去偏转换量测IMM机动目标跟踪算法的硬件实现

DCM-IMM for Maneuvering Target Tracking Algorithm Realized with Hardware

交互多模型(IMM)与去偏转换量测卡尔曼滤波(DCMKF)相结合的去偏转换量测IMM(DCM-IMM)算法是一种常用的机动目标跟踪算法。但该算法运算复杂,计算量大,用软件方法实现难以满足实际工程应用中实时性高的要求。首先,本设计将DCM-IMM算法拆解为标量运算;然后,采用块的设计思想,利用现场可编程门阵列(FPGA)的并行流水线机制,对模块内部的运算单元进行分时复用和优化,实现了浮点DCM-IMM算法的硬件设计,解决了采用传统的软件方法实现过程中存在实时性难以满足的问题;最后,通过Matlab和QuartusⅡ仿真,验证了本设计的优越性。

Combining Interacted Multiple Model （IMM） with Debiased Converted Measurement Kalman Filter （DCMKF）, the Debiased Converted Measurement IMM （DCM-IMM） is a commonly used maneuvering target tracking algorithm. However, the computation of this algorithm is time-cost and complex. The software realization of the algorithm is not adaptable for the real-time requirement in actual engineering application. We solved the problem by using hardware. Firstly, the DCM-IMM algorithm was factorized to the scalar form. Then, the parallel pipeline mechanism of field programmable gate array （FPGA） was utilized to proceed time division multiplexing and optimization of computing units within modules based on block designing method. The hardware design scheme of floating point DCM-IMM algorithm can solve the real-time problems in traditional software design. Finally, the simulations of Matlab and QuartusⅡ show the advantages of this design.