To perform cooperative tasks in a decentralized manner, multi-robot systems are often required to communicate with each other. Therefore, maintaining the communication graph connectivity is a fundamental issue when roaming a territory with obstacles. However, when dealing with real-robot systems, several sources of data corruption can appear in the agent interaction. In this paper, the effects of failure and noise in the communication between agents are analyzed upon a connectivity maintenance control strategy. The results show that the connectivity strategy is resilient to the negative effects of such disturbances under realistic settings that consider a bandwidth limit for the control effort. This opens the perspective of applying the connectivity maintenance strategy in adaptive schemes that consider, for instance, autonomous adaptation to constraints other than connectivity itself, e.g. communication efficiency and energy harvesting.