5.2.2. General Usability

We now turn to general usability issues as reported by the subjects. First, there were two instances where extreme fault-tolerance was tested during the study. In one instance, a subject reported that the phone's operating system malfunctioned in the middle of a run; but when the phone was restarted after the run, the app resumed its operation from the previous state. In this instance, the data before the crash were not lost, even though they were not submitted to the server prior to the crash.

In another instance, the server was down for a prolonged period of time, during which several runs were taken by the different subjects. Because of the built-in fault-tolerance mechanisms in the Berkeley Telemonitoring framework, no data were lost during the server downtime. The app was able to recover all the data and send them to the server once the server was back online. It is worth reporting that one subject uninstalled the app manually before the server was restarted; which caused the data not yet sent to the server to be lost. These incidents validate the fault-tolerance implementation described in [44].

Besides the aforementioned issues, we further summarize general usability reports made by the study subjects in Table 3.


**Table 3.** Summary of usability reports by the subjects.
