1: \begin{abstract}

2: 

3: The imperfection of measurements in real-world scenarios can compromise the performance of device-independent quantum key distribution (DIQKD) protocols.

4: In this paper, we focus on a specific DIQKD protocol that relies on the violation of the Svetlichny's inequality (SI), considering an eavesdropper utilizing the convex combination attack. Our analysis covers both the three-party DIQKD case and the general $n$-party scenario. Our main result is the relationship between the measurement accuracy and the extractable secret-key rate in all multi-party scenarios. The result demonstrates that as measurement accuracy improves, the extractable secret-key rate approaches $1$, reaching its maximum value when the measurement accuracy is perfect. We reveal that achieving positive extractable secret-key rates requires a threshold measurement accuracy that is consistently higher than the critical measurement accuracy necessary to violate the SI. We depict these thresholds for $n$-party scenarios ranging from $n=3$ to $n=10$, demonstrating that as the number of parties ($n$) increases, both thresholds exhibit a rapid and monotonic convergence towards unity. Furthermore, we consider a scenario involving a non-maximally entangled state with imperfect measurements, where the emission of the initial GHZ state undergoes noise during transmission, resulting in a Werner state. The study further quantifies and demonstrates the relationship between the extractable secret-key rate, the visibility of the Werner state, and the measurement accuracy, specifically emphasizing the three-party scenario. This study aims to illuminate the influence of imperfect measurement accuracy on the security and performance of multi-party DIQKD protocols. The results emphasize the importance of high measurement accuracy in achieving positive secret-key rates and maintaining the violation of the SI.

5: 

6: 

7: 

8: \end{abstract}