Statistical Inference of Serial Communication Errors Caused by Repetitive Electromagnetic Disturbances

Communication errors can be caused by transient electromagnetic disturbances, making the related protection and avoidance measures highly demanding. In this article, we propose a statistical method to infer error mechanisms of serial communication errors caused by repetitive electromagnetic disturbances. An experimental investigation has been conducted by introducing crosstalk disturbances on the transmission cable, resulting in various communication errors. To determine the corresponding failure mechanism for these errors, a hypothesis test method is introduced. First, the bit-flip error on the basis of the known mechanism is simulated through the Monte Carlo method, so that a null hypothesis can be established. Then, by applying a modified chi-squared test, researchers can investigate the influence of multiple pulses on the occurrence probability of bit errors. As a result, nonlinear effects such as energy saturation or accumulation of the transceivers can be discovered. The validated results show that the proposed method has a good performance to determine the known mechanisms of communication errors and infer new regularities with a given significant level.