Major Qualifying Project Center A Term 2011–2012



The MIT Lincoln Laboratory continues a joint collaboration with Worcester Polytechnic Institute (WPI) to have a number of their undergraduate seniors perform their Major Qualifying Project (MQP) at the Laboratory in Lexington, Massachusetts.  MQPs are available for students studying Aerospace Engineering, Computer Science, Electrical and Computer Engineering, Mathematics, Mechanical Engineering, Physics, or Robotics Engineering.  Contact the IGSD or Professor Ted Clancy in the Electrical and Computer Engineering Department at WPI for details on how to apply.
Important Dates

  • September 21, 2010
    IGSD Global Fair
    4-6 PM (Campus Center, Odeum)
  • October 28, 2010, 5 PM
    Information Session:
         IGSD Information Session for MQP Centers including Lincoln Lab
    immediately followed by…
        Lincoln Lab-hosted Q&A with Pizza
  • November 3, 2010 by 1 PM – Applications due to WPI IGSD

Descriptions of Fall 2011

Recommended for Computer Science Majors:

Recommended for Electrical and Computer Engineering Majors:

Recommended for Math Majors:

Recommended for Mechanical or Aerospace Engineering Majors:

Recommended for Physics Majors:

Recommended for Robotics Majors:


Modeling Multi-Target Track Association and Sensor Fusion
Group 31 – Systems and Architectures (2–3 students)
Dr. Stephen D. Weiner

Recommended WPI Major(s): ECE, Math, Physics

The Systems and Architectures Group is concerned with modeling the performance of Ballistic Missile Defense (BMD) systems comprised of networks of sophisticated radar and optical sensors connected by an overall battle management control structure.  In such BMD systems, information collected by several sensors on a number of targets must be combined and used to make decisions regarding which targets should be intercepted and which targets should be rejected.  For this process to be successful, it is necessary that the data from “sensor 1” on “target A” be associated with data from “sensor 2” on “target A” and not with data from “sensor 2” on “target B”.  To date, most of the analyses of this problem have involved elaborate Monte Carlo simulations which provide high fidelity modeling of the target motion, target signatures and sensor measurement capability but provide limited flexibility to consider target and sensor variations.  They also provide limited insight into those threat and defense parameters, which have the greatest influence on overall performance.

The students will specify, design, and develop simple modular parametric models of the association and fusion process to permit rapid evaluation of the overall defense performance as a function of number of targets, density of targets, velocity spread of targets, defense sensor resolution and accuracy in range and angle, any sensor measurement biases and the relative geometry of the different sensors.  Major modules will include association of crossing targets for a single sensor, handover of single targets from one sensor to another, sensor bias estimation and removal, association of objects in a target complex seen by one sensor with these objects seen by a second sensor.  For each module, the output will be the probability of success and the probabilities of different failure modes as a function of the input parameters.  These individual modules will be combined into an overall functional model whose output will be the probability of correctly selecting the object to be intercepted (the warhead) as a function of all the input parameters.  This overall model will be used in higher-level simulations of total BMD system performance.  It would be nice if versions of the model could be developed as an Excel spreadsheet(s) or a MATLAB script(s).  The output of these simple models will be compared with the output from more elaborate Monte Carlo simulations.

Students for this project could be a combination of Physics, Mathematics, and/or Electrical and Computer Engineering majors. The skills needed for this project include a background in probability and error analysis (MA 2621 or MA 2631), understanding of coordinate systems and simple target motion (PH 1110 or PH 1111; PH 2201 or ES 2503 would be helpful), and understanding of the capabilities and limitations of computers to model these processes.  Knowledge of MATLAB is also required.

The most important skill is good judgment, to model all the important factors in the problem and ignore all the unimportant factors. There are a number of people in Group 31 who can assist in acquiring the background and developing the judgment needed to address this problem.


Integration of Cloud Computing with Distributed Middleware
Group 33 – Ranges & Test Beds (2-3 students)
Matthew Leahy

Recommended WPI Major: CS

Lincoln Laboratory's Ranges and Test Beds group supports several systems that utilize distributed middleware systems.  The systems are used to control missile range operations, multi-sensor fusion sites, and are part of open, net-centric systems.  Data in these systems are "write-once, read many" and stored data should be made available immediately after creation.  They typically include a message-oriented middleware and extensive logging and storage of transmitted data (archiving).  There are several potential issues related to the archiving functions: ensuring proper data marking, network latency, and simultaneous access.  The DoD Net-Centric Data Strategy states as one of its goals "handling information only once".  Traditional design has created storage sub-systems separate from the data distribution mechanisms.  Passing the same data to multiple systems injects points for corruption and ambiguous labeling of data.  Network latency necessitates large caches and creates issues of cache coherence and data replication.

