III. MEASUREMENTS TESTBED
The smartphone used for the measurements is a Nokia N95 [2] which is running Symbian OS as operating system. When making the measurements, we used scripts to control the phone. This allowed us to measure the consumption without interacting directly with the phone, avoiding to press keys and to have the backlight of the display turned on, which could lead to imprecise results. We used Python for S60 [10], [9] as programming language to develop the scripts for testing. We observe that there is no significant penalty in terms of energy and performance by using the Python environment compared to standard Symbian/C++ for the energy levels we deal with throughout this paper. The choice of the mentioned commercial device is due to several reasons. First of all it is a 3G phone and secondly it is able to run the in-built energy profiler [4] developed by Nokia. The Nokia Energy Profiler is an application running on the mobile device that allows to make measurements without any additional hardware. It provides the values for power, current, temperature, signal strength and CPU usage. To further check the correctness of the data measured by the energy profiler on the phone, the complete setup includes the AGILENT 66319D used as millimeter as shown in Figure 3. It is connected to a PC which is running the Agilent 14565B device characterization software, a tool designed for evaluation of portable battery powered device current profiles.