
CN109239649A (zh) 一种阵列误差条件下的互质阵列doa估计新方法
Zhong et al. A distributed particle filtering approach for multiple acoustic source tracking using an acoustic vector sensor network
Rapinski et al. ZigBee ranging using phase shift measurements
JP2021524590A (ja) 波動に基づいた信号を用いて少なくとも1つの物体の位置特定を行うための位置特定方法及び位置特定システム
CN113777575A (zh) 应用于复杂环境中mimo雷达多目标参数估计评估方法
Rajan et al. Relative kinematics of an anchorless network
Yin et al. Single-step localization using multiple moving arrays in the presence of observer location errors
Li et al. A precise ultra-wideband ranging method using pre-corrected strategy and particle swarm optimization algorithm
Wang et al. Joint localization and communication study for intelligent reflecting surface aided wireless communication system
CN113360841B (zh) 一种基于监督学习的分布式mimo雷达目标定位性能计算方法
Zhong et al. Constant modulus MIMO radar waveform design via iterative optimization network method
WO2015192695A1 (zh) 分布式网络中基于gossip算法的单目标DOA估计系统及估计方法
Kumar et al. An efficient compartmental model for real-time node tracking over cognitive wireless sensor networks
Oshiga et al. Superresolution multipoint ranging with optimized sampling via orthogonally designed golomb rulers

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15810393

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15810393

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205N DATED 28/02/2017)

122 Ep: pct application non-entry in european phase

Ref document number: 15810393

Country of ref document: EP

Kind code of ref document: A1