Researchers Explore Method to Conceal Quantum Computation from Adversaries

A new approach to secure quantum computation is being explored as multi-tenant cloud platforms become increasingly prevalent. Evan J. D. Anderson and colleagues at University of Arizona introduce the concept of covert quantum computing, designed to prevent detection of computation by an adversary sharing the same quantum processing unit. The research provides a theoretical framework, utilising information theory and quantum-strategy, to analyse covertness and demonstrates that the detection information available to an adversary scales with the square root of the number of qubits in a circuit. Experiments conducted on IQM’s 54-qubit Emerald processor and IBM’s 156-qubit ibm_fez machine reveal previously uncharacterised long-range coupling effects, highlighting potential vulnerabilities and the need for improved spatial isolation in quantum hardware.

Square-root scaling defines a privacy boundary for multi-tenant quantum computation

A strong improvement over previous methods, the amount of information an adversary requires has been reduced to monitoring O(√n) border qubits. Previously, no theoretical guarantee existed on how many qubits an adversary needed to access to remain undetected, representing a fundamental limit to privacy in multi-tenant quantum systems. Establishing this scaling law allows the design of quantum circuits where computation can occur covertly, mirroring classical covert communication techniques. The significance of this lies in enabling a pathway towards practical quantum cloud security, a critical requirement as quantum computers become more accessible. Classical covert communication relies on concealing messages within seemingly innocuous transmissions; this work adapts those principles to the quantum realm, leveraging the principles of superposition and entanglement. The theoretical underpinning draws heavily from information theory, specifically utilising concepts like mutual information to quantify the amount of information an adversary can gain about the computation. The ‘√n’ scaling is particularly important because it suggests that the overhead required to maintain privacy grows sublinearly with the size of the quantum circuit, making it potentially feasible even for large-scale computations. This contrasts with naive approaches that might require monitoring a significant fraction of all qubits, rendering covertness impractical.

Experiments on IQM’s 54-qubit Emerald processor and IBM’s 156-qubit ibm_fez machine confirmed this scaling, although unexpected long-range qubit coupling effects were also observed, potentially creating new vulnerabilities. The 54-qubit Emerald processor, with its square lattice design, and the 156-qubit ibm_fez machine utilising the Heron 2 architecture, a heavy-hex lattice, both verified the square-root scaling law. Ramsey experiments, measuring the frequency of qubit transitions, were conducted on unused qubits surrounding the active computation to map crosstalk. These Ramsey experiments involved applying precisely timed microwave pulses to the qubits and measuring the probability of finding them in a specific state after a defined period. By systematically varying the measurement time, the resonant frequency of each qubit could be determined with high precision. Any deviation from the expected frequency indicated the presence of interactions with neighbouring qubits, revealing the extent of crosstalk. The choice of these two distinct architectures, square lattice and heavy-hex, was deliberate, aiming to assess whether the observed scaling held across different physical qubit layouts and connectivity patterns.

These experiments confirmed expected nearest-neighbour interactions, but also identified long-range qubit coupling effects extending beyond the predicted border, indicating a potential detection pathway not accounted for in the initial model. Leakage originating in the drive and control lines used to manipulate the qubits created this unexpected crosstalk, resulting in a vulnerability. The drive and control lines, responsible for delivering the precise signals needed to perform quantum operations, are susceptible to electromagnetic radiation and capacitive coupling, which can inadvertently affect the state of distant qubits. This leakage represents a significant challenge to achieving true covertness, as it expands the area that an adversary needs to monitor. While the demonstrated scaling limits detection to approximately the square root of the total qubit count, practical covertness requires addressing these additional, currently unquantified, crosstalk channels to achieve true security. Further investigation is needed to characterise the strength and range of these long-range couplings, and to develop techniques for suppressing them, such as improved shielding and signal filtering.

Unexpected qubit crosstalk compromises covert quantum computation security

Concealing computation from prying eyes now frames the promise of secure multi-tenant quantum computing as ‘covert quantum computing’. Long-range qubit coupling, a form of crosstalk extending beyond predicted boundaries, was detected in experiments on both IQM’s and IBM’s processors. Even with a theoretically sound strategy limiting detection to a few ‘border’ qubits, this suggests practical security is undermined by unintended signal leakage from drive and control lines. These findings do not invalidate the concept of covert quantum computing itself, but highlight a key practical challenge. The implications of this crosstalk are substantial; it effectively increases the number of qubits an adversary needs to monitor to potentially detect the computation, diminishing the privacy guarantees offered by the theoretical framework.

The observed long-range qubit coupling, stemming from drive and control line leakage, presents a genuine security risk by expanding the potential detection surface. Understanding these crosstalk mechanisms is vital, allowing for refinement of spatial isolation techniques and improved circuit design to mitigate signal leakage. This work establishes a new understanding of privacy in shared quantum computing, introducing the concept of covert quantum computation where calculations can occur without detection by another user. Employing a quantum-strategy framework, borrowed from game theory, scientists determined that an adversary need only monitor a number of border qubits scaling with the square root of the total qubit count to attempt detection. This approach allows for the design of quantum circuits where computation can occur covertly, mirroring classical covert communication techniques. Future research will focus on developing techniques to characterise and mitigate these crosstalk effects, potentially through improved shielding, signal filtering, and the development of more robust qubit control schemes. The ultimate goal is to create a truly secure multi-tenant quantum computing environment where users can confidently perform computations without fear of detection by malicious actors.

The research demonstrated that computation can occur on a quantum computer without being detected by another user sharing the same machine, a concept termed covert quantum computing. However, experiments on IQM’s 54-qubit and IBM’s 156-qubit processors revealed unexpected long-range coupling between qubits, indicating signal leakage beyond the anticipated ‘border’ qubits. This crosstalk effectively increases the number of qubits an adversary would need to monitor, thereby reducing the privacy offered by the initial theoretical framework. Scientists used a quantum-strategy framework to show that, for an n-qubit circuit, only a number of border qubits scaling with the square root of n provide detection information.