China Mil Watch Mandarin-source monitoring · Chinese military & security reporting
Independent monitor
Official PRC military media, read in the original
← Daily Brief
Modernization Doctrine Cyber & Info

Learning Science and Technology with Academicians | One Network Across the Battlefield, All Things Can Be "Intelligently Connected"

跟着院士学科技|战场一张网 万物可“智联”
PLA Daily (解放军报) 5 August 2026
View original source ↗
Yin Hao, a Chinese Academy of Sciences academician and researcher at the Academy of Military Sciences' Systems Engineering Research Institute, lays out in PLA Daily the architecture and doctrine of what the PLA terms an intelligent communications network (智能化通信网络) — a hierarchically distributed system designed to link sensors, command nodes, fire units, and logistics across land, sea, air, space, electromagnetic, and cyber domains through AI-driven 'network cognition' (网络认知) and a standardized 'interaction language' (交互语言). The article documents how AMS is framing the network not as passive infrastructure but as an active cognitive layer that converts information superiority into decision and action superiority — a doctrinal formulation that extends the PLA's published thinking on intelligentized warfare (智能化战争) into the communications domain specifically. The explicit acknowledgment of vulnerabilities — adversarial sample attacks, data poisoning, cognitive-domain deception, and computing-resource bottlenecks — is the more analytically useful content, as it records where AMS assesses the concept remains unsolved rather than battle-ready.

One Network Across the Battlefield, All Things Can Be "Intelligently Connected"

■ Li Yue, Liu Qi, PLA Daily Reporter Jia Qilong

Conceptual diagram of an intelligent communications network covering the full-domain battlefield. AI-generated image.

The military communications network is the "nervous system" connecting battlefield weapons platforms and the lifeblood of seizing information superiority. Its capability level directly measures a nation's military strength. At present, artificial intelligence technology and communications network technology are deeply converging, integrating cutting-edge technologies such as advanced electronics, metamaterials, and intelligent manufacturing, and accelerating the evolution of communications networks toward intelligentization. The intelligent communications networks born from this process are no longer mere information "pipelines." In the era of intelligentized warfare (智能化战争), they are transforming into the critical cornerstone for realizing the "intelligent connection of all things" (万物智联), as if implanting a highly intelligent, full-domain-covering "neural network" into the body of the battlefield.

In this installment of "Learning Science and Technology with Academicians," we invite Yin Hao — researcher at the Systems Engineering Research Institute of the Academy of Military Sciences and Academician of the Chinese Academy of Sciences — to explain what an intelligent communications network is.

Weaving an Intelligent Communications Network Covering the Full Domain of Operations

The future battlefield, in its breadth and complexity, far exceeds traditional understanding. Multi-domain forces spanning land, sea, air, space, the electromagnetic spectrum, and cyberspace are highly integrated, and the battlefield situation changes in an instant. The core mission of the intelligent communications network is precisely to efficiently and resiliently connect the various reconnaissance and detection equipment, command-and-control equipment, weapons and materiel, manned/unmanned combat platforms, and individual soldier systems distributed across these domains, weaving an integrated battlefield network that covers physical space, cyberspace, and even cognitive space.

This network is composed of intelligent "terminals" (such as individual soldier terminals, communications vehicles, and unmanned aerial vehicles), "nerves" (routing and switching equipment), "centers" (network management and control equipment), and integrated "nodes" (such as command posts and weapons platforms). When combat units deploy, terminals automatically register and join the network; the management and control center simultaneously directs neural nodes to establish links between terminals. Once a user requirement is submitted, the system immediately calculates the optimal path, opens a "high-speed pipeline," and delivers data precisely to its destination, achieving instant delivery of battlefield information services.

Specifically, this intelligent network supports four key networks:

First, it constitutes an invisible "ubiquitous sensing network" (泛在感知网). Detection equipment across all domains uses it to monitor signal strength and spectrum occupancy, track latency and throughput, identify task priorities and bandwidth requirements, and analyze user trajectories, communication patterns, and resource consumption patterns, forming a complete, clear, and near-real-time awareness of the battlefield situation. The network can in real time "see" user locations and "understand" user needs, thereby providing precise services.

Second, it builds an agile "command-and-control network" (指挥控制网). Key nodes including command posts at all levels, command-and-control systems, and operational mission planning platforms are tightly connected through the intelligent communications network, forming an efficient "command-and-control hub." Relying on the network's damage-resistant resilience, when some nodes are damaged, the network can sense this in real time, rapidly activate contingency plans, and direct routing nodes to autonomously negotiate and erect backup information "bridges" at break points, ensuring the continuity of command links.

