💻 computer science

Aquaman: A Transparent Proxy Architecture for Quantum Resilient Key Establishment

This paper introduces Aquaman, a transparent proxy architecture that mitigates harvest-now, decrypt-later threats by enabling quantum-resilient session-key establishment for legacy clients through an implemented multi-path key fragmentation mode that distributes ciphertext fragments across diverse network media without requiring client-side configuration.

Original authors: Tushin Mallick, Ashish Kundu, Ramana Kompella

Published 2026-05-12

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tushin Mallick, Ashish Kundu, Ramana Kompella

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Problem: The "Harvest Now, Decrypt Later" Trap

Imagine you are sending a secret letter today. You lock it in a strong box (encryption) that only the recipient has the key to. You feel safe.

But there is a villain in this story: a time-traveling thief. This thief doesn't have the key today, but they are stealing and storing your locked box right now. They are waiting for the day when they build a "Super-Magic Hammer" (a powerful quantum computer) that can smash any lock, no matter how strong.

Once that hammer is built, the thief goes back to their vault, smashes your old box, and reads your secrets. This is called "Harvest Now, Decrypt Later." Even though the hammer doesn't exist yet, the threat is real because your secrets might need to stay safe for 20 or 30 years.

The Solution: Aquaman (The Transparent Proxy)

The authors built a system called Aquaman. Think of Aquaman as a super-smart, invisible bodyguard standing at the gate of your secure network.

Usually, to get a new key for your secret box, your computer has to talk directly to the key-maker. But many old computers (like those in hospitals, factories, or old phones) are too dumb or too old to understand the new, super-secure locks (Post-Quantum Cryptography) needed to stop the time-traveling thief.

Aquaman solves this by:

Sneaking in: It sits at the edge of the network. It doesn't ask your old computer to change its settings or install new software. It just intercepts the request.
Doing the heavy lifting: It talks to the high-tech key-maker on behalf of your old computer.
Protecting the old: It ensures the key is delivered safely, even if your computer doesn't know how to use the new "magic" locks.

How Aquaman Delivers the Key: The "Multi-Path" Trick

The paper focuses on a clever way to deliver the secret key that doesn't rely on a single, unbreakable lock. Instead, it relies on confusion and diversity.

Imagine you have a secret code (the key) that is 100 digits long. Instead of sending the whole code in one envelope, Aquaman does this:

Shreds the Code: It cuts the code into 8 tiny pieces (fragments).
Locks Each Piece: It puts each tiny piece in its own small, locked box.
Sends Them Everywhere: It sends these 8 boxes to the recipient using 8 completely different delivery methods at the same time:
- One goes via Wi-Fi.
- One goes via Bluetooth.
- One goes via NFC (like tapping a phone).
- One goes via Cellular data.
- One goes via a wired Ethernet cable.
- (And so on).

The Magic Logic:
To steal the secret, the thief needs to catch ALL 8 boxes and break ALL 8 locks.

If the thief is good at hacking Wi-Fi, they might catch that box.
But to catch the Bluetooth box, they need a totally different set of tools and to be standing right next to the device.
To catch the NFC box, they need to be even closer.

The paper argues that it is practically impossible for a thief to be everywhere at once with every different tool needed to hack every different type of connection. If they miss even one box, they can't put the puzzle together, and the secret remains safe.

The Four Modes of Operation

Aquaman has four ways to work, though the authors only built and tested the first two:

Mode 1 (The Bodyguard for the Unprepared): The client (your old computer) doesn't know anything about quantum security. Aquaman handles all the complex math and sends the key fragments over different channels.
Mode 2 (The "Ghost" Delivery): If no one trusts the new locks at all, Aquaman splits the key into pieces and sends them through a mix of different networks (Wi-Fi, Bluetooth, etc.). It can even hide who is asking for the key by bouncing the request through a pool of anonymous helpers, so the thief doesn't know who to watch.
Mode 3 (The Quantum Wire): (Discussed but not built) Using a special "Quantum Key Distribution" machine that uses physics (light particles) to generate keys. This is super secure but requires expensive, specialized hardware.
Mode 4 (The Hybrid): (Discussed but not built) Using a mix of old locks and new locks together. If one breaks, the other holds.

What They Found (The Results)

The team built a prototype and tested it 1,000 times on Amazon's cloud servers. Here is what they learned:

Speed: The system is fast enough to be useful. The time it takes to send the key pieces is mostly just the time it takes for the internet to carry the data. The "cutting and shuffling" part of the process is very fast and doesn't slow things down much.
The Bottleneck: The slowest part was actually the "un-locking" of the tiny boxes on the receiving end. However, they found a way to speed this up significantly if they use a "wrapper" technique (using a temporary key to lock the pieces).
Safety: The math proves that as you add more different types of delivery channels (Wi-Fi, Bluetooth, etc.), the chance of a thief stealing the whole key drops dramatically. It's like trying to steal a diamond by breaking into a house, a bank, a car, and a safe all at the exact same second—it's too hard.

The Bottom Line

Aquaman shows that we don't have to wait for every single computer in the world to get upgraded to "Quantum-Proof" software to be safe. We can use a smart gateway (the proxy) to handle the heavy lifting and deliver keys in a way that is so scattered and diverse that even a future super-computer thief can't catch them all. It's a practical, "defense-in-depth" strategy for the quantum age.