Advances in distributed "cloud" computing systems, particularly related to large distributed file systems, can be adapted to work as part of a distributed middleware system.  Products such as GoogleFS and Amazon S3 have been proven effective on large open systems.  These systems were not designed to support the soft real-time constraints of the systems discussed above.  They were, however, designed for distributed operations and for concurrent reading.  Cloud storage systems are a prime candidate for replacing traditional network file systems (NFS, SAMBA) and block level (iSCSI) storage systems within distributed middleware systems.

This project requires 2-3 computer science students familiar with networks, network computing, distributed systems, and software engineering.  The first phase of the project will have students survey available software to support the integration of cloud storage and a distributed middleware system.  The second phase of the project will produce a working prototype/demo incorporating the cloud storage system identified during the survey.


Radar Automation in an Open System Sensor Architecture
Group 33 - Ranges & Test Beds (2-3 students)
Gregory Gimler

Recommended WPI Major: CS

Lincoln Laboratory's Ranges and Test Beds Group is developing the ballistic missile defense sensor infrastructure for the Pan-Pacific Range, associated mobile range assets, and field test beds.  So-called “open” architectures play an important role in the design of sensor systems by allowing them to be decomposed into hardware and software subsystems that can be independently developed, tested, and upgraded.  A Lincoln-developed system, the Radar Open Systems Architecture (ROSA), currently operates roughly 15 multi-mission radar sensors located in the Marshall Islands, Massachusetts, Hawaii, Florida, California, and the Caribbean.

Automation and operator situational awareness play a critical role in the success of these systems.  Students will work on an upgrade to ROSA, focused on enhanced automation using Python and 3D visualization for real-time radar control.  Students will gain valuable experience in real-time distributed control systems and radar. 

This project requires 2-3 students majoring in computer science, with strengths in the areas of operating systems, computer networks, distributed systems, and software engineering. The technologies most likely to be used during the prototyping phase of the project include Linux, C/C++, Python, and DDS publish-subscribe middleware. The desire and ability to work as an intern at Lincoln Laboratory on preparatory research during the summer prior to the MQP is advantageous.


Software Radar
Group 33 - Ranges & Test Beds (2-3 students)
Seth Hunter and Dr. John Nelson

Recommended WPI Major: CS, ECE

There is presently great interest in developing enabling technologies for Software Radar.  Software Radar is analogous to software radio where RF signals are directly sampled from the minimal receiver hardware and sampling, down conversion, signal and other processing is conducted using software techniques in special or general purpose computers.  Software Radar is based upon the same approaches but in many instances is operating at higher carrier frequencies.  A true Software Radar system would allow great flexibility with respect to use of and interpretation of agile waveform suites which could be tailored to particular needs.  It could eventually greatly expand the flexibility of radar system design and reduce costs for production of, maintenance of, and improvements to these systems.

Challenges for development of enabling technology for this area include high speed data acquisition, efficiently handling large quantities of digitized data before software down-conversion, high performance signal processing approaches, high throughput data storage, and distributed computing.  We would like to explore some of these challenges, determine limitations of present technology with respect to Radar carrier frequency based requirements, and prove the veracity of selected technologies and approaches for implementing key aspects of the Software Radar.

This project requires 2-3 students majoring in Computer Science or Electrical & Computer Engineering, with aptitude in the areas of parallel architecture, high throughput calculations, GPU processing, real time computing, traditional distributed systems, and general software engineering. The technologies most likely to be used during the prototyping phase of the project include Linux, C/C++, parallel computing, and specialized processing (eg. GPU and parallel processing environments). The desire and ability to work as an intern at Lincoln Laboratory on preparatory research during the summer prior to the MQP is highly desirable.


