Asteroid Tracker

The Asteroid Tracker challenge tasks competitors to develop optimization algorithms for Near Earth Object tracking using monolithic radar arrays. The winning solutions must first efficiently determine the necessary subarray allocation required to track all objects given a particular data set, then eventually determine the optimal configuration of the array, as well as how this optimal configuration will change if additional dishes are added. Asteroid Tracker (KaBOOM) Github repository  

Project Background

The Asteroid Tracker Challenge is part of NASA’s Asteroid Grand Challenge Series, and in this particular challenge, we are tasking competitors to optimize the use of an array of radar dishes when tracking Near Earth Objects from the time they become visible over the horizon till the point at which they cease to be visible. This tracking allows scientists to gather information from each object such as composition, spin rate, among other properties.


Near Earth Object(NEO) detection and characterization is a critical need for NASA and the United States. NASA has been directed to develop capabilities to observe, track and characterized NEOs and other deep space objects that could pose a threat to the Earth. As a result, NASA is developing concepts for a highly capable deep space radar array consisting of sets of commercially available monolithic antennas.


One of the challenges faced by NASA is determining the optimum selection of individual antennas within the array for a given track observation. This is a complex analysis and goes directly to development of the concept of operations and cost of operations (in terms of maintenance and total capacity required).


High Level Requirements

This challenge, in Phase 1 will devise the optimum subarray selection per set of object tracks provided in a predefined configuration using a fixed number of dishes . Phase 2 will build on Phase 1 results but will focus optimal optimizing configuration and looking at placement of additional dishes.


The following requirements must be met:

  • Be able to model phased array antenna beams using a predetermined set of dish and beam properties.
  • Take, as input, trajectories of a number of NEO and for each, be able to provide the optimal selection of subarrays to track the object for its entire visible path (or, if defined, a minimum time period of observation that gives sufficient scientific observation value).
  • Be able to read properties from configuration files – i.e. dish properties, array configuration, trajectory data, etc.

Atomized Project Plan

So what’s coming up, and how do I register? Simply select any of the contest names below to find out more about each specific contest, and register to participate.

Contest Start End Contest Type
KaBOOM API Documentation 04/30/2014 05/02/2014 First2Finish
KaBOOM Marathon Match Visualizer 05/08/2014 05/15/2014 First2Finish
Asteroid Tracker MM 07/29/2014 09/02/2014 Marathon Match
!!URGENT!! KaBOOM Asteroid Tracker Mac OS X Deploy Fix 09/04/2014 09/05/2014 First2Finish
!!URGENT!! KaBOOM Asteroid Tracker Windows Deploy Fix 09/04/2014 09/05/2014 First2Finish
Asteroid Tracker – Phase 1 Project Delivery 08/30/2014 08/30/2014
Asteroid Tracker – Phase 2 – Architecture for Code bundling TBD Architecture
Asteroid Tracker – Phase 2 – Code bundling into PC Executables TBD Assembly
Asteroid Tracker – Phase 2 – Scoring Function Update TBD Assembly
Asteroid Tracker – Phase 2 – Invitational Testing TBD Testing
Asteroid Tracker – Phase 2 – Marathon Match TBD Marathon Match
Asteroid Tracker – Phase 2 Project Delivery TBD


*PLEASE NOTE: The contest start and end dates are estimates only, and may shift forward or backwards depending on contest progress. To verify a contest start date, please select the contest name to view the registration page.




This challenge requires an understanding of antenna theory, radar and electromagnetics. The level of understanding is NOT greater than that which could be learned through web resources and/or undergraduate physics / engineering references. Some general references are provided, as well as specific examples and references applicable to NASA’s use cases:

Antenna Theory


Amateur Radio / Layman Resource


MIT Open Course on Electromagnetics


MIT Open Course on Receivers, Antennas and Signals


Institution of Engineering and Technology


Array Radars

MIT / Lincoln Laboratory

Fenn, et. al., The Development of Phased Array Radar Technology


MIT / Lincoln Laboratory

Herd, Phased Array Radar Basics


NASA Application (Deep Space Radar Measurement and Uplink Array)

Vilnrotter et. al., Planetary Radar Imaging With the Deep Space Network’s 34 meter Uplink Array


Davarian, Uplink Arrays for the Deep Space Network


D’Addario, Uplink Array Demonstration with Ground-Based Calibration


Background Definitions:


Monolithic Antenna (or just Antenna)

A single, standalone antenna element.

