We address the problem of detecting execution errors in communication protocols. A communication protocol is modeled as or finite state machine (FSM) that can be used as an external observer for detecting execution errors. Wang and Schwartz (1992, 1993) introduce the concept of multiple observers obtained by an adequate decomposition of the FSM. We first address the decomposition procedure from the perspective of error control codes and show that the decomposition algorithm can be restated as a simple state coding algorithm. Then, we discuss the features of fault tolerance of the resulting decomposition. We generalize the concept of multiple observers into the one of fault tolerant multiple observers. A set of observers is said to be fault tolerant if it is capable of detecting the execution errors of a protocol even when a subset of the observers is faulty. We show that error control codes can be used to generate multiple observers that are fault tolerant. We illustrate our approach on the ISO transport protocol class 4 (TP4). Finally, we give some hints on how to assign codes to the states while maximizing the fault coverage of the resulting decomposition.