Projects
| INITIATE | |
| It is seen that there is a low number of high school students who pursue Science, Mathematics and Technology in college. They feel that mathematics and science being taught to them do not have real-life applications. This results in less students focusing on the STEM areas and pursuing a profession in them. INITIATE aims at involving more and more students towards the STEM+C (Science Technology Engineering and Mathematics + Computer Science) area by using smart cars for demonstrating the various mathematics and scientific problems using smart cars. A smart car is capable of autonomous functioning. We propose to exploit a vulnerability of such a smart car and control it using an android device and use it to demonstrate the daily life applications of the concepts of Mathematics, Science, and Technology. | |
| Digital forensics and cyber-attack evaluation & prevention on real-world drones. | |
| The basic idea of the project is to exploit various vulnerabilities in an unmanned aerial vehicle and device methods to rectify them. Currently, we are exploring the capabilities of commercially available off-the-shelf drones to investigate the drone(s) that can be attacked and the team is implementing various ways to prevent an attacker from attacking the drone or collect information from the drone. As there are a number of ways in which the attack can happen on the drone, the project aims to identify these vulnerabilities. Drone forensics, on the other hand, is analyzing the hardware and software of the drone to extract information for evidence, usually after a crime. Based on deep analysis of each part of the drone, like source, purpose, and aim of the drone, it is expected that the team would be able to learn the previous locations, height, temperature, etc., where the drone was operating last. It also includes learning new technologies which can be used in flying the drones and meeting its purpose of flight | |
| Android application security. | |
| In the last few decades, smartphones have become an integral part of our everyday lives. Having the ability to handle many useful and attractive applications, smartphones sport flawless functionality and small sizes leading to their exponential growth. Additionally, due to the huge user base and a wide range of functionalities, these mobile platforms have become a target for cyber- attacks. Android being open source, securing it requires robust architecture and rigorous security programs. Android allows installation of third-party APKs, which can be potentially harmful to the device, its user or user data. Potentially harmful applications attacks can range from data collection for targeted advertisement to potentially driven attacks for users’ harm. Android requires additional security enforcements even if it has the presence of security features like user permissions, security at the kernel level, and inter-process communication security and hence application security comes in picture. Analysis at the application level can be of two types static and dynamic. The aim of this research is to develop a tool or security framework which will prevent the possibility of attack through the application. | |
| Wide body area network (WBAN) security and decision making device. | |
| Consequences occurring because of late diagnosis are disastrous. To avoid or at least mitigate these consequences we are attempting to design this system. Figure 4 shows a rough architecture of the overall idea. Several applications are possible for such a network security module. First, a wearable device which will monitor customized vital parameters according to the patient’s need. The other application might be for hospitals to reduce human effort and delay in treatments. There can be a hub for all sensors which could be secured by this security module. This device will comprise various important bio-sensors such as pulse rate sensor, blood pressure sensor, skin temperature sensor and galvanic skin response (GSR) sensor, etc. The data from these sensors will then be analyzed by the processing chip. The processor will have predefined vital parameters of human vital signs that can be used to cross reference while the process of decision making. After analyzing, if the data is found to be critically high or low, an alert will be sent to the emergency services by the device itself with the use of cellular network technology. If the reading values are high, but not critical, the data can then be transferred to the server that collects data from all these sensors periodically. Whenever the patient visits the hospital for a checkup, the RFID tag on the device can be scanned, and a report file can be automatically generated. This can help the doctor examine the changes in or recovery of the concerned patient. The data from these machines can be communicated through wired connections to a central hub, which will be a high processing unit, where the data will be analyzed in real-time. It will have wireless connectivity all over the hospital. If ever the readings go over the charts, the physician will get an alert along with the patient’s vital stats. This can help in preparation and save a lot of resources and time because time is very important in emergency situations. | |
| Distributed machine learning based real-time intrusion detection system. | |
