The main idea of this paper is very interesting. A connection between a smart watch and a smart phone is established to work with a common shared human interface using the accelerometer present in both devices to recognize user gestures, dealing with both devices as a single unified platform. These gestures are proposed as a way to overcome the limitations of direct touch on wrist-worn devices.

The paper begins with a very good review of the literature about the interaction techniques for handheld and wrist-worn devices, both individually and when they are associated. Some of these ideas are used in individual commercial devices, such as the Pebble smart watch, which can be shaken to turn on the backlight, or in iOS devices, which can be shaken to undo the last operation. Similar actions can initiate file transfers between a pair of Android devices. In fact, Pebble (and the like) should appear in the paper (in Table 1) as a device used for context and activity sensing (with applications such as Morpheuz, SwimIO, and others), operating in the background based on Buxton’s framework [1].

The authors provide a set of six types of gestures, in addition to gesture recognition methods based on the movement, orientation, and touching of both devices, for the interaction between the watch and the mobile phone. They use machine learning techniques and hard-coded heuristics for implementing their recognition system. More than 7,000 data points were taken and analyzed, providing results of over 90 percent mean accuracy in the detection of the gestures for different people. A set of applications (Duet) is used to test the usability of the proposal.

The paper is highly applicable to contemporary mobile devices.