55. SOMNILOQUY: AUGMENTING NETWORK INTERFACES TO REDUCE PC ENERGY USAGE
Name: Yuvraj Agarwal
Grad Year: 2009
Reducing the energy consumption of PCs is becoming increasingly important with rising energy costs and environmental concerns. Low power sleep states such as "Sleep" or S3 (suspend to RAM) and "Hibernate" or S4(suspend to disk) save energy, but sometimes they cannot be used because they prevent PCs from performing networking tasks, such as accepting remote desktop logins or performing background file transfers. In this paper, we present Somniloquy, an architecture that augments network interfaces to allow PCs in a "Sleep" or S3 state to be responsive to network traffic. We show that many applications can be supported without application-specific code in the augmented network interface by using application-level wakeup triggers, e.g. for remote desktop and VoIP. A further class of applications can be supported with modest processing and memory resources in the network interface, including instant messaging presence, peer-to-peer file sharing such as Bit-Torent, and long running backgroud web downloads. Our prototype implementation of Somniloquy using a USB peripheral shows significant energy saving potential in experiments, with 11x to 24x less power draw in several scenarios, which in turn translates to significant cost savings for PC users. We are currently working on deploying our system in the Computer Science and Engineering Department at UCSD, thus allowing 750+ Desktop PCs to save energy using Somniloquy. The estimatedsavings in terms of direct energy costs alone are upwards of US $50,000 per year.