Research On Crowd Evacuation Path Planning Method Based On Federated Computing And Edge Computing

In recent years,with the acceleration of urbanization,the flow of people in public places is getting larger and larger.And the potential hazards of large-scale crowd gathering in public places have also increased,with people trampling and other public emergencies occurring frequently.The video data captured by the numerous cameras deployed in the premises can assist in the evacuation of people in emergency situations.Nowadays,with the development of Io T devices and technologies,numerous cameras are deployed in large numbers in private areas.The video data collected by each camera is highly private and uploading the collected video data directly to the cloud data processing centre may cause the problem of user privacy leakage,so the data from each camera cannot be shared effectively.Due to the"data silos"problem caused by the privacy preserving of each camera,it is difficult to obtain the video data of the whole scene for path planning,which greatly reduces the effectiveness of path planning.It is a challenging problem to solve the problem of"data silos"in crowd evacuation route planning while ensuring privacy,so as to provide a global route planning service for crowd evacuation.At the same time,crowd evacuation path planning has high requirements for real-time.However,with the increase in the number of cameras in public places,it will lead to a huge amount of video data at the edge of the network,in addition to uploading the huge amount of video data from the edge nodes to the cloud data processing centre will take up a lot of network bandwidth resources,which will also largely increase the network latency.It is therefore an equally challenging problem to provide global route planning services for crowd evacuation with limited resources.To solve the above problems,this thesis proposes a research on crowd evacuation path planning method based on federated computing and edge computing.Firstly,this thesis proposes a crowd evacuation path planning method based on federated computing.The method solves the problem of data privacy preserving during crowd evacuation by constructing a privacy preserving potential field and using the architecture of federated computing.Then,based on the above methods,in order to save network resource consumption and reduce network time delay,and to meet the requirement of real-time path planning,this thesis proposes an optimal crowd evacuation path planning method based on edge computing.The method uses the AO-ABC algorithm(Aggregation Optimization-Artificial Bee Colony Algorithm)to solve for the optimal number of aggregations.Finally,based on the above approach,a crowd evacuation simulation system based on federated and edge computing is constructed in this thesis to demonstrate the theoretical analysis results of this thesis in a more intuitive way.The main work and innovations of this thesis are as follows.(1)Existing crowd evacuation path planning methods do not take into account the privacy preserving issues of each camera in large area evacuation scenarios.This thesis proposes a federated computing-based crowd evacuation path planning method,which firstly proposes a federated computing-based crowd evacuation framework and introduces the federated computing overshoot for crowd evacuation.Secondly,a privacy-preserving potential field(P~3field)is constructed to provide path planning services to the crowd while preserving privacy.The privacy-preserving potential field consists of three components:1)scene discomfort field;2)distance field;3)crowd discomfort field.Finally,we propose a privacy-preserving potential field-based crowd evacuation path planning method,where the direction of individual movement velocity is determined by the decreasing gradient of the privacy-preserving potential field value,and a density field is constructed to quantify the magnitude of individual movement velocity to obtain the individual movement velocity.The experimental results show that the federated computing-based crowd evacuation path planning method can provide a global path planning service for crowd evacuation while protecting the privacy of each camera user.(2)The existing crowd evacuation path planning methods do not take into account the optimization of network resources and the real-time nature of path planning.This thesis proposes an optimal crowd evacuation path planning method based on edge computing.The approach first proposes an intelligent collaborative edge computing architecture for crowd evacuation at the cloud edge,and constructs an edge computing-based collaborative model for crowd evacuation at the cloud edge,which enables multiple cameras deployed in large area scenarios to work together.Then considering the resource optimization problem,this thesis proposes a global aggregation optimization model which defines the aggregation optimization problem as a problem of maximum the yield of the model.(3)A crowd evacuation simulation system based on federated computing and edge computing is constructed.The simulation results show that the method in this thesis can provide privacy-preserving global path planning services for crowd evacuation,and simulate the crowd evacuation process more intuitively through realistic rendering,which is instructive for crowd evacuation.