Additionally, it constructs a modular "fire strike network" (火力打击网). According to mission requirements, management and control equipment drives routing and switching equipment to rapidly establish high-priority temporary "data channels" between fire units of different organizational structures, achieving automatic task decoupling, efficient information transmission, and on-demand aggregation of fire platforms, dynamically combining dispersed fire units into modular strike clusters that can be assembled and employed on demand, making the use of firepower more flexible.

Finally, it supports a proactive "logistics and equipment support network" (后装保障网). Support forces spanning transportation, fuel, ammunition, and medical services throughout operations can achieve full awareness of support status through the intelligent communications network, supporting the integrated processing and rapid distribution of resource scheduling information, and providing continuous, precise materiel support information services for high-intensity intelligentized operations.

Building the "Smart Brain" for Cognitive Battlefield Awareness

How does this vast network achieve "intelligence"? The key lies in endowing the network with cognitive capability, forming a network knowledge system, and building a "smart brain" (智慧大脑).

Just as humans understand the world and accumulate experience through thought, we anthropomorphize the military communications network as a cognitive subject, implanting the cognitive model of "multi-domain perception — learning and analysis — optimized decision-making" into terminals, routers, network management equipment, and other devices. These devices thereby transform into intelligent nodes with the capacity to "think." Information passes through these nodes, converging level by level from terminals to regions to centers, ultimately constructing a hierarchically distributed "network brain" that can perceive, learn, analyze, and predict multi-domain information including the battlefield environment, its own state, and user behavior, forming a deep understanding of the network and its environment — this is "network cognition" (网络认知).

If network cognition is the process of thinking, then network knowledge is the product of thinking. In the field of intelligent communications, network knowledge specifically refers to the abstract representation of network elements (nodes, links, protocols) and their interaction mechanisms. The key lies in "unified representation" (统一表征) — standardizing the description of information from physical-layer signals to application-layer requirements, in order to support understanding and application among intelligent agents.

Network knowledge encompasses three types: "data-informational knowledge" (数据信息型知识) consists of statements about objective entities in the network, capable of recording in real time objective facts such as network speed, signal strength, and interference, providing the network with a portrait of its real environment so it can "clearly see" its own state; "relational-computational knowledge" (关系计算型知识) uses models to reveal the mapping relationships and intrinsic laws among network parameters, enabling the network to "perceive" the optimal solution amid complex changes; "logical-decisional knowledge" (逻辑决策型知识) contains the experience and contingency plans for handling problems, and is the "nourishment" for the network's autonomous optimization and evolution.

Just as an experienced engineer can identify a fault by the sound it makes, an intelligent communications network with rich network knowledge also possesses the capacity to "think." It can anticipate risks, plan paths, and optimize resource allocation, achieving the leap from "passive response" to "proactive foresight," providing targeted, differentiated, and proactive services to battlefield intelligent agents, and converting information superiority into decision-making and action superiority.

In order for intelligent agents of diverse types and varying architectures to collaborate like "comrades-in-arms," the intelligent communications network has established a standardized "interaction language" (交互语言). This language transcends the interoperability of underlying data formats and pursues the sharing of high-level semantics. Through the interaction language, intelligent agents can efficiently transmit, understand, and provide feedback expressing knowledge and instructions, with capabilities including dialogue generation, semantic understanding, and autonomous conversation. For example, in response to operational tasks and drawing on prior knowledge, intelligent agents can automatically network on demand and efficiently execute tasks through multiple rounds of "dialogue" and "handshaking."

This interaction language was designed with full consideration of the harsh battlefield environment. By virtue of "semantic noise resistance" (语义抗噪) and "semantic compression" (语义压缩) technologies, even in complex electromagnetic environments with strong interference and high dynamics, core semantic information can still be precisely identified and restored; even with limited bandwidth, clear information can be conveyed with the fewest possible bits. This ensures that even in extreme environments, intelligent agents can still "hear clearly, speak plainly, and stay connected," maintaining precise and efficient coordination.

Forging a Stronger "Neural Network"

As an enabling "force multiplier" on the future battlefield, the development of intelligent communications networks faces the dual challenges of technological innovation and operational application, and every step must be taken with the utmost care.

From a technological dimension, the challenges are primarily reflected in two aspects: computing power allocation and algorithm security. Intelligentized operations impose near-exacting demands on computing power — massive volumes of sensor data, complex AI model training, and real-time collaborative decision-making all require powerful computing support. However, in the highly contested, high-dynamic battlefield environment, there are enormous bottlenecks in the scheduling and supply of computing resources. Even in a hierarchically distributed architecture that processes data as close to the source as possible, it is difficult to fully meet demand. Furthermore, artificial intelligence algorithms themselves have inherent vulnerabilities. Adversarial sample attacks can easily perturb input data; data poisoning can implant backdoors during the model training phase; the "black box" nature of algorithms makes their decision-making processes difficult to trace and audit. Once the "network brain" is induced by the adversary to generate an erroneous knowledge graph, the consequences are unimaginable.

