Andrew Poulsen and Dr. Brian Jeffs, Electrical and Computer Engineering
When images of remote galaxies/stars are observed by radio telescope, much sensitivity can be blurred by interference. This interference can be caused by both the atmosphere and man-made sources. Specifically, the Russian GLONASS satellite system transmits data at a bandwidth wherein much information important to radio astronomers is radiated by astronomical objects. A real-time DSP-based system is needed to adaptively remove the interfering signals and restore clarity to the radio astronomy signal.
During the past eight months, I have worked with two advising professors (Brian Jeffs and Karl Warnick) and other students to build a test platform with which we may evaluate the effectiveness of various adaptive algorithms. We installed an array of three 10 foot parabolic reflector antennas each with two positioning rotors for movement in both azimuth and elevation. Antenna control software was developed to track celestial objects such as the sun, moon and stars. Real-time tracking of satellites such as GLONASS and IRIDIUM was also demonstrated; it is necessary to be able to track the interference source for many adaptive cancellation architectures which will be investigated. In conjunction with the antenna control software, complex hardware was also built and debugged.
The majority of my time has been spent implementing the DSP software architecture necessary for the test platform. The DSP platform consists of a Pentek system using an array of Texas Instruments TMS320C6701 processors along with high-speed RF digital receivers. Much time was spent defining the interface between the four digital receivers and the four DSP processors.
A system capable of filling DSP memory and transferring the data to the PC host computer was developed. A Digital Integrating Radiometer is now capable of executing power computations at sampling rates from 1 MHz to 32 MHz. A floating point real-time beamformer was implemented at rates from 1 MHz to 4 MHz. Most recently, correlation computation between three separate array elements has been added, running at rates from 1 MHz to 2 MHz.
The next step in the software development is to add the Least-mean-square algorithm to the architecture. It will be the first in an array of potentially useful interference cancellation algorithms to be investigated. I plan to continue working on this project for my Masters thesis.
Programming on a real-time DSP platform can be quite difficult and cumbersome at times, though it is rewarding to solve a difficult problem. I have enjoyed my research experience and expect to be able to learn much more in the months ahead. I find it a noble cause to reduce the amount of data astronomers lose as satellites “blind” their observations with significant interference.