Technical Summary: Aquaman

Problem Statement
The paper addresses the "harvest-now, decrypt-later" (HNDL) threat, where adversaries intercept and archive ciphertext today to decrypt it retrospectively once fault-tolerant quantum computers (CRQCs) become available. This renders current public-key primitives (RSA, Diffie–Hellman, ECC) insufficient for data requiring long-term confidentiality. While Post-Quantum Cryptography (PQC) is being standardized (e.g., NIST FIPS 203/204/205), immediate universal adoption faces significant operational friction: legacy hardware cannot easily be updated, new PQC implementations have shown unexpected vulnerabilities (e.g., the break of SIKE), and PQC primitives introduce larger key/ciphertext sizes that strain bandwidth-limited protocols. Consequently, there is a need for a quantum-resilient key establishment mechanism that can be deployed in environments lacking PQC-capable software stacks.

Methodology and System Architecture
The authors propose Aquaman, a transparent-proxy architecture for quantum-resilient session-key establishment. Unlike explicit proxies requiring client configuration, Aquaman operates at the network edge of a trusted network, intercepting session-key requests without modifying client devices. The system comprises three logical components:

Quantum Key Management Server (QKMS): Generates session keys, fragments them, and dispatches them across heterogeneous channels.
Transparent Proxy: Intercepts client traffic, forwards requests to the QKMS, aggregates fragments received via diverse channels, and forwards the reconstructed key to the client. Crucially, the proxy never sees the key in plaintext.
Client: Requests keys and reconstructs them from received fragments.

Aquaman supports four operating modes, though the paper focuses on implementing and evaluating the first two:

Mode 1 (PQC Offloaded): Clients without PQC connect classically to the proxy, which performs PQC Key Encapsulation Mechanism (KEM) operations with the QKMS on the client's behalf.
Mode 2 (Classical Multi-Path): When no party trusts PQC, the system splits the session key into $n$ ciphertext fragments. These fragments are distributed across structurally diverse transmission media (e.g., Wi-Fi, Bluetooth, NFC, cellular, Ethernet). Reconstruction requires all fragments.
Modes 3 & 4 (Not Implemented): The paper discusses but does not implement QKD key delivery (via SKIP/ETSI interfaces) and classical/PQC hybrid handshakes, noting these are well-covered in existing literature.

Security Model and Formal Argument
The security of Aquaman's multi-path mode (Mode 2) does not rely on the computational hardness of a single algorithm against quantum attacks. Instead, it relies on the practical infeasibility of an adversary simultaneously compromising every channel in a structurally diverse set.

Threat Model: The adversary has a limited "capacity budget" $B$ to compromise $d$ distinct medium types. Compromising different media (e.g., Wi-Fi vs. NFC) requires distinct physical proximity, hardware, and operational footprints.
Formal Proof: The authors formalize the recovery probability. If a key is split into $n$ fragments distributed uniformly across $d$ media types, and the adversary can fully compromise a budget of $B$ media types, the probability of recovering the key decays as $(B/d)^n$ .
Theorem 1: The paper proves that with uniform fragment distribution, the maximum recovery probability is minimized when fragments are spread evenly, and this probability decays exponentially with the diversity dimension $d$ .

Key Contributions

Transparent Proxy Architecture: A novel design that extends quantum resilience to legacy clients lacking PQC stacks by offloading key establishment to the network edge.
Multi-Path Key Fragmentation: A mechanism that splits session keys across heterogeneous physical media (not just IP paths), leveraging the high cost for an adversary to simultaneously compromise diverse transmission types.
Formal Security Analysis: A capacity-based model proving that key recovery probability decreases exponentially with the diversity of transmission media, independent of specific algorithmic hardness assumptions.
Prototype Implementation: A working prototype implemented in Python (Flask) and deployed on AWS EC2, supporting Wi-Fi and Bluetooth channels, with an NFC-based bootstrap mechanism for initial trust establishment.

Evaluation Results
The authors conducted 1,000 runs of the prototype to evaluate latency:

Latency Dominance: End-to-end latency is dominated by network transmission time, not by the cryptographic overhead of fragmentation, shuffling, or multi-path dispatch.
PQC Overhead: In scenarios utilizing a PQ KEM (Mode 1), the additional latency was bounded (approx. 100–300 ms) and largely parallelizable with network transit.
Decryption Bottleneck: In the non-PQ scenario, per-fragment asymmetric decryption was the primary computational cost (approx. 717 ms). The authors note this can be optimized by using symmetric encryption for fragments wrapped by a single asymmetric key, reducing decryption time to microseconds (as observed in the PQ scenario where the tunnel is already symmetric).

Significance and Claims
The paper claims that Aquaman demonstrates a practical path to achieving session-key confidentiality against quantum adversaries today, using commodity classical media.

Complementarity: The design is presented as complementary to the standardization-track migration to PQC. It can serve as a standalone hedge in PQ-incapable environments or as a "defense-in-depth" layer stacked on top of PQC KEMs to survive potential future breaks in specific PQC implementations.
Operational Viability: The evaluation suggests that the multi-path overhead is negligible compared to standard network transmission, making the architecture deployable without significant performance penalties.
Modesty: The authors explicitly state that Modes 3 and 4 are not implemented and that the security model assumes independent compromise of media types, acknowledging that correlated capabilities (e.g., a nation-state controlling multiple infrastructure types) would require further modeling. The work focuses on the architectural novelty of the proxy and the multi-path mechanism rather than introducing new cryptographic primitives.