From an application dimension, the intertwining and superimposition of complex electromagnetic environments and cognitive-domain confrontation pose unprecedented tests to the network's resilience and reliability. Competition in the electromagnetic spectrum domain is growing increasingly intense; both sides in a conflict will inevitably use means such as strong jamming, high-power microwave, and hard-kill destruction to render the adversary's intelligent agents into "kites with severed strings." At the same time, attack methods in the cognitive domain are more covert — by fabricating battlefield situations, injecting false feature information, and disseminating deceptive instructions, they induce the adversary's command decision-making to deviate from the correct course, achieving "subduing the enemy without fighting" (不战而屈人之兵). Therefore, the intelligent communications network must not only be able to withstand hard-kill damage at the physical layer, but must also possess the defensive capability to "distinguish right from wrong" (明辨是非) in the cognitive domain.

Building and developing a military intelligent communications network is a grand and arduous systems engineering endeavor. Building such an endeavor requires top-level design from a commanding height, and requires assembling teams and forces to conduct joint research and collaborative innovation across multiple domains. We must adhere to autonomous and controllable development (自主可控), consolidate algorithm security, and forge a truly functional and battle-ready intelligent "nerve center," providing a solid guarantee for winning future wars.

Layout design: Wu Huaijiang

Original Chinese
战场一张网 万物可“智联” ■李玥 刘琦 解放军报记者 贾启龙 智能通信网覆盖全域战场概念图。AI图片 军事通信网,是联结战场武器平台的“神经系统”,是夺取信息优势的命脉所在。其能力高低,直接衡量着一个国家的军事实力。当前,人工智能技术与通信网络技术深度融合,集成先进电子、超材料和智能制造等前沿技术,加速推动通信网络向智能化方向演进。由此诞生的智能通信网络,已非单纯的信息“管道”。在智能化战争时代,它正蜕变为实现“万物智联”的关键基石,如同为战场肌体植入一套高度智能、全域覆盖的“神经网络”。 本期“跟着院士学科技”,我们邀请到军事科学院系统工程研究院研究员、中国科学院院士尹浩,向您讲解什么是智能通信网。 织就覆盖作战全域的智能通信网络 未来战场,其广度与复杂度远超传统认知。陆、海、空、天、电、网等多域力量高度融合,战场态势瞬息万变。智能通信网的核心使命,正是将分布于这些领域的各类侦测设备、指控设备、武器装备、有/无人作战平台以及单兵系统等,高效、韧性地联结在一起,织就一张覆盖物理空间、网络空间乃至认知空间的一体化战场网络。 这张网络由智能“末梢”(如单兵终端、通信车、无人机)、“神经”(路由交换设备)、“中枢”(网络管控设备)及综合“节点”(如指挥所、武器平台)构成。作战单元展开,终端自动入网注册;管控中枢同步指挥神经节点,在终端间建立链路。