| The new generation of LEO satellite constellation broadband network is an important part of the space-ground integration network.LEO satellite constellation network,with its advantages of strong survivability,wide coverage,flexible deployment and low constraints,can provide high-speed and high-quality services.By deploying computing resoureces on the satellites and making request served by them,the pressure of the cloud node can be reduced and the task processing efficiency also can be improved.This research has important theoretical and practical significance.In order to reduce the latency of request response and the energy consumption in the LEO satellite constellation broadband network,this thesis proposes multi-level offloading algorithms.There are three main research contents in this paper.Firstly,some element models should be established to support the design of offloading algorithms and the construction of the simulation platform.It includes network model,computing model,request model and so on.Secondly,two application scenarios for LEO constellation network were proposed.They are sparse user scenario and dense user scenario.After that,according to the number of the requests in different scenarios,the offloading range of satellite has been divided into three main categories,which are single satellite,multi satellite in the same orbit and multi satellite in different orbits.This thesis designed different strategies for each offloading range.For single satellite offloading,the single satellite multilevel computing delay matching offloading algorithm（ss MCD）is proposed by this thesis.And an algorithm called multi satellite priority offloading algorithm（ms PS）is designed for the same orbit multi satellite offloading.In addition,for the multi-level network of extraterrestrial multi satellites,based on the utilization of satellite resources and task delay,this thesis proposes the algorithm of extraterrestrial multi satellite delay resource comprehensive evaluation（DRCE）by using projection pursuit comprehensive evaluation method and simulated annealing method.Finally,based on average request response delay（ARRD）and average request energy consumption（AREC）,this thesis uses STK and OPNET to design network simulation and verify the performance of the algorithms.For the the two application scenarios,the ARRD and AREC were tested under the two application scenarios without satellites which have computing capability.The results show that enabling the on-board resources of the access satellite can greatly reduce the ARRD and AREC in the sparse user scenario.In the dense user scenario,although ARRD can be reduced,AREC will increase.Then,for the ss MCD algorithm,it shows that the delay and energy consumption of the comparable basic random algorithm are reduced by25% and 58%.At the same time,when the access satellite is overloaded,the application of ms PS offloading algorithm in this thesis can further reduce ARRD by 9% on the basis of single satellite algorithm.DREC algorithm is suitable to be used in middle and low latitudes,and can further reduce ARRD and AREC by 6% and 17% compared with ms PS algorithm. |
