Towards Sustainable Portable Computing through Cloud Computing and Cognitive Radios Vinod Namboodiri...

15
Towards Sustainable Portable Computing through Cloud Computing and Cognitive Radios Vinod Namboodiri Wichita State University

Transcript of Towards Sustainable Portable Computing through Cloud Computing and Cognitive Radios Vinod Namboodiri...

Towards Sustainable Portable Computing through Cloud Computing and Cognitive Radios

Vinod NamboodiriWichita State University

Sustainability

• The World Wide Fund for Nature, United Nations Environment Programme, and World Conservation Union define sustainability as follows:

• Sustainability is improving the quality of human life while living within the carrying capacity of supporting eco-systems.

Computing Plays a Role

• Anywhere from 3-7% of global energy attributed to Information and Communication Technologies (ICT)

• That is why we have this workshop!

Sustainability – Portable Devices

Energy consumed off the grid

Laptops

Desktops

Data Cen

ters

Mobile Phones

Mobile In

frastr

ucture

Internet

0

50

100

150

200

250

300

350

400

450

Sectorwise electricity consumption (million MWh)

Electronic Waste

Somavat et. al, e-Energy 2010, with updated results

Does not include data center cooling costs

Existing Approaches

• Energy-aware schemes to maximize battery lifetime– Energy efficient protocols at various layers of the

stack – Cross-layer approaches

• Do not necessarily address energy consumed from the grid

• Do not address electronic waste problem

Hardware or Software Approach?

• New hardware could be more energy-efficient• New hardware = more electronic waste!

• Software upgrades through improved protocols, drivers, OS can also lead to energy-efficiency

• Minimizes device replacement

• Favor software approaches where possible

A Proposed Solution

• Rely on Cloud Computing paradigm– portable device executes all applications remotely– more like a thin-client

• Example– Game of Chess on Smartphone– Play locally or online

Application executed locally

on device hardware

Server(s)

Application executed on remote server over a communication network

Non-Cloud Architecture Cloud Architecture

• Periodic hardware upgrades needed on device due to limited local resources

• Periodic hardware upgrades lead to more waste

• Application execution with limited resources could be energy-inefficient for portable devices

• Non-Sustainable

• Fewer or no hardware upgrades needed on device; needed only on server(s)

• Rare hardware updates results in less waste• Application execution on remote, powerful

servers could be energy-efficient• More Sustainable

• Communication will be bottleneck• For portable devices, wireless medium

will have heavy contention• Cognitive Radio could be the answer,

if found energy-efficient

WLAN Access

Cognitive Radios

Courtesy Broadband Wireless Networking Lab, Georgia Tech

Courtesy Anonymous Source

Why Cognitive Radios?

• State-of-the-art solution to wireless spectrum congestion– Can continuously hunt for spectrum that is less

congested– Implemented mainly in software; software

upgrades can keep optimizing communication energy consumption

Are Cognitive Radios Energy Efficient?

• Save Energy– By finding spectrum with • Less contention• Better channel conditions

• Waste Energy– Scanning is a • power-intensive process• Delay inducing process

Merits of CognitionTo Energy Consumption- better channel- less contention

Demerits to obtaining Cognition- power intensive- time consuming

Physical Layer

Channel Conditions

Higher Layer

Node distribution, Channel scanning time, Number of nodes, etc.

Factors under Study

Cognitive Radio Result

Notes:Two radios used; one for scanning one for communicationNode conttention only factor differentiating channelsNumber of Channels Considered = 20. All scanned.

Cloud vs Non-Cloud

C1; Google DocsC2: Office Live

NC: Microsoft Office with WiFi off

Future Work

• Consider many cloud based applications• Understand cloud based network traffic and

optimize energy for communication• Consider cloud based application scenarios

and impact on energy consumption under the cognitive radio model