战术级分布式计算框架  

The Research on Tactical Level Distributed Computing Framework

作　　者：杨峰 谢帅 张宗兴 闫雯罡 张光亚 YANG Feng;XIE Shuai;ZHANG Zongxing;YAN Wengang;ZHANG Guangya(The Technological Research Department of Unit 32124 of PLA,Yanji 133000,China)

机构地区：[1]解放军32124部队技术研究室,吉林延吉133000

出　　处：《火力与指挥控制》2023年第10期55-64,共10页Fire Control & Command Control

摘　　要：随着智能化装备的发展,战术场景中出现大量智能终端,使得战场数据采集和交换大大增加,如果将这些计算资源通过分布式计算框架整合起来,就能更好地发挥人工智能和大数据分析算法的作用,进一步增强终端的感知能力和自主决策能力。提出一种名为魔捷的轻量级分布式计算框架,包含执行器、本地调度器和全局调度器等模块,在调度效率、容灾能力和安全性能等方面做了针对性的设计,具有跨平台和多语言支持能力,程序实现上具有高内聚、低耦合和易扩展的特点,更加适用于无人化战术场景,能够灵活分配或汇聚战场末端的算力,从而提升智能化装备的战术性能。With the development of intelligent equipment,a large number of intelligent terminals have emerged in tactical scenes,greatly increasing the amount of battlefield data collection and exchange.If the computation resources are integrated by a distributed computing framework,then the role of benefit artificial intelligence and big data analysis algorithms will come into play better,the awareness and autonomous decision-making capability of the terminals will be further enhanced.A lightweight distributed computing framework called Moj is proposed,including such modules such as executor,local scheduler,and global scheduler,etc.And the targeted designs are made in terms of scheduling efficiency,disaster tolerance and security performance,etc.With the ability to support cross platforms and multiple languages,the program implementation has the characteristics of high cohesion,low coupling and easy expansion.It is more suitable for unmanned tactical scenes,and can flexibly allocate or converge the computing power on the end of the battlefield so as to improve the tactical performance of intelligent equipment.

关 键 词：分布式计算 人工智能 无人化作战 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]