Smartphones have become ubiquitous personal devices so that much of sensitive and private information will be saved in the phone, and users have their own unique behavioral characteristics when using smartphones, so, to prevent private information from falling into the hands of impostors, there is a kind of identity authentication system based on user's behavioral features while the user is unlocking. However, due to the impact of environmental factors, changes of gesture will introduce bias into the feature data, which results in a diminishment of the system performance. To solve this problem, we propose an implicit identity authentication system based on keystroke behaviors, and it is the first attempt to consider the changes of a user's gesture. This system collects five keystroke features in the background and analyzes to identify different users without additional hardware supporting. We present our work with an experimental study, and our experiments show that the accuracy of identity authentication system we proposed is up to 99.1329%. Comparing with the identity authentication system without considering the impact of gesture changes, the EER of the system considering the impact of gesture changes is decreased by 1.2514%.

Smartphone is broadly applicable to the human activity recognition (HAR) mobile devices. However, energy consumption becomes a big obstacle to such mobile devices of real‐time monitoring. In order to solve this problem, this paper presents a method of activity recognition based on energy‐efficient schemes. In terms of data acquisition and processing, energy‐efficient schemes adopt the best sample rate and extract the most effective feature combinations in accordance with the different activities, so as not to increase energy consumption; while in terms of recognition algorithm, we adopt the improved structure of multi‐class support vector machine, combine it with the probability of activity occurrence, so as to reduce the time complexity of recognition. This method can minimize energy consumption greatly under the premise of maintaining higher recognition accuracy. Moreover, this paper adopts mobile cloud security technology to reduce potential risk of the smartphone's data transmission and processing. We present our work with an experimental study, and our experiments show that the accuracy of activity recognition based on energy‐efficient schemes we proposed is up to 90.6%. In addition, this method will save 51.0% energy than that when sample rate and extracted features are, respectively, fixed at 100Hz and combined features. Copyright © 2016 John Wiley & Sons, Ltd.