用户需求一旦提出,系统立即计算最优路径,打通“高速管道”,让数据精准直达目标,实现战场信息服务的即时交付。 具体而言,这张智能网,托起四大关键网络: 首先,构成一张无形的“泛在感知网”。各域侦测设备通过它监测信号强度、频谱占用情况,追踪时延、吞吐量,识别任务优先级、带宽需求,分析用户轨迹、通信模式、资源消耗规律,形成对战场态势完整、清晰、近乎实时的认知。网络可以实时“看见”用户位置,“了解”用户需求,从而提供精准服务。 其次,打造一张敏捷的“指挥控制网”。各级指挥所、指挥控制系统、作战任务规划平台等关键节点,通过智能通信网紧密联结,形成高效的“指挥控制中枢”。依托网络的抗毁韧性,部分节点受损时,网络可实时感知,迅速启动预案,指挥路由节点自主协商,在断点处架设备用信息“桥梁”,确保指挥链路的畅通。 另外,构建一张模块化的“火力打击网”。根据任务需求,管控设备驱动路由交换设备,在不同建制火力单元间快速搭建高优先级临时“数据通道”,实现任务自动解耦、信息高效传输和火力平台按需聚合,将分散的火力单元动态组合为即用即组的模块化打击集群,使火力运用更加灵活。 最后,撑起一张主动服务的“后装保障网”。贯穿作战的运输、油料、弹药、卫勤等保障力量,可通过智能通信网实现保障状态全感知,支撑资源调度信息融合处理与快速分发,为高强度智能化作战提供持续、精准的物资保障信息服务。 打造认知战场的“智慧大脑” 这张庞大的网络如何做到“智能”?关键在于让网络具备认知能力、形成网络知识体系,打造“智慧大脑”。 如同人类通过思维认识世界、积累经验,我们将军事通信网拟人化为认知主体,将“多域感知—学习分析—优化决策”的认知模型植入终端、路由器、网管等设备,它们就会由此蜕变为具备“思考”能力的智能节点。信息通过这些节点,从终端向区域再向中心逐级融合汇聚,最终构建起分级分布式的“网络大脑”,可对战场环境、自身状态、用户行为等多域信息进行感知、学习、分析、预测,形成对网络及其环境的深刻理解——这便是“网络认知”。 如果说网络认知是思考的过程,那么网络知识便是思考的成果。在智能通信领域,网络知识特指对网络元素(节点、链路、协议)及其交互机制的抽象表达。其关键在于“统一表征”——将物理层信号到应用层需求的信息进行标准化描述,以支撑智能体间的理解与应用。 网络知识包含三种类型:“数据信息型知识”是对网络中客观事物的陈述,能实时记录网速、信号、干扰等客观事实,为网络提供真实环境画像,让其“看清”自身状态;“关系计算型知识”通过模型揭示网络参数的映射关系与内在规律,让网络在复杂变化中“洞察”最优解;“逻辑决策型知识”包含处理问题的经验和预案,是网络自主优化演进的“养分”。 就像经验丰富的工程师能听声辨位判断故障,拥有丰富网络知识的智能通信网也具备“思考”能力。它能够预判风险、规划路径、优化资源配置,实现从“被动响应”到“主动预见”的跨越,为战场智能体提供针对性、差异化、主动式的服务,将信息优势转化为决策与行动优势。 为了让种类繁多、架构各异的智能体如“战友”般协同,智能通信网建立了标准化的“交互语言”。这套语言超越了底层数据格式的互通,追求高层语义的共享。通过交互语言,智能体可以高效传递、理解、反馈表达知识与指令,具备会话生成、语义理解、自主对话等能力。例如,针对作战任务,结合先验知识,智能体可通过多轮“对话”与“握手”,按需自动联网,高效实施任务。 这套交互语言设计时充分考虑了严酷的战场环境。凭借“语义抗噪”和“语义压缩”技术,即使在强干扰、高动态的复杂电磁环境下,核心语义信息依然能被精准识别和还原;即使带宽受限,也能用最少的比特传达明确的信息。确保极端环境下,智能体仍能“听得清、说得明、联得上”,保持精准高效的协同。 锻造更强健的“神经脉络” 作为未来战场上的赋能“倍增器”,智能通信网的建设面临着技术创新与实战应用的双重挑战,每一步都需慎之又慎。 从技术维度看,挑战主要体现在算力调配与算法安全两个方面。智能化作战对算力提出了近乎苛刻的要求——海量传感器数据、复杂的AI模型训练、实时的协同决策都需要强大的算力支撑。然而在强对抗、高动态的战场环境下,算力资源的调度与供给存在巨大瓶颈。即使在分级分布式架构体系下尽可能让数据在靠近源头的地方处理,也难以完全满足需求。此外,人工智能算法本身存在脆弱性。对抗样本攻击可以轻易扰动输入数据;数据污染可以在模型训练阶段植入后门;算法的“黑箱”特性导致其决策过程难以溯源和审计。一旦“网络大脑”被敌方诱导生成错误的知识图谱,后果不堪设想。 从应用维度看,复杂电磁环境与认知域对抗的交织叠加,对网络的韧性和可靠性提出了前所未有的考验。电磁频谱领域的争夺日趋白热化,作战双方势必会利用强干扰、高功率微波、硬毁伤等手段,尽力使对手的智能体沦为“断线风筝”。与此同时,认知域的攻击手段更加隐蔽——通过伪造战场态势、注入虚假特征信息、传播欺骗性指令,诱导对手指挥决策偏离正确轨道,实现“不战而屈人之兵”。因此,智能通信网不仅要能抵御物理层面的硬杀伤,还要具备在认知域“明辨是非”的防御能力。 建设发展军事智能通信网,是一项宏大而艰巨的系统工程。建设这样的工程,需要高屋建瓴的顶层设计,需要汇聚团队和力量在多个领域联合攻关、携手创新。我们必须坚持自主可控,筑牢算法安全,锻造出真正管用、顶用的智能“神经中枢”,为打赢未来战争提供坚实保障。 版式设计:吴淮江