Project 1: Measuring the effects of daytime activity levels on sleep quality (led by Dr. Metsis)
The adverse effects of sleep deprivation, caused by various sleep disorders, are widely acknowledged by the research community. Traditional medicine focuses on diagnosing and treating such disorders via sleep studies. However, daily activities that affect sleep quality are often hard to monitor and take into consideration during diagnosis. There have been some preliminary efforts that looked at the effects of daily urban lifestyle and activities on sleep quality. However, such efforts are limited as daytime activities cannot be easily monitored outside the lab, and they mainly rely on self-reports. Also, current sleep monitoring standards do not allow for a long-term sleep monitoring, that will quantify the effects of lifestyle changes on sleep quality. Metsis has a track record of developing non-invasive sleep monitoring technologies, that can be used at home. In this project, we will develop methods for the collection and computational analysis of human activity data, to assess the effects of daytime activity levels on sleep quality. The collected data will be analyzed for correlations between daily activity levels and sleep quality, using machine learning and standard sleep quality indexes.
Project Plan: In this project, students will:
- Develop a Java desktop application which extracts the activity data collected by a smart-watch over a period of time (using the available API of the device), and provides some basic visualizations.
- Develop machine learning-based tools to analyze the data by segmenting the continuous collected signals into separate activities, and classifying those activities into a predefined set of classes (e.g. walking, running, sitting, driving, sleeping, etc.), and intensity levels.
- Utilize existing tools developed by the PI and develop new sleep analysis tools to determine sleep quality from sensor data. The last item will build upon the Summer 2015 REU project, in which Dr. Metsis mentored an REU student to develop sleep analysis methods from Polysomnography data.
Project 2: Modeling Tools and Risk Assessment for Smart Infrastructure (led by Dr. Guirguis)
Smart infrastructure will be more connected than ever with various components sensing and sending critical information to decision making entities. For example, smart meters in a neighborhood can sense various power loads and send this information to power generation and distribution components to adapt how power is generated, distributed and consumed. While such connectivity offers efficient methods to operate these systems and the convenience to manage them, they do open doors for cyber attacks to be mounted. Cyber attacks on such infrastructures continue to be one of the major threats that face the vision of smart cities. In one study, it was shown that two weeks of power loss in the Los Angeles County can cause an economic loss of $20.5B.
The goal of this project is to develop modeling tools to assess the risk of cyber attacks on smart infrastructure which rely on networking components and communication channels that are typically subjected to interference and cyber attacks. The operation of a smart infrastructure can be modeled as a state transition diagram.
Generally speaking, states can be classified into three sets: (1) safe states, (2) critical states and (3) catastrophic states. Safe states are ones in which the degree of uncertainty, coupled by the information received and decisions made would not lead to being in a catastrophic state. Critical states are those in which the potential to transition into a catastrophic state is higher and the system should recover back to a safe state. Catastrophic states are the ones in which an attack is imminent. There are few challenges that need to be tackled for the model above to be applicable for various smart infrastructures.
The first challenge is to understand the basic operations in these applications and map them to the model above without oversimplification. This would likely lead to a huge number of states and actions. The second challenge deals with the curse in dimensionality in which there are exponential number of states and actions. We will look into approximation algorithms. Finally, the third challenge deals with the issue of time-scales; there is a wide range of timescale in which information arrives and decisions are made. The model above must be modified to deal with a wide range of time-scales since it operates in steps. This can be done by mapping some decisions/information over a certain number of steps rather than over one step.
Project Plan: In this project, students will:
- study the operation of some case studies (e.g., smart power meters) in terms of the information available and the decisions made. Develop risk assessment models for those systems by mapping their operations to the model above,
- develop plausible adversary models, given the attacker’s knowledge about the operation of the smart infrastructure and the power he/she has (e.g., can manipulate information and/or decisions), and
- assess the risk involved and advocate policies to ensure the correct and safe operation of the smart infrastructure.
Project 3: Multi-UAV System for Accomplishing Collaborative Aerial Missions (led by Dr. Gu)
Unmanned aerial vehicles (UAVs or drones) are becoming more accessible to consumers due to the fast advancement of mobile and sensor technology and the dramatic drop in cost. Consumer UAVs are prominent mobile computing platforms to accomplish missions in a wide range of use cases, such as farm and livestock monitoring, infrastructure inspection, community security patrol, emergency response, filmmaking and so on.
Many use cases need to deploy multiple UAVs to accomplish aerial missions collaboratively. Such multi-UAV operations can increase area coverage and reduce response time greatly. But, the current practice is mostly single-UAV oriented. A mission is planned as a set of waypoints and paths. Then, the plan is divided and assigned to individual UAVs. The single-UAV practices obviously cannot handle multi-UAV missions for several reasons. First, multiple UAVs will have unbalanced air times over different flight paths and their battery capacities. Second, the safety of multi-UAV operations requires more attention due to higher chances of UAVs’ in-air failures and collisions. Third, the multi-UAV system is more susceptible to attacks that can sabotage communication and control among UAVs and ground stations.
