Research On Distributed Fusion Algorithms Under Time-asynchronization And Different Fields-of-view

In the distributed sensor network,each node exchanges multi-target posterior distribution with neighboring nodes and combines the information of all neighboring nodes for fusion.However,due to the difference of functions,configuration clocks and detection ranges among nodes in the distributed sensor network,nodes are unable to detect the same target information at the same time in the same field-of-view(Fo V)so that distributed fusion algorithms based on generalized covariance intersection(GCI)and weighted arithmetic average(WAA)rules fail.Whether it can ensure that the target information represented by the information to be fused is consistent is the key to ensure the effectiveness of distributed fusion algorithms under time-asynchronization and different Fo Vs.In this dissertation,the problem analysis,method research and experimental verification are carried out on multiple sampling rates,clock offset,different Fo Vs and unknown Fo V information involved in distributed fusion algorithms under time-asynchronization and different Fo Vs.The main contents are as follows.1.Aiming at the distributed fusion problem under time-asynchronization,a distributed fusion algorithm based on time-stamp difference(TD)under GCI rule,named TD-GCI,is proposed.It can use the time-stamp information and C-K equation to achieve the time-alignment of the multi-target posterior distribution between nodes for the scenario with multiple sampling rates,which ensures the effectiveness of GCI fusion algorithm.Further,a distributed fusion algorithm based on the time-offset estimation(TE)under GCI rule,named TE-GCI,is proposed.It can use random sampling or multiple clock hypotheses strategies to estimate the time-offset for the scenario with clock offset and unknown transmission delay,which ensures the effectiveness of the time-alignment process before fusion process.2.Aiming at the distributed fusion problem under different Fo Vs,a distributed fusion algorithm based on intensity decomposition(ID)of probability hypothesis density(PHD)filter under GCI rule,named ID-GCI-PHD,is proposed.It can achieve the match and fusion between sub-intensities for the scenario where the Fo V information is unknown,which improves the tracking performance of the GCI-PHD fusion algorithm when the target mis-detection occurs.Moreover,another distributed fusion algorithm based on Fo V information under WAA rule,named FID-WAA-PHD,is proposed,which improves the accuracy of target cardinality estimation for the scenario where the Fo V information is different but known and time-invariant.Further,the ID-WAA-PHD algorithm based on clustering method implementation,named CID-WAA-PHD,is proposed,which avoids the wrong space decomposition caused by the unreliable Fo V information for the scenario with unreliable or unknown Fo V information.3.Aiming at the distributed fusion problem under time-asynchronization and different Fo Vs,an asynchronous ID-WAA-PHD distributed fusion algorithm based on clustering method,named ACID-WAA-PHD,is proposed.It can both realize time-alignment and adaptive space decomposition,which expands the application ranges of WAA-PHD fusion algorithm.