Green Edge Network
P.I.: Young H. Cho, University of Southern California
While rapid growth of the Internet has enabled the great economic and societal benefits, it has also begun to demand more of the energy production and distribution infrastructure. This comes at precisely the time when there is a global concern about our planet's energy consumption. There appears to be no silver bullet that will solve the energy crisis; rather, reducing consumption wherever possible may be the only sustainable long-term solution.
In this proposal, we focus on understanding the principles and methods for the design of green edge networks. By at least one estimate [Mogul07], the total power consumption of edge networks is of the same order as data centers, so even moderate improvements in energy-usage in an individual device can result in non-trivial savings overall. Obtaining these moderate improvements in the energy-efficiency of edge networks is challenging for two reasons. First, the heterogeneity in edge technologies presents different tradeoffs, requiring diverse energy-management strategies. Second, edge networks exhibit significant and unpredictable dynamics in their workload. Moreover, component energy usage in these devices varies significantly even across instances of the same device as a result of variability in the manufacturing process, or due to ambient thermal conditions.
Intellectual Merit. These challenges motivate a unique cross-cutting approach that leverages our teamí»s combined expertise in low-power electronics, link-layer technologies, and energy-efficient network subsystem design and architecture. In this proposal, we will: a) devise a deep energy-inspection architecture that encompasses a broad range of edge devices and networking technologies and incorporates innovative hardware designs for subsystem-level monitoring and control of energy usage; b) explore run-time energy adaptation at various levels of the network, enabled by this inspection architecture; c) examine coordination mechanisms for controlling edge network energy usage which will allow coordinated energy management across components and devices, enabling more aggressive energy savings. Our research will be driven by an innovative heterogeneous green edge network test-bed which will help us explore a large part of the space of run-time energy management techniques for edge networks, and will provide conclusions that generalize beyond the current generation of devices.Broad Impacts. Our proposal will have broader impacts on several fronts. First, our proposal can have significant societal benefit, targeted as it is on sustainable technologies. Second, the techniques that we will develop for energy efficiency in context of edge networks have potential broader applicability to other areas of computing: large server systems, mobile devices, and consumer appliances. Indeed, we envision a future where a fine-grained energy accounting system will be an integral part of all computing devices. Beyond its impact on technology, our project will also contribute to workforce development by training EE and CS students for a future driven by green technologies through innovative courseware at our institutions, and hands-on test-bed experience. As part of the broader education effort, our project will involve underrepresented groups at various levels. Our institutions already have significant efforts in this direction (at CENS and with a diversity-focused REU-site at USC) and our project, especially with its experimental focus on green technologies, can effectively engage underrepresented students. Finally, the infrastructural components resulting from our research (hardware, software, models) will be useful for and made available to the broader education and research community.