Predictive Resource Management for Wearable Computing

Abstract

Achieving crisp interactive response in resource-intensive applications such as augmented reality, language translation, and speech recognition is a major challenge on resource-poor wearable hardware. In this paper we describe a solution based on multi-fidelity computation supported by predictive resource management. We show that such an approach can substantially reduce both the mean and the variance of response time. On a benchmark representative of augmented reality, we demonstrate a 60% reduction in mean latency and a 30% reduction in the coefficient of variation. We also show that a history-based approach to demand prediction is the key to this performance improvement.

The Weight-Watcher Service and its Lightweight Implementation

“…If the provider's policy is not to limit, but to maximize the number of receivers, it must adapt gracefully, 1 by degrading its quality of service (or application fidelity [12], [13], [16]), i.e., by increasingly compressing the stream. But first of all, in order to implement its resource-aware behaviour through timely adaptations to the fluctuations of its resource environment, the software running on the provider's device must be informed on the amount of resources currently available, as well as the amount of resources required by the piece of code that is currently to be executed (from here on, the piece of code on which the Weight-Watcher gives a prediction in terms of resource consumption will be named an action).…”

“…All kinds of actions, e.g., methods or event handlers, can be profiled as long as they are executed more than once (as in any history-based learning approach, e.g., [12], [13], [19]). In this paper, we handle the following Memory, CPU, Network, Energy, and Time resources, even though the techniques presented are not limited to these specific resources.…”

An energy-aware framework for coordinated dynamic software management in mobile computers

“…break; 16: end if 17: end for 18: end if 19: Figure 8 shows the detailed result of an experiment with the goal set to 1400 seconds and for executing 24 video clips. The Measured line in Figure 8(a) shows how the battery energy supply changes over time, i.e., the energy dissipation rate of mpeg player.…”

“…An application adaptation framework is discussed in [5], but it does not target realistic applications. A recent work [19] provides examples of how the computation fidelity of mobile applications can be altered. However, it does not present a methodology or framework for accomplishing this task automatically, and it targets system delay reduction instead of energy saving.…”

COCOA : ConversationBased Service Composition for Pervasive Computing Environments

“…However, providing accurate metrics measures for the selection of services prior to their execution requires special care, since this relates to predicting the service's resource consumption. The prediction of service metrics can be carried out based on histories [11,24,13], which has been proved to be accurate and efficient [13].…”