MAV Swarm Parallel Simulation Based On KD-PARSE

It is safe, efficient and low-cost to perform a military mission with MAV (Micro Air Vehicle), such as searching, reconnaissance, electronic jamming etc. They will have an important role to play in future wars and can not be substituted. Due to the small size, poor abilities of load and defense of a single MAV, it is necessary and possible to build up MAV swarms. It will play a more powerful force to carry out a mission with MAV swarms. Now China has developed a variety of MAV models. In order to arm our army early with the new weapon, it is necessary to study the behavior of a MAV swarm using simulation technology. The dissertation realizes a MAV swarm model and implements it on KD-PARSE– a parallel simulation engine. And the speedup of parallel simulation and scalability of simulation model have been improved with original methods.The research of parallel simulation technology is still at it's initial stage in China. Firstly, the dissertation discusses some basic concept of Parallel Simulation, and has an in-depth study of the parallel simulation engine– the core component module of KD-PARSE. This part of the dissertation also studies the modeling framework of KD-PARSE, and establishes a good foundation of knowledge for following works.The next part is modeling and programming of something relevant to a MAV swarm parallel simulation. First of all, MAV and MAV swarm have been established with the method of modeling based on agent. Before starting a search mission with MAV swarm, the problem of routing must be resolved– to determine the shortest search path. The dissertation parallelizes the basic PSO algorithm, and realizes a program resolving MAV swarm routing problem with it. Then a parallel simulation program of MAV swarm flying along an established path has been accomplished. At last a simple GUI program has been accomplished, which is used to give a direct look into the trail each MAV in simulation.Then some parallel simulation experiments have been designed and the corresponding analyses have also been carried out. First of all, the effect to parallel simulation speedup of different Simulation Object disassembling method, different time manage method, and different number of CPUs has been studied through comparative method. Then the speedup is improved through the process of using distributed data management mechanism. And the modeling scalability is increased by using grouping sub-MAV-swarm and correspondingly partitioning the flight space. The overall study is reviewed and the future work is presented at the end of the dissertation.