varunsaravagi / compute-offloading-cloudlets

Summer fellowship work- CMU Intel labs

Description

This repository contains work done as part of Summer fellowship during Summer 2015 at Carnegie Mellon University under Kiryong Ha, Ph.D Student under Prof. Mahadev Satyanarayanan and Padmanabhan Pillai, a senior researcher at Intel Labs.

In this work, we have tried to analyze the feasibility of using Cloudlets to speed-up standalone mobile based javascript applications. Cloudlet is a new architectural element that arises from the convergence of mobile computing and cloud computing. It is being actively researched upon by the research group under Prof. Satya. More information on cloudlet can be had from Elijah

We took an already existing javascript application (which can be easily modified to increase both computation and data size) and split it to run in client and server mode. Both client and server are pure javascript based and attempt had been made to do minimal changes to the code base of the original application.

Frameworks

NodeJS has been used to write the server side of the application. Socket.io has been used to communicate between the client and the server. More details on how exactly the application was partitioned can be had from the Partition folder.

Directory Structure

• Partition: This directory contains the work and experiments done to split the application into client and server side.
• Provision: This directory contains the code to dynamically provision the application on the cloudlet or have it run purely on client side.

compute-offloading-cloudlets's Issues

Show simulation

Perform simulation on the server

Do update as many times as the browser refresh rate

Show FPS information

Display simulation effect on screen