【軍傳媒／軍事科技】近年來，中國對周邊海域持續採取灰色地帶戰術，破壞區域規則並衝擊既有國際秩序。俄羅斯亦在歐洲周邊複製類似手法，使水下關鍵基礎設施（包括離岸風電、海底油管與輸電纜線）在過去 5 至 10 年間面臨愈加頻繁的人為威脅。2022 年 9 月 26 日連接俄羅斯與德國的海底天然氣管道 Nord Stream 1 與 Nord Stream 2 在波羅的海海底遭到爆炸破壞。2024年11月18日連接立陶宛和瑞典的海底電纜，以及連接芬蘭和德國的數據通信海底電纜C-Lion1被中國籍貨輪 Newnew Polar Bear 拖錨切斷。而自2014到2024的十年間，約有1,012艘被偵測到的俄羅斯籍可疑船隻在北海活動，其中至少950 次靠近海底纜線或管線的1公里內，正是此一威脅的最佳案例。
面對此一新型態安全環境，民主國家已開始強化對關鍵基礎設施的長期監視與警戒能力，而前篇文章提到的「分散式聲學感測」（Distributed Acoustic Sensing, DAS），正是目前逐漸受到重視的技術之一。
AI 的突破帶來關鍵轉折。過去難以從噪訊中辨識的微弱震動，如今能透過人工智慧與大數據分析被重新解讀。然而必須釐清的是：DAS 並非一套獨立、全面性的偵測系統，而是一種「輔助性」感測來源。其最佳效益來自與其他偵測手段——如雷達、聲納與 EO/IR 光學識別技術——整合後所形成的多源情資架構。
DAS 技術本身並不新穎，問世已逾 40 年，最初應用於採礦與地震監測領域。如今隨著 AI 能力提升、以及國際社會對水下基礎設施安全需求的激增，這項技術再度受到重視。其核心概念為：利用海底光纖所接收的震盪訊號，透過 AI 進行聲紋分析與分類，而使其真正具備實用價值的基礎，即為其背後的「數據庫規模與品質」。
為深入理解此技術的最新應用，本刊特別訪談 BAE Systems 水下系統部專案經理 Matt Willsher。他指出，DAS 是具成本效益且能明顯提升偵測準確度的輔助型系統，「它不是完整的解決方案，但透過高度整合，卻可大幅提升整體海域警戒能力。」
Willsher 強調，DAS 的真正價值在於「預警」：它能協助使用者辨識潛在的可疑活動，讓軍方或政府單位提前部署與反應。例如，藉由聲紋辨識模式，系統可辨識接近海底基礎設施的可疑船隻，並結合岸基雷達、EO/IR 影像或其他感測器，迅速引導艦艇或航空器前往查證。
在系統架構上，DAS 的部署方式相對簡單。只需在海纜上岸點設置訊號集合器接收震動訊號，再將其輸入整合情報平台，與岸基雷達與其他感測器共同融合，即可形成一個初步的海域監控網。
然而，DAS 仍存在其侷限性。其一，由於依賴海纜作為感測媒介，偵測範圍主要集中在纜線周邊；儘管如此，由於 DAS 的核心目的正是保護這些關鍵基礎設施，因此此限制仍在可接受範圍內。其二，海況、海流、溫度等自然條件會大幅影響聲波傳遞與系統判讀距離；若對手刻意採取低噪音接近，也會降低預警效率。此外，系統建立後需要長時間累積聲紋資料，才能達到穩定且成熟的辨識能力。
目前國際間並未交換此類敏感的聲紋資料庫，主因涉及國安與軍事機密。若台灣未來建置相關系統，勢必得自行建立聲紋資料庫。雖然其投入需具長期性，但一旦累積完成，便可如同冷戰時期的大西洋「海底監聽長城」般，提供台灣周邊海域更高解析度的感測能力。結合 C4ISR 架構——雷達、EO/IR、聲納、無人機與海巡目視資料——便可形成更完整的「台灣周邊海域共同作戰圖像」。
全球水下關鍵基礎設施正面臨愈加頻繁的灰色威脅，各國也因此投入更多資源發展相關技術與系統。DAS 與 AI 所帶來的跨領域整合能力，正逐漸成為下一階段水下保護戰略的重要拼圖。台灣需要未雨綢繆，逐漸建立自身的人工智慧分析資料庫，對於有敵意有威脅的水面水下航行器建立聲紋分析，除了能預警海底電纜被破壞之外，隨這科技進步，水下戰場環境監視能力對四面環海的臺灣將愈發重要。
A New Generation of DAS- and AI-Driven Early-Warning Capabilities
In recent years, China has intensified gray-zone operations across surrounding waters, undermining regional norms and disrupting the established international order. Russia has adopted similar tactics around Europe, creating mounting risks for undersea critical infrastructure—offshore wind farms, subsea oil pipelines, and power or data cables. Over the past five to ten years, these assets have faced increasingly frequent man-made threats.
