Research On Earth Observation Task Cooperation For Distributed Satellites System

Earth observation is a very important approach to gaining earth observation information, improving earth system science, space information science and so on. With the development of modern small satellites' techniques especially that of satellites' autonomous operation, there is an increasing desire in many organizations to use distributed satellites system (DSS), including constellations and fleets, to accomplish more and more complex earth observation mission objectives by autonomous spacecraft working together collaboratively in DSS. The vision of what can be achieved from space is no longer bound by what an individual satellite can accomplish. Rather, the functionality can be spread over a number of cooperating satellites. The reasons for this paradigm switch are many, including the increased production rates to decrease unit cost, and better performance in terms of mission science, fault tolerance, reconfigurability and upgradeability. With these far reaching benefits, however, comes a new set of challenges, including relative navigation, configuration maintenance, distributed control, multi-satellite cooperation and so on. The key technology that will enable multiple, distributed satellites to achieve their potential is coordinated intelligent autonomy. Whatever, DSS open up the opportunity to achieve new capabilities while using relatively inexpensive, perhaps disposable components. Based on it, the objective of this paper is to make a general research on relevant theory and methods of earth observation tasks cooperation for DSS in the dynamic circumstances by Multi-Agent theory.In detail, Earth observation task cooperation for DSS in this paper is defined as: considering the characteristics of autonomous operation cooperatively for DSS and the requirement of dynamic observing environment sufficiently, various observing tasks are allocated to the satellites in DSS optimally with constraint satisfaction of these tasks. The ultimate object of task cooperation is to make use of negotiation mechanism to accomplish general observation task effectively on the basis of the lowest consumption of energy, so as to improve the rate of accomplishments.Currently, the research on earth observation task cooperation for DSS is a brand new topic, no matter domestically or abroad. So, it makes our research important and significant in both of theory and practice. By analyzing the characteristics of autonomous cooperation model for earth observation DSS, we pointed out the autonomous control architecture model for DSS, and task cooperation mechanism based on it. Cooperation mechanism includes negotiation and task optimal allocation. The main contents and conclusion of this paper are outlined as follows:Firstly, according to the basic characteristics and generally functional requirement of earth observation DSS, we hold that it is necessary and feasible to adopt Multiple Agent System (MAS) theory to instruct the research on autonomous control of DSS. The qualitative analysis and comparison of commonly-used autonomous control architectures of DSS is discussed. Based on it, we depict the system through organization structure and control structure. So, a hybrid systematic hierarchical structure model with two levels is put forward to fulfill autonomous coordination demands of DSS under the dynamic circumstances. Moreover, the study on satellite autonomy at home and abroad is still in the early stage, and there are not any guidelines or frameworks to follow till now. A novel agent model on satellite control architecture is outlined, and the components, functions and relationships of agents in the architecture are analyzed. It is designed for multi-satellites coordination and satellite self-government, thus it has generality to some extent.Secondly, there are some drawbacks in traditional contract net protocol for DSS, so we extend the protocol by MAS theory. In this paper, the conversation policies for extended contract net protocol and a formation model for BDI mental state, such as trust degree, cooperation frequency and so on, are proposed in accordance with selfless characteristics of satellites agent in DSS. Combining the extended protocol and these models, a negotiation mechanism in DSS is presented. A series of detailed rules of action based on BDI agent are set forth in the mechanism to ensure a better cooperation between agents and higher flexibility of system.Thirdly, according to the features and some basic hypotheses about earth observation task cooperation for DSS, the mathematical model on task optimal allocation is put forward. Based on the model and BDI-extended contract net protocol, we study task optimal allocation for DSS in all the processes of announcing, bidding and awarding, and figure out corresponding algorithm. Based on it, considering DSS is under the dynamic circumstances, the allocation for some typical dynamic tasks is made further study.In announcing process, considering frequent communication burdening network load, we raise a method for screening bidders, which involve the historic result about completing similar task, and set forth corresponding multi-level First-N random selection algorithm.In bidding process, citing the conception of backward time slack, we judge whether bidder enable accomplish the observation task by inspecting task constraints satisfaction both of time window constraint and energy constraint. Meanwhile, we list corresponding testing algorithms of time window constraint and energy constraint, and analyze their time complexity.In awarding process, task allocation is switched to set-covering. So, with the principle of lowest general energy-consumption for completing observation task, we figure out a kind of greedy heuristic algorithm, testify its convergence, and give its searching supremum and time complexity. Moreover, we compare allocation effect of the algorithm with simple greedy algorithm through a case.Finally, we design a prototype of multi-agent based earth observation task cooperation of distributed satellite system, and offer a case to explain how the depicted model, algorithm and prototype works and to identify reasonability and efficiency of our task cooperation theory.