安全与密码学2020arxiv.org编辑精选
Air-FI：从气隙隔离计算机生成隐蔽 Wi-Fi 信号
Air-FI: Generating Covert Wi-Fi Signals from Air-Gapped Computers
进阶 · 约 45 分钟读完前置知识：电磁泄漏与侧信道基础、Wi-Fi 物理层基础、计算机体系结构基础
导读
这篇论文研究如何从物理隔离（air-gapped）的计算机中隐蔽地外泄数据：利用内存总线产生的电磁泄漏，在 2.4 GHz Wi-Fi 频段内编码信息，使附近的普通 Wi-Fi 网卡（如通过测量信道噪声）即可接收解码。据评论，用现成 Wi-Fi 芯片作接收端时速率约为每秒 10 比特，属于极低带宽的隐蔽信道。
HN 关注点：其巧妙之处，同时讨论焦点在于这是否只是又一个'纸上谈兵'的侧信道研究——前提是目标机器已被植入恶意软件。
HN 评论精选
这几条评论里：质疑 1 补充 2 亲历 1
质疑标题夸大：实际是 Wi-Fi 频段的噪声编码，速率仅 10bps，且该实验室类似论文工厂，前提是机器已中毒，学术意义大于实用。
标题有误导性。这并不是生成 Wi-Fi 信号，而是在 Wi-Fi 频段生成非 Wi-Fi 信号（本质上是噪声），以编码信息的方式被现有 Wi-Fi 芯片检测到（例如测量信道噪声），形成一个极低带宽的通信信道。用现成 Wi-Fi 芯片作接收端，速率约 10 比特/秒。这确实有趣且聪明，但作为安全研究者，我认为这个实验室有点像'论文工厂'：他们把计算机任何可被远程接收的随机辐射包装成突破物理隔离的手段。问题在于，一旦接受以下两点：1) 如果你的隔离机器已被植入能做这件事的恶意软件，你早就输了；2) 如果真要防御这种威胁，你的'物理隔离'得是一个隔绝声音、电磁干扰及一切物理传输介质的密封金库——那么这整个研究方向就纯属学术游戏了。他们做过 LED、温度、电感噪声、超声波等。这些东西当然都可行，很巧妙，但并不冷门也不难实现，而且全都依赖目标机器上先有恶意软件。计算机本来就是嘈杂的设备，在噪声中编码信息我们几十年前就知道了。
marcan_42查看英文原文
Title is misleading. This isn't generating Wi-Fi signals. It's generating non-Wi-Fi signals (i.e. effectively noise) in the Wi-Fi frequency bands, in a way that encodes information and can be detected using existing Wi-Fi chipsets (e.g. measuring channel noise), as a very low bandwidth communications channel. The bit rate with off the shelf Wi-Fi chipsets as receivers is ~10 bits per second. This is definitely interesting and clever, but as a security researcher I will say I consider this particular research lab a bit of a paper mill. Their entire schtick is they pick any random emission from a computer that can be picked up remotely, and hype it up as an airgap-defeating measure. The thing is, once you accept that either 1) if your air-gapped computer has malware that can do this, you've already lost, or 2) if you need to be resilient to that, your "air gap" needs to be a sealed vault insulated from sound, EMI, and any other physical transmission medium, then this whole body of research becomes purely academic. IIRC they've done LEDs, temperature, inductor noise, ultrasound, etc. When I first started looking into this I could come up with a good dozen of their ideas without looking. Of course all this stuff works. It's cute, it's clever, but it's not particularly obscure nor difficult to make work, and it all relies on having malware on the target machine to begin with. Yes, computers are noisy beasts, and you can encode information in the noise; we've known this for decades, picking a new technique and implementing it isn't particularly interesting after the fifth or sixth time. Personally, I find their techniques primitive and boring. Like here, they set the RAM bus frequency to 2.4GHz (DDR-2400). Yes, of course, if you do that, then RAM traffic generates noise in the Wi-Fi band. Then they just used the most boring and trivial encoding possible on top of that. I don't remember ever reading one of their papers and thinking they'd used a clever technique. It always seems to be "pick a leak vector, then do the least amount of work possible to make it work at a few bits per second and get a paper out". The interesting side channel papers are those where you can get information from emissions (e.g. crypto keys) without malware on the target computer, during normal operation. Those are real threats. And this is not what this group is researching.