In the case of the NEO Array each monolithic antenna consists of a 12-meter Cassegrain parabolic reflector antenna, antenna pedestal (structure) and the electronics and motors necessary to point, steer and transmit/receive signals. For the NEO Array, each monolithic antenna is essentially a standalone “ground station”. Each physical (real world) antenna has a specific antenna pattern “fingerprint”. These patterns must be measured to determine the actual pattern and how it differs from theoretical.


Antenna Array (or just Array)

A combination of multiple monolithic antennas into an integrated system. The NEO array will use phased array signal processing techniques to actively combine the transmit and receive signals to achieve much greater performance (power, range, sensitivity, etc) than could be achieved by monolithic antennas individually. The “Array” in the context of the NEO array refers to the complete set of monolithic antennas and the infrastructure required to support and operate them.


Antenna Subarray (or just Subarray)

This refers to a specific combination of monolithic antennas used to support a given tracking event. A “subarray” can consist of any number of monolithic antennas from one to the total number of antennas in the entire array system. A subarray is used to synthesize a RF beam for a specific tracking event.


Antenna Beam

This refers to the path along which most of the desired energy (and sensitivity) of an antenna is focused. This is also referred to as the “main lobe” of the antenna pattern.


Array (or Subarray) Beam

This refers to the synthesized antenna beam resulting from the combination (through constructive interference and superposition of RF energy) resulting from the simultaneous operation of properly phased monolithic antennas that make up the subarray.

REF (Antenna Array Analysis in MATLAB):

REF (Subarray Analysis in MATLAB):

Side Lobes

This refers to directivity that is not in the direction of the main antenna beam. These are typically undesirable. Side lobes are a result of the interference (constructive and destructive) and superposition of RF energy as it is influenced by the design of the antenna. All antennas (other than the theoretical isotropic) have sidelobes which can be predicted mathematically. Real-world antennas generally follow the theoretical predictions, but will have additional effects introduced by imperfections in the physical antenna.


Antenna Gain

The “gain” is the amount of “increase” or “amplification” that a signal receives as a result of the physics of the antenna. An antenna with a gain of 1 (or 0 dB) is referred to as “isotropic” or “omnidirectional” and is used as a reference level. All real antennas have some gain which is reflected by the directivity of the RF energy transmitted in the beam or sensitivity of the received signal. This is a zero-sum process as power “added” to one direction must be made up by power being “removed” from another.


The parabolic reflectors used in the NEO array are highly directional and designed to produce very focused beams in the direction of the dish axis, with very little power being radiated (or sensitivity provided) off-axis. All real-world antennas have imperfections, however, and so some power and sensitivity is available “off axis”. These are referred to as the “side lobes” of the antenna pattern. It is a goal to minimize the effects of the side lobes.



The process by which the amplitude and phase of signals transmitted / received by individual monolithic antennas comprising the subarray are combined (through signal processing) to yield a synthesized beam resulting from the subarray.


Array Gain

The maximum directivity (or sensitivity in the receive path) of the synthesized beam resulting from the combination of antennas in the subarray. In an ideally combined system, the resulting transmit power in the synthesized beam is increased by a factor of N^2 where N is the number of antennas in the subarray.


Azimuth and Elevation

Measures used in a spherical coordinate system to identify a direction in celestial space.



The object of interest for the Radar observation. This could be a satellite / spacecraft or a natural object in space such as a planet or asteroid.



The path traced by a target object across the sky.


Orbital Elements

Used to specify the orbital characteristics of a spacecraft / space target. Along with


Two Line Element set

“Shorthand” notation specifying a space object’s characteristics and orbit. Used to plan for radar observations, compute target tracks, and plan / observe object trajectories.


