Quantum Anti-Spyware Protocols for Fault Injection Detection
Nakshathra Nazer Karakkadan · 2024
Multi-tenancy schemes for quantum computing hardware are vulnerable to fault injection attacks through crosstalk effects. Research studies have experimentally demonstrated the change in output probabilities of victim circuits where the victim and the adversary share the same quantum hardware. To detect adversary circuits, this paper proposes compile time and simulation based protocols to scan quantum computer programs for malicious code patterns. In the compile time approach, with the help of databases of known malicious patterns, the number of gates and the repeating patterns in the code structures are tracked and validated for malicious intent. In the simulation based approach, Idle Tomography (IDT) using pyGSTi software framework is employed for crosstalk error rate extraction. Hamiltonian and Stochastic errors of relevant qubit pairs in accordance with the quantum hardware’s connectivity mapping are analyzed for the detection of unrecognized malicious adversary patterns. Both the protocols output metrics and error rates information that can be used as a "confidence ruler" over a concerned quantum hardware. These protocols safeguard honest quantum circuit parties from possible interference and resulting tampered computational outcomes brought about by adversary parties.