Entanglement witnessing with untrusted detectors

We consider the problem of entanglement detection in the presence of faulty, potentially malicious detectors. A common-and, as of yet, the only-approach to this problem is to perform a Bell test in order to identify nonlocality of the measured entangled state. However, there are two significant drawbacks in this approach: the requirement to exceed a critical, and often high, detection efficiency, and much lower noise tolerance. In this paper, we propose an alternative approach to this problem, which is resilient to the detection loophole and is based on the standard tool of entanglement witness. We discuss how the two main techniques to detection losses, namely the discard and assignment strategies, apply to entanglement witnessing. We demonstrate using the example of a two-qubit Bell state that the critical detection efficiency can be significantly reduced compared to the Bell test approach.