BBPSO-PF algorithm based on GPU parallel optimization

doi:10.13245/j.hust.210303

	BBPSO-PF algorithm based on GPU parallel optimization
	Cao J(曹洁)1,2,3 ; Hu WD(胡文东)1; Wang JH(王进花)2 ; Yu P(余萍)2
	2021-03-23
发表期刊	Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition)
ISSN	1671-4512
卷号	49 期号:3 页码:12-17
摘要	To address the problem that the particle filter algorithm cannot be fully parallelized in the resampling process due to particle interaction, a parallel bare bones particle swarm optimization (BBPSO) algorithm was introduced to optimize the particle filter algorithm based on graphic processing unit (GPU). Firstly, in the aspect of the algorithm structure, the process of resampling was optimized benefiting from the easy parallelism of BBPSO algorithm. At the same time, the multi-threaded architecture of GPU was used to process the data of each particle swarm in paralle, which means each thread is responsible for one particle swarm, to avoid the shortcoming of the low-parallelism of particle filter resampling owing to particle interaction. Finally, an efficient particle cluster data storage structure was designed based on the principle of memory access in GPU alignment and merging, so as to accelerate the access speed of particle cluster data by reducing memory access transactions and further improve the real-time performance of the algorithm. The experimental results show that this method can improve the real-time performance of the algorithm while ensuring the accuracy of the algorithm. © 2021, Editorial Board of Journal of Huazhong University of Science and Technology. All right reserved.
关键词	Digital storage Graphics processing unit Memory architecture Monte Carlo methods Particle interactions Particle swarm optimization (PSO) Bare-bones particle swarm optimizations Graphic processing unit(GPU) Multithreaded architecture Parallel optimization Particle clusters Particle filter Particle filter algorithms Real time performance
DOI	10.13245/j.hust.210303
收录类别	EI
语种	中文
出版者	Huazhong University of Science and Technology
EI入藏号	20211710250313
EI主题词	Clustering algorithms
EI分类号	722 Computer Systems and Equipment ; 722.1 Data Storage, Equipment and Techniques ; 723 Computer Software, Data Handling and Applications ; 903.1 Information Sources and Analysis ; 922.2 Mathematical Statistics ; 931.3 Atomic and Molecular Physics
引用统计	无
文献类型	期刊论文
条目标识符	https://ir.lut.edu.cn/handle/2XXMBERH/148415
专题	电气工程与信息工程学院
作者单位	1.兰州理工大学计算机与通信学院; 2.兰州理工大学电气工程与信息工程学院; 3.兰州理工大学甘肃省城市轨道交通智能运营工程研究中心
第一作者单位	计算机与通信学院; 电气工程与信息工程学院; 兰州理工大学
第一作者的第一单位	计算机与通信学院
推荐引用方式 GB/T 7714	Cao J,Hu WD,Wang JH,et al. BBPSO-PF algorithm based on GPU parallel optimization[J]. Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition),2021,49(3):12-17.
APA	曹洁,胡文东,王进花,&余萍.(2021).BBPSO-PF algorithm based on GPU parallel optimization.Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition),49(3),12-17.
MLA	曹洁,et al."BBPSO-PF algorithm based on GPU parallel optimization".Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition) 49.3(2021):12-17.