Multi-attribute operation policies for multi-RAT smartphone

Abstract

Todays smartphones integrate multiple radio access technologies (multi-RAT), e.g., 3G, 4G, WiFi, and Bluetooth, etc. Moreover, state-of-the-art smartphones can activate multiple RAT interfaces simultaneously for the parallel transmission. Therefore, it is becoming more important to select the best RAT set among the available RATs, and determine how much data to transfer via each selected RAT network. We propose Energy, Service charge, and Performance Aware (ESPA), an adaptive multi-RAT operation policies for smartphone with supporting system design and multi-attribute cost function for smartphones Internet services including multimedia file transfer and video streaming services. ESPAs cost function incorporates battery energy, data usage quota, and service specific performance, simultaneously. These attributes are motivated by the growing sensitivity of todays smartphone users to these attributes. Each time the individual attributes are calculated and updated, ESPA selects the optimal RAT set that minimizes the overall cost. It can activate only the best one RAT interface or exploit multiple RATs simultaneously. The primary benefit of the ESPA is that it enables the smartphone to always operate in the best mode without the need for users manual control

the energy saving mode if the remaining battery energy is becoming nearly depleted

the cost-saving mode if the remaining data quota is almost running out

or, the performance-oriented mode if remaining data quota and battery energy are both sufficient. From Chapter 2 to Chapter 4, we cope with file transfer, video streaming, and standby mode for our proposed algorithms. The proposed algorithms are based on the service specific cost or utility models, which also take into account practical issues related to user satisfaction metrics. First, for file transfer mode, we apply the transfer completion time as the performance metric, and the energy consumption and service charge for downloading a specific size of file are simultaneously considered. Furthermore, we especially take into account a problem that the computational complexity exponentially increases as the number of available RATs increases. We propose a heuristic linear search algorithm to find the optimal RAT set without significant performance degradation. Secondly, for video streaming mode, we consider the HTTP-based video streaming model exploiting multipath with LTE and WiFi networks. Based on analysis of the energy consumption and data usage for the video streaming services, we propose a multi-RAT based video streaming algorithm that balances between the video quality, i.e., the performance metric, and the total playback time with currently given battery energy and data quota. Finally, we cope with the battery energy leakage issue of the smartphone in the standby mode due to intermittent traffic generated by some applications running on background. We analyze the energy-consuming factors in the standby mode and smartphone usage patterns of multiple users, and then, propose a usage pattern-aware deep sleep operation algorithm to save the battery energy in the standby mode. Simulation results based on real measurement data of the smartphone show that the ESPA algorithms indeed choose the best operational mode by maintaining dynamic balance among the performance, energy consumption, and service charge considering the currently provided services and the remaining resources.