Ranveer Chandra, Steve Hodges, and Anirudh Badam from Microsoft Research and Jian Huang from Georgia Tech present an overview of three such offload techniques that they have developed. The first offloading technique they developed as part of their aim to reduce the power consumption of mobile devices is called Somniloquy. Somniloquy is designed to let a range of tasks with basic computation, memory, and peripheral requirements execute while the rest of the device is in one of the four ACPI G1 states, typically S3 or S4. This is achieved by adding a low-power secondary processor that can be operated independently of the main CPU, such that it can continue to execute lightweight tasks when the CPU and associated subsystems are in a G1 state. The secondary processor is designed to run lightweight appl ications autonomously on behalf of the main CPU. Another approach developed is WearDrive. WearDrive can improve the battery lifetime by offloading storage from a wearable device to a nearby phone. The key insight behind WearDrive is that the battery-powered RAM in a mobile device can be considered persistent as long as the battery capacity is monitored to ensure that the RAM remains reliably powered. The third approach is Mobile Assistance Using Infrastructure (MAUI) system based on two key insights. First, executing a task on a more powerful device takes less time, thereby consuming less energy. Second, the mobile device can go to a low power state while the task is being executed at a remote server.
- green computing
- offloading techniques
- pervasive computing
- power management
ASJC Scopus subject areas
- Computer Science Applications
- Computational Theory and Mathematics