Projects
To Do
- Form a team of at least 2 members; register your team.
- Register your project (i.e., say what you will do).
- Schedule project presentation
- Present project
- Demonstrate project - during final examination period
- Deliver project documents during final examination period
Project Ideas
Teams
In the table below, all 30-minute meetings occur on Wednesday in the GSL.
| Start Time | Members | Project Description & Goals |
|---|---|---|
| 3:00 | Gaurav Kumar Mehta Shashank Agarwal Yiming Li |
Execution time based Scheduler with performance monitor Our Scheduler will schedule tasks based on the Average execution time taken on computers. Computer with lowest execution time will get more tasks based on the difference in execution time. Our approach will use push model for assigning tasks to computer. We will be using Multi Core feature of HW5. We will also show a graphical display for showing performance and of computers (might add few more features to it). Our application will include TSP and N-Queens problem. You can also see the current status and progress of our project at http://code.google.com/p/javacentric/wiki/JavaCentricClustering Project Goals
|
| 3:30 | Nupur Garg Jaideep Nijjar |
Implementing the MapReduce framework in Java In our MapReduce implementation we will have a client, a master and workers(2 types:Reduce, Map). A user defines the map and reduce task and sends it over to the master. The user can specify the number of maps and reduce tasks to create. Input files are stored on a common file system.We are going to give the client a graphical user interface. We are also planning to support multiple clients(simultaneously). Applications that we are planning to implement are: word count, distributed grep and statistics on web crawling logs. Project Goals
|
| 4:00 | Scott Bonebrake Adam Doupe Gaurangi Tilak Krithika Ananthakrishnan |
Implementing RMI via a Proxy Server on the Google Android Phone We will be implementing a subset of the Java RMI interface for the Google Android Phone. This interface will be as completely compatible with java.rmi as possible. Due to differences between the serialization of the Android VM and Sun VM, we will be using a remote web service to proxy the RMI calls. This process will be as transparent to the Android developer as possible. We will attempt to use this framework to implement a Mandelbrot Android application that uses RMI calls to a Compute Cluster similar to the homework assignment to do the computation. Project Goals
|
| 4:30 | Prudhvi Chaitanya Dhulipalla Murali Krishna |
This project is to build a substantial MapReduce application that runs over Hadoop cluster. We would like to simulate a real world application which involves computation of connected components for a given undirected graph. Imagine we have a collection of satellite images covering a large wooded region. This data has been analyzed to produce a grid of boolean values, with a resolution of one or two meters. At each grid point, a value of 1 indicates the presence of a tree, while a value of 0 indicates a clear spot without a tree. The problem is all about measuring the extent to which fire spreads in this forest when a careless hiker drops a match at a location (x, y). Also, we would like to measure the expected number of trees that get burnt when a tree at some random position (x, y) is struck by lightning. Here, the assumption is that flames can spread from a burning tree to an immediately adjacent tree in any direction (rectilinearly or diagonally), but flames can't jump across a clear spot. Clearly we can model this process with an undirected graph in which the nodes of the graph correspond to the trees and there is an (undirected) edge betwen adjacent trees. Project Goals
|