Because a mobile device operates on a finite supply of energy contained in its battery, energy is one of the key resources that needs to be used carefully [99]. In the context of mobile clouds, the cost of participation (such as power consumption) should be less than the benefit gained [100]. Also, it will enable the mobile device to take appropriate component level action to minimize unnecessary energy consumption, and hence, lengthen the system’s life span by unloading unneeded software components, redeploying energy intensive components to more resourceful hosts, and collocating frequently communicating hosts, as suggested in [101]. For these reasons, being aware of a device’s energy usage is vital.
In the following we discuss the research on energy consumption, mainly focusing on work on energy profiling and energy usage estimation.
Energy profiling. In PowerScope [102] the authors present a profiling tool for mobile applications. The tool maps power usage to specific code components in applications and the operating system, allowing an analysis of power draining procedures. Using this analysis, the developers can modify their software to be more energy efficient. The authors report a 46% energy saving by using PowerScope to profile an adaptive video application run on the Odyssey platform [44]. Experiments were carried out with several laptops and pocket computers. Profiling is done offline after collecting data, to ensure no overheads are added to the analysis. A digital multimeter measures the electric current used by the profiling computer and the energy profile is generated using these correlated current measurements. The apps can only be profiled on an open source operating system however, as small modifications to the kernel are required.
In [103], Rice and Hay present a power consumption measurement framework, specifically for mobile phones. In particular, they explore the effect of message size and send buffer size when transmitting data on two Android 1.5-based phones. Power usage during connection to a WiFi network and idle power costs for WiFi, 3G and 2G are also discussed. Power consumption is measured by sampling the voltage drop across the phone battery and a high precision resistor. The mobile device first downloads the test script from a central server and uploads the results to the same server. The central server aligns the test results with the traces and logs. Their findings can be summarized as follows:
- 1.
- When connecting to a WiFi network through DHCP, a significant portion of the time and therefore power, are taken by the ARP Probe packets and the delay between them.
- 2.
- In the case of idle power, WiFi has the lowest energy cost, followed by 3G and 2G respectively. However, it should be noted that the locations of the base stations will also affect radio transmission.
- 3.
- Although it would be logical to expect that the energy cost per byte will decrease as the message size increases, results show that this is not the case. There appears to be a sharp increase of power cost from sending a 7 kB message to 8 kB. However, the reason for this is not clear.
- 4.
- The choice of buffer size can significantly affect the power consumption.
The work discussed previously measures the energy via hardware. A different approach is to take the measurements via software to query battery levels as done in PowerSpy [104], implemented in the Windows operating system. PowerSpy operated in two stages; event tracking and analysis. In the event tracking stage, the application is run and tracked for CPU time, I/O activity and energy consumption. In the analysis stage, the data acquired in the previous stage is processed. To filter out the energy consumed by I/O activities, an estimation of energy usage by various devices to run particular tasks (such as energy used by the disk to read 1 kB) as specified by the devices manufacturer, is used. This estimation is subtracted from the total of recorded energy use, and the remainder is taken as the energy taken up by CPU threads. Next, energy consumption is drilled down to individual threads, on the assumption that CPU power usage is proportional to the number of CPU cycles spent on a thread.
Work done by Cano et al. [105] and [106] provides insight into the energy consumption of the Bluetooth protocol. The focus of their work is on the different states of Bluetooth such as Startup, Standby, Inquiry, Connection and especially the low power modes provided by the protocol.
Energy usage estimation. Work done by Chiyoung et al. [101] and Seo et al. [107]focused on the energy (electrical) consumption estimation of Java based pervasive systems at the level of its system level components. The initial estimation in [101] is done prior to runtime—during construction time. Then the estimation is refined during runtime automatically depending on certain system parameters such as size of data exchanged over the network, inputs to the components’ interfaces, and invocation frequency of components’ interfaces. The exact formulas for calculating the estimation are given in [107]. Their methodology gives an Energy Cost Framework as follows:
- •
- Overall Energy cost = Computational energy cost + Communicational energy.
- •
- System Energy cost = Overall Energy cost + Overall infrastructure energy cost
