Tamper Detection against Unitary Operators
This work addresses security for quantum storage devices against tampering, extending classical cryptography to quantum capabilities, but it is incremental as it builds on prior classical results.
The paper tackles the problem of detecting tampering by quantum adversaries using unitary operators, extending classical tamper detection codes to the quantum setting. It shows that quantum tamper detection codes exist for unitary families of size up to 2^{2^{αn}} with α<1/6, provided the operators are not too close to identity, and also constructs non-malleable codes for classical messages under similar constraints.
Security of a storage device against a tampering adversary has been a well-studied topic in classical cryptography. Such models give black-box access to an adversary, and the aim is to protect the stored message or abort the protocol if there is any tampering. In this work, we extend the scope of the theory of tamper detection codes against an adversary with quantum capabilities. We consider encoding and decoding schemes that are used to encode a $k$-qubit quantum message $\vert m\rangle$ to obtain an $n$-qubit quantum codeword $\vert {ψ_m} \rangle$. A quantum codeword $\vert {ψ_m} \rangle$ can be adversarially tampered via a unitary $U$ from some known tampering unitary family $\mathcal{U}_{\mathsf{Adv}}$ (acting on $\mathbb{C}^{2^n}$). Firstly, we initiate the general study of \emph{quantum tamper detection codes}, which detect if there is any tampering caused by the action of a unitary operator. In case there was no tampering, we would like to output the original message. We show that quantum tamper detection codes exist for any family of unitary operators $\mathcal{U}_{\mathsf{Adv}}$, such that $\vert\mathcal{U}_{\mathsf{Adv}} \vert < 2^{2^{αn}}$ for some constant $α\in (0,1/6)$; provided that unitary operators are not too close to the identity operator. Quantum tamper detection codes that we construct can be considered to be quantum variants of \emph{classical tamper detection codes} studied by Jafargholi and Wichs~['15], which are also known to exist under similar restrictions. Additionally, we show that when the message set $\mathcal{M}$ is classical, such a construction can be realized as a \emph{non-malleable code} against any $\mathcal{U}_{\mathsf{Adv}}$ of size up to $2^{2^{αn}}$.