Research On The Evolution Mechanism Of Device-collaborative Computing Networks Based On Propagation Dynamics

With the iterative update of mobile communication technology and artificial intelligence technology,the scale of terminal devices is expanding,data traffic in communication networks is surging,and a variety of computation-intensive services are emerging,putting forward higher requirements on the consumption of computing resources.Cloud computing and mobile edge computing technologies have been proposed to a certain extent to break the limitation of computing capacity of terminal devices.However,due to the high requirements of base stations and edge servers for deployment locations,the coverage of cloud computing and edge computing networks is limited.With the increasing maturity of device-to-device technology(D2D technology),the emergence of devicecollaborative computing networks has made the participation of user devices in collaborative computing offloading another effective way to relieve network data pressure and enhance user experience,compensating for the communication network coverage defects finitely.However,the potential risks associated with over-reliance on devicecollaborative computing should not be underestimated.When a large number of computation tasks flood into device-collaborative computing networks,terminal users experiencing computation task buildup will offload computation tasks to neighboring idle users for collaborative computing,if the existing users with free resources in the terminal network are not enough to support all the offloading tasks to complete collaborative computing,it will cause more terminal users in the surrounding area to accumulate computation tasks and thus make the devices participating in device-collaborative computing short of their own resources.They not only have no free resources to provide collaborative computing,but also cannot meet their own task demands,thus causing the device-collaborative network to lose its advantages of efficiently and conveniently assisting cloud computing and edge computing and enters a state of paralysis.Therefore,in this thesis,we apply propagation dynamics theory to study the evolution mechanism of device-collaborative computing networks,analyze the propagation process of computation task offloading based on D2D technology,and investigate the critical conditions to ensure the stability and effectiveness of them.The following work is specifically carried out.First,a propagation dynamics model underlying the devicecollaborative computing network is constructed based on the linear threshold model for the task offloading transmission process in the devicecollaborative computing offloading context.The proposed dynamics model is analyzed and discussed by applying the edge-based compartmental theory,and the propagation dynamics equations are established.The saddle point of the dynamics equations is derived by mathematical solution,which corresponds to the emergence of the network paralysis state,and thus the critical condition that will lead to task overload of more users in the device-collaborative computing network is obtained.The rationality and accuracy of the proposed dynamics model were verified by simulation on Paul Erdos-Alfred Renyi Random Network(ER random network)and Scale-free Network(SF network)respectively.The effects of idle user service capacity threshold,overloaded user collaborative offloading probability,and network degree distribution heterogeneity on propagation process of device-collaborative offloading are investigated.Secondly,the heterogeneity of user behaviors in device-collaborative computing network is analyzed,considering that with the gradual improvement of terminal device intelligence,when user nodes have some decision-making skills,they can actively choose different computing offloading methods,including:partial offloading and full offloading.The impact of different offloading methods on the propagation process of task offloading is discussed from the perspective of resource and energy consumption.A more realistic propagation dynamics model of the devicecollaborative network is designed and optimized.Subsequent mathematical derivation is carried out to obtain the critical conditions for the emergence of network paralysis.The simulation platform is built to simulate the accumulation of resource and energy consumption under partial and full offloading respectively,and the simulation test results match with the theoretical prediction results.On this basis,the impact of the proportion of offloading tasks and the proportion of overloaded users with different offloading methods on the evolution of device-collaborative networks with user behavior heterogeneity is further explored.In this thesis,an analytical model based on propagation dynamics theory is proposed.The model is to explore the evolution mechanism of device-collaborative computing networks in response to the problems of over-reliance on device-collaborative computing.The problems mainly refer to the accumulation of computation tasks for a large number of mobile terminal users,the shortage of computing resources of idle devices,and the paralysis of device-collaborative computing networks that rely on the assistance of D2D technology.The established propagation dynamics model and simulation results have good reference significance for the deployment of device-collaborative computing networks.