This research will focus on developing a multi-UAV system that can coordinate multiple UAVs in a collaborative aerial mission to maximize the aerial time of UAVs and ensure safety and security in aerial operations. We will investigate battery consumption models on various UAV’s maneuvers and flight conditions. We will develop multi-UAV mission planning algorithms to balance UAVs’ aerial time with regards to their battery capacities. We will study UAV collision avoidance mechanisms to improve safety on multi-UAV operations. We will examine multi-UAV communication protocols and identify and remedy security threats to ensure the security of multi-UAV missions. The tools developed in this research will be added to the UAV communities (such as Ardupilot and Paparazzi UAV) to advance the status quo of consumer UAV applications from single UAV to multiple UAVs.
Project Plan: In this project, students will:
- Develop multi-UAV simulation components in the SITL tool,
- Develop aerial time and battery consumption models based on real UAV data,
- Explore safety and security issues in several UAV communication and control protocols,
- Develop multi-UAV mission planning algorithms and tools.
Project 4: Vision-based Automated Vehicle Activity Alert System (led by Dr. Tesic)
In the data-rich operational world of smart connected cities, it is increasingly difficult to sift through the vast amount of information coming from the wide array of camera sensors feeds. For example, emergency responders will be able to pick up a live feed of CCTVs, cameras mounted on unmanned aerial systems (UAS) or cameras in the vicinity of the event. Typically, cameras are used as “eyes” of first responders AFTER they have been alerted of the incident (e.g. traffic accident, fire, theft, unusual activity).
In this project, we will focus on developing tools for automated activity alerts from existing camera feeds. The idea is to automate the alert system, and in conjunction with other available contextual information (e.g. location, information obtained through 911 or 311, web activity), use the existing camera system as an “eye” WHEN the event happens, and thus increase the efficiency of first responders. The problem: camera sensors are cheap, but transforming the incoming video data streams from surveillance cameras around the city into actionable items still demands expensive human processing. For object recognition in overhead low-resolution video datasets, the best algorithms struggle with objects that are small (distant car) or with the distorted view of the parking lot from the UAV (sun glare), where humans have no issues in recognizing cars in such videos. However, recent advances in deep neural networks allow us to design and train algorithms to reliably detect objects from overhead cameras.
The goal of the project is to detect unusual or unexpected activity in the camera feeds and alert the end user to review the data and take appropriate action. In this work, the student will address fundamental questions regarding user experience and automated alerts, and design a metric of alert usefulness, a balance of too many false alerts and too many missed alerts for the end user.
Project Plan: The students will gather data from overhead city imagery e.g. VIRAT, NEOVISION, and Car Parking Lot Dataset captured by drone cameras to start. The students will:
- adapt open source deep learning network and user interface to detect and display detection of these moving objects (people, cars) in the video datasets;
- apply machine learning approaches to detect activities and create alerts if unusual activity is detected; and
- connect contextual data (location, camera info, user feedback) to improve the system. The goal of this project is to demonstrate the usability of state-of-art deep learning for smart city alert generation.
Project 5: Content Synchronization in Device-to-Device Communication in Smart Cities (led by Dr. Chen)
In this project, we investigate the problem of content synchronization between devices in S&CC. Consider a set of vehicles moving on the streets to form an intermittently-connected network. Each vehicle is equipped with wireless devices which are capable of short-range communication [3, 24]. Each device may store a number of files according to user interest: like city maps, news from a website, music, video clips and so on. Assume a file is copied to a set of mobile devices and it is updated over time. When two devices move into their communication range, they will mutually synchronize their files to the latest versions. When a file is updated in one vehicle, it is challenging to synchronize the content of the file in a set of vehicles in an intermittently connected environment.
Mobility is one of the most important characteristics in device-to-device communication. Various mobility models have been proposed in the past, such as the Random Walk model, Random Waypoint Model, and group mobility models. To mimic the vehicle movement in a city street network, we adopt the City Section mobility model.
In this project, we aim to answer the following questions: Consider an intermittently-connected network where node mobility is restricted to city streets and information exchange relies on device-to-device communications. When the content of a node is updated, how can the common content in other nodes be synchronized to the newest version? How long does it take for the synchronization process to finish? We will design synchronization strategies and find out the expected synchronization delay of these strategies. Though analytical results exist for other mobility models such as Random Walk and Random Waypoint, little attention has been paid to analyze the delay of content synchronization based on the City Section mobility model.
Project Plan: Our plan for this project includes the following tasks:
- In the first two weeks, the students will learn the concepts of device-to-device communication, content synchronization, the City Section mobility model and do an analysis of the properties of the City Section mobility model.
- In the next three weeks, the students will work on designing content synchronization strategies in device-to-device communication in a smart city environment and conduct a theoretical performance analysis of the proposed strategies.
- In the remaining weeks, the students will conduct simulations to test the performance of the synchronization strategies and compare the simulation results with the theoretical ones.