Trustworthy Computing using Untrusted Cloud-Based Quantum Hardware

TL;DR

This paper addresses the challenge of ensuring trustworthy quantum computing when using untrusted cloud-based quantum hardware by modeling adversarial tampering and proposing adaptive strategies to mitigate its impact, significantly improving result reliability.

Contribution

It introduces a model for adversarial tampering on quantum hardware and proposes an adaptive shot allocation heuristic to enhance trustworthiness and performance.

Findings

Approx 30X and 1.5X improvement with uniform shot distribution.

Maximum 5X improvement for hybrid classical algorithms.

Adaptive heuristic yields up to 190X improvement in pure quantum workloads.

Abstract

Security and reliability are primary concerns in any computing paradigm including quantum computing. Currently users can access quantum computers through a cloud based platform where they can run their programs on a suite of quantum computers. As the quantum computing ecosystem grows in popularity and utility it is reasonable to expect that more companies including untrusted or less trusted or unreliable vendors will begin offering quantum computers as hardware as a service at varied price or performance points. Since computing time on quantum hardware is expensive and the access queue could be long the users will be motivated to use the cheaper and readily available but unreliable or less trusted hardware. The less trusted vendors can tamper with the results providing a suboptimal solution to the user. In this paper we model this adversarial tampering and simulate its impact on a number of pure quantum and hybrid quantum classical workloads. To guarantee trustworthy computing for a mixture of trusted and untrusted hardware we propose distributing the total number of shots equally among the various hardware options. On average we note approx 30X and approx 1.5X improvement across the pure quantum workloads and a maximum improvement of approx 5X for hybrid classical algorithm in the chosen quality metrics. We also propose an intelligent run adaptive split heuristic leveraging temporal variation in hardware quality to users advantage allowing them to identify tampered or untrustworthy hardware at runtime and allocate more number of shots to the reliable hardware which results in a maximum improvement of approx 190X and approx 9X across the pure quantum workloads and an improvement of up to approx 2.5X for hybrid classical algorithm.