Smartphones as a platform for advanced measurement and processing

Current smartphones house a wide variety of sensors and the processing power has increased heavily in recent years. These capabilities can be exploited to implement smartphones as mobile sensing and processing devices in a wide range of applications. This paper discusses two initial biometric applications which were implemented on a standard PC and a Nexus S smartphone to analyze the relative processing power of each. In addition to providing benchmarks for the processing power of the smartphone and PC platforms, such a biometric implementation also serves as a proof-of-concept measurement and processing application. For the initial experiments a speaker recognition system was developed in Java and a face recognition system was developed in C++ (native Android code for the smartphone) using existing open source software libraries. The non-native Java application was found to be about 30 times slower on the phone than on the PC and the native C++ application was about 21 times slower.