Imperial College Energy and Performance Colloquium 2012 ICEP 2012

Although the capability of smartphones has increased manifold over the past decade, the battery density has barely doubled. Smartphones now have multi-core processors, a high-resolution display, several sensors, and a faster mobile interface – all of which quickly drain the battery, while the battery capacity is limited. In this talk I will present one approach to improve battery lifetime: offloading. The core idea is to move as much computation and other tasks away from the application processor such that the SoC components can stay longer in low power state. We have investigated offloading to several components: a low power processor, the cloud, the GPU and the network interface. We will focus on the first two in this talk. Offloading to a low power processor [1]: In the context of laptops and tablets, sleep states such as S3 (suspend to RAM) save energy, but sometimes they cannot be used because they prevent devices from performing networking tasks, such as accepting incoming connections or performing background file transfers. In this system, called Somniloquy, we present an architecture that augments network interfaces to allow devices to be responsive to network traffic even when the main processor is asleep. 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 VoIP. A further class of applications can be supported with modest processing and memory resources in the network interface, including instant messaging presence and peer-to-peer file sharing. Offloading to the Cloud [2]: In another work, called MAUI, we present a system that enables finegrained energy-aware offload of mobile code to the infrastructure. Previous approaches to these problems either relied heavily on programmer support to partition an application, or they were coarse-grained requiring full process (or full VM) migration. MAUI uses the benefits of a managed code environment to offer the best of both worlds: it supports fine-grained code offload to maximize energy savings with minimal burden on the programmer. MAUI decides at run-time which methods should be remotely executed, driven by an optimization engine that achieves the best energy savings possible under the mobile device’s current connectivity constrains.