亲历亲历分享：利用坏网卡干扰/不干扰同轴以太网的状态来传比特，是类似的隔离穿透手法。
这让我想起我同事曾用来从一台'物理隔离'设备提取数据的黑客手段。那台电脑的以太网卡坏了，无法发送数据包，但可以干扰同轴以太网段上连接的其他网卡（所有网卡通过同一根导线电气相连）。于是他写了一段代码，通过'干扰'和'不干扰'两种状态来传输比特。那是在 DOS 上，由于可以直接访问硬件而绕过操作系统，实现起来更容易。
praptak查看英文原文
This reminds me of a hack that my colleague used to extract data from an "airgapped" device. The device was a computer that had a broken Ethernet card. It wasn't able to send packets but could jam the other cards connected to the Ethernet segment (coax Ethernet - all cards in the local network were galvanically connected to the same wire). So he wrote a piece of code to transmit bits via the "jammed" and "unjammed" states. This was on DOS, so it was easier due to all the hardware being directly available without OS getting in the way.
补充指出内存频率与 2.4 GHz Wi-Fi 频段巧合重合是该方法可行的关键。
大多数内存恰好运行在 Wi-Fi 所需的 2.4 GHz 频率上（无论是否超频），这真是个相当大的巧合。
erdewit查看英文原文
It's a pretty big coincidence that most memory runs at the required Wi-Fi frequency of 2.4 GHz (overclocked or not).
补充联想到对超级 AI 进行可靠物理隔离的难度。
这让我想到，要把一个超级 AI 真正物理隔离起来，恐怕会非常困难。
Sosh101查看英文原文
This reminds me about how hard it would probably be to keep a super AI airgapped.
相关资源
从这篇论文所有 HN 讨论帖的全部评论里整理出来的链接。
历次讨论
- 2020Hacker News 讨论帖294 分 · 105 评
同领域还值得读
如何入侵一台已关机的电脑：在 Intel ME 中运行未签名代码
How to hack a turned-off computer, or running unsigned code in Intel ME
这篇 Black Hat 2017 白皮书讨论 Intel Management Engine（ME）的安全问题：如何在 ME 中运行未签名代码，以及攻破处于关机状态的电脑的相关…
安全与密码学2017 · 543 分 · 107 评专业 · 约 40 分钟编辑精选比特币：一种点对点的电子现金系统（2008）
Bitcoin: A Peer-to-Peer Electronic Cash System (2008)
中本聪 2008 年发表的比特币白皮书，提出了无需可信第三方的点对点电子现金系统。
安全与密码学2017 · 536 分 · 394 评入门 · 约 45 分钟编辑精选GPRS 加密算法 GEA-1 的密码分析表明存在故意弱点
Cryptanalysis of GPRS Encryption Algorithms GEA-1 suggest intentional weakness
首次公开了对 GPRS 加密算法 GEA-1 和 GEA-2 的密码分析攻击。
安全与密码学2021 · 487 分 · 107 评专业 · 约 90 分钟编辑精选Signal 消息协议的形式化安全分析
A Formal Security Analysis of the Signal Messaging Protocol
用形式化方法分析 Signal 消息协议（TextSecure）的安全性：作者在多阶段密钥交换模型下，借助自动化证明工具对协议的密钥协商与 Double Ratchet 密钥更新…
安全与密码学2016 · 353 分 · 230 评专业 · 约 120 分钟编辑精选A2：模拟恶意硬件
A2: Analog Malicious Hardware
论文研究如何在芯片设计阶段植入模拟电路形式的恶意硬件后门，使其难以被常规检测发现。
安全与密码学2016 · 315 分 · 94 评专业 · 约 75 分钟编辑精选