Key Concepts:


Array Radar

The purpose of the array radar is to detect, track, measure and characterize targets such as spacecraft, asteroids and planets from LEO to deep space. Radar works by measuring the characteristics of the RF energy reflected back off of a distant target. An RF signal is transmitted, it travels to the target (losing power and experiencing distortions). The signal is reflected by the target (with some changes to the signal resulting from material characteristics of the target). The signal then travels back to the receiver where is it captured and analyzed. Round trip time-of-flight tells you the range. Doppler shifts inform you of motion and rotations. Polarization and other electrodynamic changes to the RF energy tell you something about the material properties. Finally, precise measurement of the RF signal can provide images of the target object (eg. pictures of an asteroid).


Radar is an N^4 problem. This means that the power received by the radar receiver will be approximately (but not greater than) the 4th root of the power transmitted. This is because one-way RF transmission is an R^2 problem. As a RF beam propagates through space it spreads over an increasingly large surface (think of the surface of an expanding balloon… as the radius increases so does the surface covered by the same beam angle, and so less power per unit area is available). Once the signal reaches the target and is reflected back the process must begin again, only now the “transmitting” power is only as much as was available at the target… And so the total power loss for the round trip goes as 1/N^4… 1/N^2 out and 1/N^2 back. This makes sensing targets very far away extremely difficult.


Using a phased array for radar (especially a phased array of individually very powerful transmitters) significantly improves this. As the main beam of the phased array combines the power from each element antenna the power is added together at the target. Since its actually the electromagnetic field strength (measured in volts/m^2) that is being combined, and power goes as the square of voltage, we achieve an increase of N^2 in the beam.


To achieve this it is absolutely vital that the signals reach the target precisely in phase. As such, we need to know very accurately the physical placement of each transmitting / receiving antenna as well as the pointing of the individual element main lobes. To achieve this each site is surveyed precisely and mapped. Each antenna reflector is very carefully measured to determine the individual radiation pattern (to measure the non-ideal shape) and those “real world” parameters are used to compute the actual beam phases in the signal processing system.


Array Operations


Selection and positioning of each monolithic antenna is vital. Individual antennas must be able to see and track the target across the sky and must not be blocked by neighboring antennas (shadowed). Additionally, the RF beam side lobes from each antenna can and will interfere and so their placement is critical as is the selection of which antenna to use in each subarray.


For a an array consisting of a large number of antennas, operations will be something like this…


A given target will be selected (say an asteroid). The basic two line elements will be known so that a target track can be predicted. This tells us where we need to point the main beam of the array and what the expected path will be across the sky. It also tells us when we can expect the target to “rise” and “set” at the horizon.


Given the target (size, approximate distance, desired measurement) a decision will be made as to how much radar power will be used for the observation. This will determine the number of monolithic antennas required to be combined as a subarray to support the track. As power is a factor of N^2 where N is the number of antennas used, this can easily be determined. To this minimum number some margin will be added in case of failure of an individual antenna during the observation.


Having determined the number of antennas required, we need to decide specifically WHICH antennas to use for the track, and when (or if) to add in or remove antennas during the track. This selected set will comprise the SUBARRAY for the track, and can be as few as one antenna (only a single dish tracking) or as many dishes as are in the entire array (full array support).


The selection of antennas needs to take into account several things:

  • Shadowing. There is a geometric limit below which a given antenna cannot usefully be used because its beam would be pointing directly at another dish. This not only is limiting to the measurement, it can be dangerous to the equipment (and possibly personnel) that would be in its beam. So, the geometries of each selected antenna (including their predicted view paths throughout the track) must be identified and deconflicted.
  • RF interference. The constructive and destructive interference produced by arraying antennas is what yields the benefit of the main beam. It also produces a set of periodic peaks and nulls that are not desired. These peaks and nulls are much more pronounced as the periodicity of the antenna spacing increases – this means that the more regular the spacing of the antennas (with respect to wavelength of the transmitted signal), the worse the interference pattern will be. As such, we want to select antennas that are placed as “aperiodically” as possible. A uniform grid is a very bad idea here.