Indoor Marsupial UAV
Group 76 – Control Systems Engineering (2-3 students)
Michael Boulet and Byron Stanley
Recommended WPI Major(s): Robotics, Electrical and Computer Engineering

Navigating unstructured indoor environments can be difficult for ground vehicles.  In particular, the military may need to map out buildings with rubble, debris or other obstacles that make the ground difficult for ground vehicles to move over or survey.  Autonomous air vehicles can provide surveillance information while avoiding the difficulties posed by indoor terrain.  One of the primary limitations, however, for small air vehicles, is the time of flight.  As such, a combined ground and air platform system could provide a significant advantage over one or the other alone.
Students on this project will automate and demonstrate a UAV system that is capable of launching off of a packbot and avoiding obstacles within a room.  In particular, the focus will involve developing and successfully implementing robust algorithms and sensors so as to enable the UAV to autonomously land and take off from the packbot, avoid walls and obstacles, and provide feed back camera video to a ground station.  While the vehicle may be teleoperated for exploration, a secondary objective is to successfully autonomously explore a room.  A beacon and landing mount may be implemented on the packbot to assist with locating and landing.  A quad-rotor or other unmanned vehicle will be provided for use with this project.  The final demo scenario will include one or more of the three objectives: launch and land; obstacle avoidance; and/or room exploration. 

For this project we desire robotics engineering majors, perhaps complemented with ECE majors.  Robotics majors should have completed the Unified Robotics sequence through RBE 3002 (or, RBE 400X, if possible).  It would be helpful for one or more members of the team to have experience in wireless networking (CS 3516 or ECE 3308), control engineering (ES 3011) and/or digital signal processing (ECE 2312).

(Note: One robotics team will be created to work on one of the two projects listed: “Indoor Marsupial UAV” or “Realistic Outdoor Sensing.”  Project depends on funding priorities as well as the interests and skills of student applicants.)


Realistic Outdoor Sensing
Group 76 – Control Systems Engineering (2-3 students)
Michael Boulet and Byron Stanley

Recommended WPI Major(s): Robotics, Mechanical Engineering

The output quality of sensors frequently used on outdoor mobile robotic systems degrades with time due to accumulation of dust/dirt/mud on the sensor aperture and/or moving parts. Additionally, many of the sensors were designed for indoor, controlled-climate use and were re-purposed for outdoor environments.  Realistically fielding autonomous robotics will require sensors that can reliably operate outdoors for extended periods of time without human maintenance (e.g. lens wiping).

Students on this project will create a design that modifies a commercial off-the-shelf (COTS) sensor to operate outdoors for extended periods of time without significant performance degradation.  In particular, the design must be robust and generally applicable to any sensor with a viewing window or lens.  At least one of the sensors must provide planar range data (LIDAR) and at least one sensor must provide 2D images (camera).  The testing area, lidar, and camera would be provided ahead of time by MITLL.  It is important that the solution be generalized to most visual sensors, as a single custom solution is of limited value.

For a final demonstration, the newly modified sensor(s) will be mounted to a mobile platform (does not need to be unmanned or autonomous) alongside conventional / unimproved sensors. The sensor set is powered-on and set to collect data during an extended drive (8 hours) through conditions simulating environments in which a robot might operate. These conditions will include dust and mud splash.  Sensor output will be analyzed to asses the improvement relative to the standard robot sensors.

For this project, we desire robotics engineering majors, perhaps complemented with ME majors.  Robotics majors should have completed the Unified Robotics sequence through RBE 3002 (or, RBE 400X, if possible).  It would be helpful for one or more members of the team to have experience in mechanical mechanisms (ME 3310) and/or control engineering (ES 3011).

(Note: One robotics team will be created to work on one of the two projects listed: “Indoor Marsupial UAV” or “Realistic Outdoor Sensing.”  Project depends on funding priorities as well as the interests and skills of student applicants.)


Determining Initial Orbit of Satellites using Optical Sensors
Group 91 – Space Control Systems, (2-3 students)
Dr. Zoran Spasojević

Recommended WPI Major: Math, Physics

The Space Control Systems group at MITLL develops technologies to detect and track satellites and other objects in space using optical sensors. One of the primary objectives of observations is to determine the orbits of satellites orbiting the earth. Because optical sensors produce only direction vectors (range is not known), one is challenged with computing accurate orbits using as few measurements as possible.

Modern techniques for orbit determination proceed in two stages. The first stage assumes a two body system and produces initial orbits based on Kepler's laws. In the second stage, a full force equation is expanded in the Taylor's series about the initial orbit. Various estimation techniques are then used to truncate the Taylor's expansion to a numerically manageable form. Therefore, for high precision orbits, computing accurate initial orbits is critical.

