Activity Detection Using RF & Sensor Data Fusion

Sponsor: Bosch | Fall 2023

This project aims to leverage RF sensor and BLE technologies to enhance the detection and monitoring of a wide range of human activities, from basic movements to complex actions including yoga poses, utilizing ML and DL techniques.

The focus includes data acquisition from sensor nodes, and development of tailored ML models for activity detection, while embedded programming on provided hardware is not in scope. 

Current ML Models for activity detection do not cater to a wide range of activities and do not emphasize on physical wellness for elderly people that this project targets.

Control of a Passively Steered Rover

Sponsor: CMU Robotics Institute | Fall 2023

The project focuses on developing the Zoe 2 rover, a compact and technologically advanced successor to Zoe 1, equipped with a 3D kinematic model-based controller. The primary use case is for the rover to navigate precisely on challenging terrains such as deserts and other rough landscapes on Earth. The aim is to enhance the rover's effectiveness in terrestrial exploration, showcasing its capabilities in real-world scenarios.

Monitor Creative Cloud Express with AIOps

Sponsor: Adobe | Fall 2022

The team built a Machine Learning system, based on incoming metrics, to help site reliability engineers (SREs) analyze the large amounts of data and provide monitoring and anomaly detection. Given the incoming metrics from Prometheus, the team decided the right statistics for the metrics in the system and created corresponding alerts for these statistics.

Smartphone-based Peer-to-peer Blockchain-enabled Compute & Workload Sharing

Sponsor: Honda Development & Manufacturing of America | Fall 2022

Today, smart cars rely on cloud based services for computational needs such as predicting efficient routes and preventing accidents. Can we offload computation tasks to idle cars? In the team's project, they designed a system where idle cars may perform computation tasks requested  by other cars and earn cryptocurrency in return. 

Security Analysis of Consumer IoT System 

Sponsor: Carnegie Mellon University | Fall 2021

With the rapid increase of IoT devices, the attack surface of in-home networks has grown significantly, imposing a security risk since these devices have become a primary target for cybercriminals. Many different types of attacks could be avoided with an anomaly detection system that aims to reduce the risk of IoT device attacks. The team developed the foundation for an anomaly detection system that specifically performs data collection and device type identification in the network. To achieve this, different tools and methods were used to complete each task including Scapy (powerful packet manipulation program), nProbe (extensible NetFlow probe/collector), and Machine Learning (ML) algorithms. Training of the ML models using a decision tree classifier and random forest classifier resulted in an accuracy of 97% and 98% respectively. These results contribute to the foundation of a more complex IoT anomaly detection system that will apply an ML detection model for each identified IoT device.

BLOSEM: Blockchain for Optimized Security and Energy Management 

Sponsor: SLAC | Fall 2021

Establishing end-to-end security of critical grid infrastructure such as grid assets is one of the most challenging areas in energy infrastructure management. Compromised assets installed in the energy grid can expose the entire energy delivery system to cyber threats and risk the collapse of critical national infrastructure. This project involves working with SLAC National Accelerator Labs to enhance grid resilience against cyber attacks by developing a blockchain-based distributed, decentralized, and privacy-preserving supply-chain management system. The system is used to verify grid asset integrity through the supply chain and identify if the assets were tampered with during transit. Particularly, the system is designed to protect against clone attacks and false data injection attacks on grid assets.

Internet-of-Things Data Fusion

Sponsor: National Security Agency | Fall 2020

As their proliferation increases, Internet-of-Things (IoT) devices continue to generate more data and new types of information. Such devices are increasingly used by personnel in military installations, creating digital dust that presents a growing operational security concern for National Security Agency and Department of Defense organizations. This project team seeks to better understand how different IoT datasets could interact with each other and provide intelligence to adversaries. 

Serverless SIEM/Log Management Solution 

Sponsor: Procter & Gamble | Fall 2020

As a multinational company, P&G has Security Operations Center (SOC) teams that rely on SIEM (Security information and event management) systems to collect and aggregate event logs from across the enterprise to analyze and respond to security incidents. However, SIEM solutions have been expensive and difficult to scale. Therefore, this project team seeks to design and implement a scalable SIEM based on the existing cloud service infrastructure.