模拟训练中可靠组播技术研究与实现

Design of Reliable Multicast Model For Naval Training Simulators

该文针对舰艇模拟训练系统的需要,提出一种可扩展的可靠应用层组播模型,它采用心跳机制与序列号相结合的方法进行差错检测和基于NAK否定信息的ARQ机制进行数据差错恢复,同时设计了相应的机制来保障组播数据的可靠传输。该模型弥补了普通应用层组播协议在可靠性传输上的不足,使通信网络具有良好的可扩缩性、可选择和可靠性功能。实验证明:基于心跳机制的差错控制模型可以很好克服传统方法可靠组播通信的"反馈风暴"和NAK算法中的缓冲区无法控制的问题,与基于TCP的单播可靠方式相比,该模型不需要中心节点服务器,节省了网络资源,网络带宽占有率相比也减少约两倍。

For the reliable and real-time requirements of naval training simulators, this paper proposes a new scalable and reliable Application Layer Multicast （ALM） model. Serial integrated with heat beating are used to discover a error, ARQ mechanism based on NAK feedback is modified to correct a error, and the feasible structure are putted forward for the multicast to realize the reliability of data transmission. It is imported to make up the drawbacks of unreliable delivery ALM protocols, and improves the network efficiency and the system scalability. The result of experiments on naval training simulators shows that the mechanism based on heat beating is a perfect method for solving the traditional problem of feedback explosion and the increasing buffer size of the NAK model. Compared with TCP, This scheme reduces about 200% on the network bandwidth, and is more economical without the sever computer.