Nithya Kumar

Develop a program using POSIX threads to find the number of occurrences of an input search string in a given new-line delimited source.
Searches are parallelized among threads where each thread operates on a different data block (not necessarily terminated by a newline). The number of matches per block (partial result of a thread) may be stored in a global array that is aggregated in the main thread (after joining with other threads) to produce a final count.

Implemented two new scheduling policies, viz., Exponential Distribution Scheduler and Linux-like Scheduler in XINU OS.

1) Exponential Distribution Scheduler
Scheduler chooses the next process based on a random value that follows the exponential distribution. When a rescheduling occurs, the scheduler generates a random number with the exponential distribution, and chooses a process with the lowest priority that is greater than the random number. If the random value is less than the… Show more

Implemented two new scheduling policies, viz., Exponential Distribution Scheduler and Linux-like Scheduler in XINU OS.

1) Exponential Distribution Scheduler
Scheduler chooses the next process based on a random value that follows the exponential distribution. When a rescheduling occurs, the scheduler generates a random number with the exponential distribution, and chooses a process with the lowest priority that is greater than the random number. If the random value is less than the lowest priority in the ready queue, the process with the lowest priority is chosen. If the random value is no less than the highest priority in the ready queue, the process with the largest priority is chosen. When there are processes having the same priority, they are scheduled in a round-robin way.

2) Linux-like Scheduler
Scheduler divides CPU time into continuous epochs. In each epoch, every existing process is allowed to execute up to a given time quantum, which specifies the maximum allowed time for a process within the current epoch. The time quantum for a process is computed and refreshed at the beginning of each epoch. If a process has used up its quantum during an epoch, it cannot be scheduled until another epoch starts. On the other hand, a process can be scheduled many times as long as it has not used up its time quantum. Show less

Implemented readers-writers lock system in XINU adopting a priority inheritance protocol to avoid priority inversion problem.
The priority inheritance protocol solves the problem of priority inversion by increasing the priority of the low priority process holding the lock to the priority of the high priority process waiting on the lock.

1. Developed a MapReduce application framework to detect online spam reviewers and by building & comparing two infrastructure models viz., Hadoop HA on EC2 (with EBS) and EMR (with S3) to handle huge volumes of input data

2. The project successfully detects spammers on different product reviews big datasets and also evaluates the performance of MapReduce on the built EC2 and EMR clusters.

1. SciBot is a Slackbot, a lighter version of the Scrum master. It intends to assists the product owner (or Admin) and the product dev team adhere to the Scrum framework and ease their work in accomplishing the goal

2. It assists software dev teams manage and maintain Scrum Agile; Supports interaction with Slack and GitHub Rest APIs

Developed a memory mapping scheme with support for demand paging, backing store management and page replacement (Aging and Second Chance)

The project focuses on developing an open source high performance Big Data Analytics Cloud infrastructure ​that will offer Analytics as a Service​ using OpenStack Platform 10.
Host OS - RHEL 7.3 with KVM + QEMU as type-2 hypervisor.
Base image for providing Platform as a Service is Ubuntu-16.04 LTS ​(Xenial Xerus) qcow2 image.
Two customized images - one for DBs (low speed but high storage) and another for analytics (high speed low storage).
The analytics image is mainly customized… Show more

The project focuses on developing an open source high performance Big Data Analytics Cloud infrastructure ​that will offer Analytics as a Service​ using OpenStack Platform 10.
Host OS - RHEL 7.3 with KVM + QEMU as type-2 hypervisor.
Base image for providing Platform as a Service is Ubuntu-16.04 LTS ​(Xenial Xerus) qcow2 image.
Two customized images - one for DBs (low speed but high storage) and another for analytics (high speed low storage).
The analytics image is mainly customized with: Nutch, Solr, Mortar and Portrait. The database image is configured with: Cassandra and PostgreSQL. Show less

Developed a movie recommendation application to recommend movies and genres to users based on their previous ratings by collaborative filtering using Apache Spark and kNN classification.

The project aims to develop algorithms to track systems buffer pool, implement FIFO buffer replacement policy and perform Undo-Redo recovery management.

Database application to assist people track and manage health status information using JDBC and Oracle DB integrated into a web GUI.

The project attempts to build a health management system to keep track of patient's health (both healthy and otherwise) with a collaborative effort from doctors and health supporters.

People can track their observations about certain health indicators and activities of daily living whose patterns can be reflective of their health status. Tracking… Show more

Database application to assist people track and manage health status information using JDBC and Oracle DB integrated into a web GUI.

The project attempts to build a health management system to keep track of patient's health (both healthy and otherwise) with a collaborative effort from doctors and health supporters.

People can track their observations about certain health indicators and activities of daily living whose patterns can be reflective of their health status. Tracking such information can help create alerts when something is wrong or when patients become slack about health management. Show less

A novel multi-level encryption scheme using enhanced RSA and MVC algorithms to make the cloud data safe from any kind of attacks is introduced.

The system is designed for a community of users where any user can upload a file and share a secret key among all the users, thereby resulting in a mutually trusted multi-owner file sharing framework.

Implemented using Java and Eucalyptus cloud.

An enhanced and secured RSA key generation scheme is proposed based on four prime numbers (instead of the conventional 2 prime number based RSA) along with a complex computation for the encryption key.

Real-time cryptanalysis was performed to prove the efficiency of the scheme.

Implemented using Java.

Developed an efficient search space reduction methodology for retrieving matching image from large scale face image database using improved and novel data mining algorithms.

This work was done as a part of the project proposal titled “Face Recognition System with video analytics” funded by Baba Atomic Research Centre-Board of Research in Nuclear Sciences (BARC-BRNS).

Done under the guidance of Dr. V. Vaidehi, Director, AU-KBC Research Centre, Madras Institute of Technology, Anna… Show more

Developed an efficient search space reduction methodology for retrieving matching image from large scale face image database using improved and novel data mining algorithms.

This work was done as a part of the project proposal titled “Face Recognition System with video analytics” funded by Baba Atomic Research Centre-Board of Research in Nuclear Sciences (BARC-BRNS).

Done under the guidance of Dr. V. Vaidehi, Director, AU-KBC Research Centre, Madras Institute of Technology, Anna University, Chennai.

Implemented using OpenCV and C++. Show less

Designed a three-tier framework for categorized and secured data storage.

Formulated a boosted clustering algorithm to ensure optimal information retrieval and high-level personalization.

Funded by Centre for Technology Development and Transfer (CTDT), Anna University, Chennai, India.

Done under the guidance of Dr. P. Varalakshmi, Assistant Professor (Sr. Gr.), Department of Computer Technology, Madras Institute of Technology, Anna University, Chennai.

Implemented… Show more

Designed a three-tier framework for categorized and secured data storage.

Formulated a boosted clustering algorithm to ensure optimal information retrieval and high-level personalization.

Funded by Centre for Technology Development and Transfer (CTDT), Anna University, Chennai, India.

Done under the guidance of Dr. P. Varalakshmi, Assistant Professor (Sr. Gr.), Department of Computer Technology, Madras Institute of Technology, Anna University, Chennai.

Implemented using Java. Show less

Developed a secure web application using HTML, J2EE, AJAX and Web services to store confidential information (like passwords and credentials).

In the back end, the data is stored in encrypted format and upon retrieval it is decrypted.