Hu Xiaoxuan

The Framework for Multi Agent Concurrent Negotiation

In the multi-agent systems, the resources for the task and the ability of each single agent may be limited.An agent cannot do the whole task by itself, it need the help of other agents. So, the task needs to be decomposed into several subtasks. Some of the subtasks will be carried out locally by the original agent,while others need to be done non-locally with other agents. So, during the negotiation process, agents must make sure the order of each subtask and the assignment of the features for all the subtasks. In this paper, we are mainly concerning about the high level of the negotiation, and discuss about how to make the negotiation more efficient. We suggest doing negotiations concurrently, and propose a concurrent negotiation framework based on the Semi Markov Decision Process by using the concurrent action model.The framework helps agents to decide when to make decision in the negotiation. Experimental results show the performance of the model, and compare the outcome with difference termination scheme.

A two-phase genetic annealing method for integrated Earth observation satellite scheduling problems

This paper investigates an integrated approach to Earth observation satellite scheduling (EOSS) and proposes a two-phase genetic annealing (TPGA) method to solve the scheduling problem. Standard EOSS requires the development of feasible imaging schedules for Earth observation satellites. However, integrated EOSS is more complicated, mainly because both imaging and data transmission operations are of equal concern. In this paper, we first establish a mixed integer linear programming model for the scheduling problem using a directed acyclic graph for determining candidate solution options. Then, we optimize the model by applying the TPGA method, which consists of two phases in which a genetic algorithm is first employed, followed by simulated annealing. Detailed designs of the algorithm integration and algorithm switching rules are provided based on reasonable deductions. Finally, simulation experiments are conducted to demonstrate the feasibility and optimality of the proposed TPGA method.