| Distributed denial of service (DDoS) attacks make an organizational network services unavailable by continuously overwhelming its servers with unsolicited traffic. Inexpensive Internet subscriptions and easily accessible attack tools led a vast increase in volume, size, and sophistication of these attacks in the past few years, as noted by the most recent attack on Twitter and other popular websites. According to the prediction of Cisco Visual Networking Index (VNI), DDoS incidents will reach up to 17 million in 2020, a threefold increment compared to 2015. Therefore, detection and mitigation of these attacks in real-time become a prime consideration for any organizations.The team is working on the design and development of an IDS that could run in near real-time to address the issues mentioned earlier. Figure 5 shows this architecture. The importance of feature extraction in the development of a real-time IDS serves as the primary motivation of this work. To develop a system for efficient feature extraction, we used Netmap and Apache Spark. Both tools are open-source. Netmap is capable of capturing packets at line-rate in contrast to the previously mentioned tools, hence, will help in reducing the false-positives. Apache Spark is a distributed computing framework that processes data more efficiently compared to Hadoop and will make our system faster for feature extraction. We used our implemented system for feature extraction of TCP-based network traffic and evaluated the results for attacks modeled on CAIDA attack dataset. | |
| Human machine training (HMT) in training of medical first responders. | |
| The aim of the project is to develop a Synthetic Assistant (SA) that uses computer generated audio to assist first responders in treating patients and helping in the Transfer of Care (TOC) process. The SA is implemented in Human Machine Teaming (HMT) architectural process in a Live Virtual Constructive (LVC) environment which is used for training first responders. This is expected to support the emergency care-taker by undertaking various computing intensive operations, enabling coalescing of man and machine strengths. This SA will assist each caretaker trainee individually and as a group.This is an OFRN funded project. Recent Demo | |
| Emotion detection in Augumented Reality(AR) environment using hologens. | |
| HoloLens is based on the term mixed reality. That means it blends the objects made in 3D in our physical world. Wearing the HoloLens one can interact with the holograms, along with preserving the peripheral vision with the help of see through lenses, the holograms travel with the one using it in the holographic frame. HoloLens can be used to interpret emotional states exhibited by the individuals (not wearing the HoloLens) within a wearer’s field of view and also can give feedbacks. This can be done through the eye tracking engine, image and audio processing engine. The research is on how the emotions (smile, frown, anger, murmurs) can be captured. Using the captured emotions, we can interpret what the other individual is thinking. This has different scenarios like when one goes to a wedding; this will be used to find out people we know, and while in the scenario of a professor it will be used to tell the environment of the class, like which student is interested or which is not. The feedback will be given from the database of stored interactions. This feedback will help the wearer to change how he will approach people so as to look more interesting. Till now we have created a small working model using Unity, Visual Studio and hologram emulator. Figure 8 (a) shows this environment. We have added gaze and a gesture to it to make it look real. Figure 8 (b) demonstrates how the user would respond when the system detects that the user is in a happy mood. | |
| Cyber-Security in web development and OWASP based projects. | |
| The Open Web Application Security Project is an online community which creates freely-available articles, methodologies, documentation, tools, and technologies in the field of web application security.The aim of the research is to develop projects based on Open Web Application and security projects. | |
| Deep learning based security model for Intelligent Channel Sensing Network. | |
| Increased use of space communication has introduced new requirements for spectrum sensing, dynamic spectrum allocation and intelligent networking for a network to provide resilient space communication. The research study materials1 focused on an architecture for an Intelligent Channel Sensing Cognitive networking with security as one of the influential parts. In brief, the architecture shows the formation of the cognitive network which enables space-automation. Starting with adaptation capabilities in a changing communication environment, it also performs channel sensing, dynamic spectrum allocation, signal & interference detection. On top of these functions, the network is made secure from attacks & threats caused by the inherent properties of the Intelligent Cognitive Network. As the Cognitive Network is capable of communicating in a wide spectrum of bands, it is vulnerable to attacks which might degrade the performance of the network and the learning itself. Software security and Hardware security are the two parts addressed by the security portion of the network architecture. The software security is again a two-fold implementation, namely, (i) Security against any attacks or malicious activity affecting network functionality and (ii) Preventing adversarial learning i.e. affecting the brain of the network and manipulate its learning. This is a funded research by the OFRN (Ohio Federal Research Network), $200K subcontract. Total project award is $800K including university/industry partners. | |
| The Organ Failure Assessment Module (OFAM) using contextual anomaly detection (CAnD). | |
| Developing The Organ Failure Assessment Module (OFAM) which implement a novel semi-supervised contextual anomaly detection (CAnD) approach to predict organ failure by identifying sequences that may not appear anomalous on their own, but are anomalous when taken in the context of what has happened around that sequence. | |
| Human computer interaction with multi-modal robotic arm. | |
| Developing a multi-modal robotic arm whose inputs are voice and gesture. The idea is to make a system more efficient and error free by increasing the accuracy through fusion of several modalities. |