There are three techniques using 3 direction vectors to determine the 6 orbital parameters
for initial orbits. These techniques are due to Laplace, Gauss, and more recently Escobal. Each technique has its strengths and deficiencies. One of the strengths of Gauss's technique is that it works better than the other two when the measurements are taken close together.

All three techniques fail, however, when the 3 direction vectors are coplanar, and this is exactly the situation we face because many satellites and sensors orbit the earth in the same plane. In practice the preference is to take closely spaced measurements with optical sensors and thus obtain initial orbits. A generalization of Gauss's technique is being developed that can handle 4 coplanar vectors and thus eliminate the degenerate condition that exists when using only 3 direction vectors. Students will be involved in formulating scenarios for testing the range of applicability of these methods for initial orbit determination using 4 coplanar direction vectors.

Students should have complementary backgrounds in physics and mathematics. Knowledge of orbital mechanics is desirable but not necessary. The students will implement all algorithms in either C, Java, or MATLAB.


Task Scheduling of Optical Sensors for Multiple Users
Group 91 – Space Control Systems, (2-3 students)
Dr. Zoran Spasojević

Recommended WPI Major: CS, Math

The Space Control Systems group at MITLL develops technologies to detect and track satellites and other objects in space using optical sensors. The Optical Processing at Lincoln (OPAL) software package is a key part of this effort. Mission planning algorithms for sensors are an essential part of successful data collection and are an integral part of the OPAL package.

We frequently use multiple sensors to obtain desired information about satellites, scheduled at different times for observations. With multiple satellites being observed, scheduling conflicts often occur. The best observation time for two different satellites may coincide, while a sensor can only devote its resources to one satellite. We would like to develop efficient scheduling algorithms to avoid such conflicts and effectively assign multiple satellites to multiple sensors.

Multiple users often have similar observation requirements for multiple sensors and multiple objects. Different users among the same user group will have different priorities. The tasks they submit to different sensors may also be ranked according to their importance. We would like to develop algorithms for effective scheduling of multiple users and their submitted tasks to multiple sensors taking into account user and task priorities and sensor properties.

Students will develop and implement multi-user scheduling algorithms. Students will also work on formulating the means to evaluate the effectiveness of such algorithms. This is usually done by defining an objective function which assigns numeric value to each scheduling opportunity. Students will develop their own algorithm evaluation criteria.

Students should have complementary backgrounds in mathematics and computer science. Mathematics is needed to develop the theoretical aspects of the algorithms, while computer science is needed to efficiently implement the algorithms in Java. Familiarity with solving assignment problems and linear programming techniques is a plus. For successful completion of the project, it will be highly advantageous to students to start the internship in June of 2011.


Searchable Archive For Data From Large-Scale Field Exercises
Group 101 – ISR Systems and Architectures
2-3 Students
John Collins

Recommended WPI Major: CS

Lincoln Laboratory’s Intelligence, Surveillance and Reconnaissance Systems And Architectures group focuses on analysis of large complex systems supporting America’s defense and intelligence communities.  Critical to this focus is collection of large volumes of data via participation in field exercises that demonstrate sensor technologies in realistic contexts.   Such efforts, which typically occur several times per year, generate large quantities of data [Gigabytes to Terabytes] that support a wide variety of analysis activities across multiple organizations within the defense research community. Different exercises are not organized within any single program, but instead meet the needs of different government sponsors.  As such, exercises vary widely in their goals, methods, and types of sensor data collected. Consequently, no common system exists for archiving, organizing, and disseminating these data sets.  There is a need for a centralized archive that enforces common conventions for data representation, correlates data products in time and space, supports searches for archived products, and organizes data sets for dissemination.

Students participating in this project will design and implement a system for archiving data from field exercises. Participants will work with subject matter experts to analyze requirements for representing data from the wide range of sensor types that are used in these exercises.  Data may include still imagery and video from airborne cameras, radar-based detection of moving targets, GPS-based tracking of vehicles or individuals, and time-stamped activity logs compiled by exercise participants. The archive should assure that all data files use well-defined and fully documented formats, following recognized standards whenever possible.  The key metadata for all data files (such as spatial and temporal coverage, sensor and platform parameters, etc.) should be maintained in a database to support searching and cross-referencing.  A particular challenge will be accommodating formats and data collection practices that vary widely across different field exercises.  The system should not only provide a means to represent data from past exercises, but should also be extensible to accommodate future exercise data that may vary in type, format, and organization from what has been handled in the past.  The implemented system will use a web-based interface that provides methods for searching for data products based on various criteria (spatial, temporal, etc.), and exporting data in a well organized format.