On 26 September 2022, the Nord Stream 1 and Nord Stream 2 natural gas pipelines connecting Russia and Germany were severely damaged by underwater explosions in the Baltic Sea. On 18 November 2024, an undersea power cable linking Lithuania and Sweden, as well as the C-Lion 1 data cable running between Finland and Germany, were severed after the Chinese-flagged cargo vessel Newnew Polar Bear dragged its anchor across them. From 2014 to 2024, approximately 1,012 suspected Russian vessels were detected operating in the North Sea—at least 950 times venturing within one kilometer of seabed cables or pipelines. These incidents illustrate the growing severity of the threat.
In response to this evolving security landscape, democratic nations have begun reinforcing long-duration monitoring and early-warning capabilities for critical infrastructure. One of the increasingly prominent technologies—introduced in the previous article—is Distributed Acoustic Sensing (DAS).
A major turning point has emerged with the rise of artificial intelligence. Weak vibration signals that were once indistinguishable from noise can now be reinterpreted using AI and big-data analytics. It is essential, however, to clarify that DAS is not a standalone or comprehensive surveillance system. Instead, it is an auxiliary sensing layer, providing greatest value when fused with other sensors such as radar, sonar, and EO/IR optical identification systems.
DAS itself is not a new technology—it has existed for more than 40 years, originally applied in mining and seismic monitoring. Today, with rapid advances in AI and growing global concerns over subsea infrastructure security, DAS is regaining strategic importance. Its core concept is simple: vibrations captured through subsea fiber-optic cables are analyzed via AI-driven acoustic classification. Its actual operational value, however, depends on the scope and quality of the underlying acoustic database.
To better understand the latest applications of this technology, Military Media interviewed Matt Willsher, Program Manager at BAE Systems’ Underwater Systems division. Willsher noted that DAS is a cost-effective auxiliary system that significantly enhances detection accuracy. “It is not a complete solution,” he emphasized, “but when highly integrated, it can greatly strengthen maritime awareness.”
Willsher highlighted that the true value of DAS lies in early warning. It helps operators identify potentially suspicious activity, allowing military or governmental agencies to mobilize and respond ahead of time. Through acoustic-signature matching, DAS can identify vessels approaching undersea infrastructure and, when fused with shore-based radar, EO/IR imagery, or other sensors, can rapidly cue naval or aerial units for verification.
In terms of system architecture, DAS is relatively straightforward to deploy. By installing a signal collector at the landing point of an undersea cable, vibration data can be gathered and then fed into an integrated intelligence platform—fused with land-based radar and other sensors—to form an initial maritime monitoring network.
Nevertheless, DAS does have limitations. First, because it relies on seabed cables as sensing media, its detection range is concentrated along the cable’s immediate vicinity. However, since DAS is specifically intended to protect these assets, this limitation is acceptable. Second, environmental variables—sea state, currents, temperature—can significantly influence acoustic propagation and detection range. Adversaries deliberately using low-noise approaches may also reduce warning effectiveness. Moreover, the system requires long-term accumulation of acoustic data before its identification performance becomes stable and mature.
Currently, nations do not exchange such sensitive acoustic databases due to national security and military confidentiality concerns. Should Taiwan pursue similar systems in the future, it will need to build its own acoustic-signature repository. Although this demands long-term investment, once established, it could function much like the Cold War–era Atlantic SOSUS network—providing higher-resolution sensing around Taiwan’s maritime approaches. When integrated into a C4ISR framework—radar, EO/IR, sonar, UAVs, and Coast Guard visual reporting—it can contribute to a comprehensive Common Operating Picture (COP) of Taiwan’s surrounding waters.
As global undersea infrastructure faces increasingly frequent gray-zone threats, nations are directing more resources toward technologies and systems that can protect these critical assets. The cross-domain fusion capabilities enabled by DAS and AI are becoming essential building blocks in the next phase of undersea-security strategies. For Taiwan, the priority should be to prepare early—gradually developing a domestic AI-driven acoustic database for suspicious surface and subsurface contacts. Beyond warning of cable sabotage, advancements in underwater-environment sensing will only grow more critical for an island nation surrounded on all sides by the sea.