Research On Target Tracking Algorithm Of IoT Sensor Fusion Based On Multi-agent

As an important means of connecting hardware and terminal networks,the Internet of Things has entered our lives gradually.In the Io T,as the carrier of perceiving target information in the environment,the single type of sensor cannot reflect the situation of the environment.As a result,the joint inference of environmental changes by multiple types of sensors has become a current trend and has become a hot spot for us to pay more and more attention.Target tracking technology is an important technology of the Io T to assist people in target tracking management.First of all,in the context of the data fusion of Io T sensors,a data fusion structure based on the Io T is proposed for the three-layer physical structure of the Internet of Things,the function and structure model of data fusion.Secondly,the nature of Agent and Multi-agent System(MAS)are introduced.It is proposed to build an Io T sensor data fusion model based on Agent.For the target tracking problem,an Agent-based target tracking fusion model at the location level is further proposed.And for the probability hypothesis density algorithm of random finite set,an Agent-based target tracking fusion model is proposed.In the Agent-based target tracking model,single target and multiple targets are tracked separately.Then,for the single target tracking problem in clutter environment,the method of data association and state fusion are used to research single target tracking.Under the Agent-based centralized fusion model,the nearest neighbor and probabilistic data association algorithms are compared and Kalman is used.to filtering state fusion,the data association of probability data association obtained through experimental comparison has a better tracking effect on single target.Finally,for the multi-target tracking problem,the classic Bayes algorithm and particle filter are first introduced.In order to avoid the data under target tracking computational dimension explosion phenomenon caused by correlation,the probability hypothesis density algorithm of the random finite set is combined with the particle filter algorithm,the RBPF-PHD algorithm is proposed based on the Rao-Blackwellized idea.Under the distributed tracking fusion structure based on Agent,the algorithm is proved by simulation It can effectively solve the problem of missing numbers.