Suggested background includes CS 3431 (Databases I), CS 3733 (Software Engineering), and CS 4241 (WebWare).

Heterogeneous Computing Infrastructure for Image and Signal Processing
Group 102 – Embedded and High Performance Computing, (2 Students)
Hanh Kim and Dr. Huy T. Nguyen

Recommended WPI Major: ECE

The Embedded and High Performance Computing Group is involved in the design and development of advanced signal processor technology based on hybrid architectures that use a combination of programmable digital signal processors (DSPs), general purpose graphics processing units (GPGPU), field programmable gate arrays (FPGAs), and application-specific integrated circuits (ASICs). Systems with throughput on the order of 100s of billion operations per second (GOPS) are required to meet the needs of the next generation of applications.

The seamless incorporation of a high-performance coprocessor (GPGPU, FPGA, or ASIC) into a software application is highly desirable. Besides the apparent benefit of application acceleration, this capability will support the implementation of a target system in multiple stages using technologies of increasingly higher performance and risk factors (e.g., DSP, GPGPU, FPGA, and ASIC). This project is aimed at designing and developing an infrastructure with a high-level middleware library that will allow the use of high-performance IP cores to support heterogeneous, reconfigurable signal processors capable of delivering 100s GOPS of computation in the challenging form factors required for embedded processors found in radars, sonars, and communication systems. The use of FPGA and GPGPU as embedded processors within different host processor architectures will be explored.

A total of 2-3 electrical and computer engineering students with suitable backgrounds will work under the direction of Dr. Michael Vai and Dr. Huy Nguyen to design and implement the infrastructure. Performance, scalability, and flexibility of a prototype implementation will be demonstrated in a modern radar application. This will be an excellent project for engineering students wishing to work on the forefront of computer technology in the important field of reconfigurable, embedded processing.

Preferred candidates will have successfully completed coursework in discrete linear systems, (ECE 2312); assembly programming, (ECE 2801 or CS 2011); and computer architecture or digital embedded systems, (ECE 2801, ECE 3803, or ECE 4801). Preferred candidates will also have proficiency in FPGA design with VHDL (ECE 3810), familiarity with software/hardware drivers (e.g., PCIe, SRIO, etc.), proficiency in MATLAB, and an understanding of linear algebra, FIR filters, and the FFT.

Also desirable: programming experience in C and MATLAB; computer system interfacing and network experience.


Beacon Locator
Group 108 - Tactical Defense Systems (2-3 students)
Dr. Brie Howley, Scott Bailie, Lisa Basile

Recommended WPI Majors: ECE, Physics, Math

The Tactical Defense Systems group works on air defense issues, in particular, air vehicle survivability and vulnerability of United States Air Force (USAF) aircraft to weapons systems.  For air defense, as well as many other applications, it is desirable to be able to locate a target emitting some type of beacon signal.  The target may be cooperative, in the case of an aircraft carrying an RF beacon or it may be non-cooperative in the case of an insurgent who is emitting a signal with his cell phone.  A network of sensors can receive the target signal, and compute its position using interferometry or time difference of arrival (TDOA) techniques. 

The students working on this project should design and prototype a direction finding (DF) system for determining angle of arrival of a radar beacon signal.  Students will need to conceptualize the system, adequately model the system to determine appropriate system requirements, design and fabricate a prototype, and demonstrate performance capabilities.  Real-time algorithms should be developed to discriminate the beacon’s signal and perform angle estimations.  The team should build a display to give system operators situational awareness about the location of the target.  The system should be extendable to support DF of multiple discrete targets.

For this project, students could be a combination of ECE, CS, physics and math majors.  Knowledge of RF systems, digital signal processing, coordinate systems, and real-time software is desired.  All team members need not have experience in all of these areas.  Rather, we are looking for an interdisciplinary team whose members have the necessary skills to work on this project collaboratively.       


The work described in this web page was and will be performed at Lincoln Laboratory, a center for research and development operated by MIT. The opinions, interpretations, conclusions, and recommendations expressed in this web page are those of the authors and not necessarily endorsed by MIT, the U.S. Air Force, or the United States Government.
Employment at MIT Lincoln Laboratory and/or participation in these projects is restricted to U.S. citizens.
The work described in this web page is sponsored by the Department of the Air Force under Air Force Contract FA8721-05-C-0002.


top of page