To add to the complexity, all antennas produce side lobes. Real antennas produce more “jagged” patterns than ideal models. While a very good approximation can be made by using ideal models, we need to compute the interference patterns using actual measured patterns from the antennas.


Once the determination of specific antennas is finalized, the track is planned and executed. The dishes of the subarray track together (steering the beam as the target crosses the sky) and the radar observations are made. Once the track completes, the subarray antennas are “released” back to the pool to be assigned to a new track event.


There will be multiple, simultaneous tracks by the array and so several sets of subarrays must be identified and allocated in real-time.


Project Personnel

Barry Geldzahler
Program Executive,
NASA Chief Scientist for Space Communications and Navigation

Barry Geldzahler received BA and BS degrees in Astronomy and Mathematics from the University of Illinois, and a PhD in Astrophysics officially from the University of Pennsylvania – unofficially from the National Radio Astronomy Observatory with about 2 years spent at the Max Planck Institute for Radio Astronomy. This was followed by post-doctoral positions at MIT and NRL. His specialty is radio astronomy, specifically connected link and very long baseline interferometry. During his career in astronomy, Dr. Geldzahler has published in optical, radio, xray, UV, and gamma ray astrophysics. Dr. Geldzahler has worked in industry for 15 years on a variety of spacecraft missions including Clementine, and at APL on MSX, and NEAR (Near Earth Asteroid Rendezvous). He has been at NASA since 2001 where he has been the program executive for NEAR, Galileo (Jupiter orbiter), Stardust (comet flyby), the Planetary Data System, the Deep Space Network, and the Near Earth Network. Since 2009, Dr. Geldzahler has been NASA’s Chief Scientist for Space Communication and Navigation.

Jason Soloff
Systems Engineering Lead
NASA Human Exploration and Operations Mission Directorate

Mr. Soloff serves as a Systems Engineering Lead for the NASA Human Exploration and Operations Mission Directorate. In this capacity, Mr. Soloff is responsible for ensuring systems engineering of Human Space Flight programs. Mr. Soloff is also responsible for development, coordination and integration of international communications and space networking technologies, and is a member of NASA’s delegation to the Space Internetworking Architecture Group of the IOAG. Mr. Soloff’s experience includes staff positions in the GSFC Microwave & Communication Systems Branch, the JSC Avionics Systems Division, and as the Lead of the Avionics & Communications Office for the Constellation Program. Mr. Soloff’s other program experience includes the Global Precipitation Measurement Mission (GPM), the Lunar Reconnaissance Orbiter (LRO), the Space Shuttle, and the International Space Station. Mr. Soloff holds undergraduate and masters degrees from the Pennsylvania State University, and a Graduate Certificate in Space Systems Engineering from the Stevens Institute of Technology.


Karim R. Lakhani
Lumry Family Associate Professor of Business Administration, Harvard Business School
Principal Investigator, Harvard-NASA Tournament Lab at the Institute for Quantitative Social Science.

Karim R. Lakhani is the Lumry Family Associate Professor of Business Administration at the Harvard Business School and the Principal Investigator of the Harvard-NASA Tournament Lab at the Institute for Quantitative Show more

Rinat Sergeev
Data Scientist, Harvard-NASA Tournament Lab
Institute of Quantitative Social Sciences, Harvard

Dr. Rinat Sergeev is a Data Scientist at the Harvard-NASA Tournament Lab (NTL). Rinat works as a lead science and technical expert on exploring and utilizing crowdsourcing approaches in application to Big Data challenges,Show more

Jason Crusan
Director, Advanced Exploration Systems Division
NASA Human Exploration and Operations Mission Directorate

As Director for the Advanced Exploration Systems (AES) Division with the Human Exploration and Operations Mission Directorate (HEOMD) at the National Aeronautics and Space Administration (NASA), Jason Crusan is the senior executive, manager, principle advisor and advocate on technology Show more