CAJ | 학술논문

完全分布式的机动目标跟踪是传感器网络等应用中亟待解决的关键问题。本文针对变拓扑非完全连通网络,提出一种基于网络共识的多模型信息滤波器( Consensus based Multiple Model Information Filter, C-MMIF)。 C-MMIF基于标准IMM框架,保证了估计最优性；并通过构造目标运动模式概率和状态估计的信息滤波形式,使节点间运算相互独立。同时,每个独立节点仅需与其相邻节点通讯,利用平均网络共识分布式优化算法对自身信息状态进行更新,实现节点间对目标运动模式及状态的一致估计。最后在无人机与地面传感器网络协同对地机动目标跟踪场景下进行算法仿真验证,结果证明该方法可以在无融合处理中心且网络拓扑变化情况下,使各节点实现对机动目标的一致有效跟踪。
완전분포식적궤동목표근종시전감기망락등응용중극대해결적관건문제。본문침대변탁복비완전련통망락,제출일충기우망락공식적다모형신식려파기( Consensus based Multiple Model Information Filter, C-MMIF)。 C-MMIF기우표준IMM광가,보증료고계최우성；병통과구조목표운동모식개솔화상태고계적신식려파형식,사절점간운산상호독립。동시,매개독립절점부수여기상린절점통신,이용평균망락공식분포식우화산법대자신신식상태진행경신,실현절점간대목표운동모식급상태적일치고계。최후재무인궤여지면전감기망락협동대지궤동목표근종장경하진행산법방진험증,결과증명해방법가이재무융합처리중심차망락탁복변화정황하,사각절점실현대궤동목표적일치유효근종。
To track a maneuvering target in a fully distributed way is an emerging and critical task in important sen-sor network applications. A consensus based multiple model information filter ( C-MMIF) is presented for not-fully-connected sensor networks with variable topology. The C-MMIF follows standard interacting multiple model ( IMM) filter framework which ensures the optimality of the estimation. Through applying information filtering for-mation on target’ s motion model probability and state estimation, parallel calculation among sensors are achieved. Meanwhile, with the average consensus distributed optimization algorithm, each sensor only needs to communicate with its adjacent nodes and update its own information states to achieve the global consensus on estimation of motion model probability and system states. Simulation test of cooperative target tracking using an unmanned air vehicle ( UAV) and ground sensor network are conducted. The results validate that, in a sensor networks with variable to-pologies and without fusion processing center, the C -MMIF can make every node effectively track maneuvering target.