Models And Algorithms For Distributed Earth Observation Satellite System Online Coordination Scheduling Under Communication Constraints

The current demand for earth observation increases significantly,the earth observation task is more complex than before,the emergency cases and changes on observation environment need to be faced.In this situation,online coordination of Distributed Earth Observation Satellite System(DEOSS)is characterized with its onboard data processing,onboard planning & scheduling and onboard collaboration.It makes full use of the system's own distribution and autonomy to realize the rapid response,efficient scheduling and cooperative observation for emergency cases,in order to improve the observing performance of the whole system.This paper mainly studies the design of online collaboration architecture,the single-satellite online scheduling mechanism and the multi-satellite co-scheduling models & algorithms under the communication constraint to solve the online task co-scheduling problem of distributed earth observation satellite system under communication constraints in order to meet the needs of practical applications.The main research results are summarized as follows:(1)Analyze and design a centralized-distributed online collaboration architecture and a decentralized online collaboration architecture for DEOSS.On basis of the current multi-agent system collaboration architecture and general multi-satellite collaboration architecture,the main constraints,practical requirements and key architectural design points of multi-satellite online collaboration are described,then a novel centralized-distributed collaboration architecture and novel decentralized collaboration architecture are introduced.The intelligent level and corresponding configuration of autonomous functions of an Agent within the architecture,and the information flow among agents are developed in detail,so as to lay the foundation for proposing the algorithms corresponding to the different architectures.(2)Constructs a single satellite online task scheduling mechanism based on revision and progressive methods.Under the communication constraints,onboard computation constraints and dynamic arrival of urgent tasks,the problem of single satellite online scheduling problem within satellite network environment is divided into two decision sub-problems: one is when-to-schedule and the other one is how-to-schedule.For when-to-schedule,a hybrid scheduling point decision mechanism based on rolling horizon and urgent task accumulation threshold is developed.For how-to-schedule,Mixed Integer Linear Programming(MILP)model is proposed.According to the proposed scheduling point decision mechanism,both the progressive method of complete re-scheduling and the revision method of schedule repairing are adopted,and then two online scheduling heuristic algorithms are proposed.In the case of experimental results,for when-to-schedule,the developed hybrid mechanism is superior to the completely periodic scheduling mechanism on into-scheduling percentage of urgent tasks and total profit.For how-to-schedule,the proposed two algorithms are superior to the recently published methods.(3)Propose a variety of online co-scheduling algorithms referring to different online collaboration architectures under communication constraints.Homogeneous multi-satellite on-line co-scheduling problem under communication constraints is described,especially the definition of the communication time window,the time availability of batch tasks and the time availability of urgent tasks for the communication constraints,and then MILP model for each sub-problem with the arrival of urgent task batch is established.Two online coordination scheduling algorithms based on the market mechanism are proposed under centralized-distributed architecture: Single task-Contract Net Protocol algorithm(SI-CNP)and Batch task-Contract Net Protocol algorithm(BA-CNP).Another two algorithms are proposed under decentralized architecture: modified consistency-based bundle algorithm with synchronous communication(m-CBBA)and modified asynchronous consistency-based bundle algorithm with asynchronous communication(m-ACBBA).The experimental results show that m-CBBA is preferred for efficiently balancing the total profit and communications if the communication cost is high,and if the communication cost is low,m-ACBBA is the best choice for obtaining the high total profit and high successfully scheduled rate of urgent tasks.(4)Systematically study and solve the heterogeneous multi-satellite online co-scheduling problem for compound tasks under communication constraints.Firstly,the compound tasks and sub-task constraints are described.The compound tasks are divided into proactively generated tasks that can be scheduled once and the progressively generated tasks that need incrementally scheduling.Then,the sub-problem is divided and modeled as MILP.Finally,a heterogeneous multi-satellite online coordination mechanism based on the General Partial Global Programming(GPGP)is constructed in order to improve the proposed online co-scheduling algorithms under the decentralized architecture.The experimental results show that if the communication cost is high,it is more advantageous to carry the imaging instruments as much as possible.To improve the task completeness of the compound task,it is necessary to ensure a high profit coefficient.