Research Horizons

Laser Amplifiers for Space Communications

The implementation of laser and optical technology in space-based communications systems will potentially offer benefits over radiofrequency (RF) platforms. The higher-frequency optical carrier accommodates higher data rates, and the reduced diffraction, or beam spreading, enables greater security. This latter feature also reduces the size requirements on the antenna (or telescope) apertures, which potentially can reduce size and mass of the overall space platform. The trade-off for replacing RF with optical technology is that pointing requirements become much more demanding, and the long-term reliability of key photonic components in a space environment has yet to be proven.

An important component for space-based laser communications is the fiber amplifier. These devices are employed in both transmitters and receivers to compensate for the decrease in signal power per unit area due to propagating the long distances typical of space links. As explained by Todd Rose of the Photonics Technology Dept., fiber-based high-power amplifiers (HPAs) in the range of 1–10 watts are envisioned for the transmitters, and low-noise amplifiers (LNAs) in the range of 10–100 milliwatts are envisioned for the receivers. LNAs operating at a wavelength of 1.55 microns have been developed by the terrestrial telecom industry and are currently used in all terrestrial optical communication networks. Fiber LNAs for space applications are based on standard telecom fiber designs. They typically have micron-size cores doped with erbium. Light propagates down the core and is amplified by the erbium, which is optically pumped by low-power semiconductor laser diodes, Rose said. HPAs function similarly to their lower power counterparts, but have added features such as additional doping with ytterbium to improve pump absorption and specialized cladding around the cores to accommodate pumping with high-power laser diodes. Depending on the data rate requirements, HPAs and LNAs can amplify a single wavelength channel or a set of wavelength-multiplexed channels for higher capacity. Standard telecom components now support single-channel rates up to 40 gigabits per second, and systems multiplexing 32 or more channels have been demonstrated.

Todd Rose (right) and Jose Linares of the Photonics Technology Department have been studying the effects of radiation on fiber amplifiers for laser communications using Aerospace's cobalt-60 source.

"The success of fiber LNAs in terrestrial systems makes them attractive candidates for use in space," Rose said. The terrestrial telecom industry qualifies these parts according to a set of standard industry guidelines. "While these standards serve as a good starting point, additional qualifications are needed for the space environment—such as a greater thermal range, radiation tolerance, and operation through thermal cycles and in vacuum." Work is ongoing to modify terrestrial LNAs to achieve space qualification.

The HPAs present greater engineering challenges because they involve more exotic structures, pumping schemes, and chemistry. Device efficiency is also an important parameter for space applications because power is limited aboard spacecraft, Rose said. Furthermore, HPAs in the 1–10 watt range are not needed for terrestrial telecom at present, so their rate of development has been slower than that of LNAs.

Rose has been leading a team, which includes George Sefler, Heinrich Muller, George Valley, and Jose Linares that is conducting research on both LNAs and HPAs to evaluate their state of development and reliability. In particular, the team has been investigating the effects of radiation on amplifiers. When exposed to radiation, these devices develop "color centers" that absorb signal and pump light, which affects performance. The team's research indicates that these color centers reduce gain, but—unexpectedly—not the noise figure in these devices. "Most tests of unpumped amplifier fibers indicate that their degradation would be too severe in a space radiation environment to support a communication link over a mission life of 10 or more years," Rose said. Interestingly, under pumped or operational conditions, the formation of color centers is reduced by a photoannealing process that is not fully understood. Part of the team's research centers on how pump power and wavelength affects this annealing process and to what extent the annealing might render the fibers usable in space.

Another important goal of the Aerospace research effort is to develop models to predict the performance of HPAs doped with erbium-ytterbium. The modeling effort is more difficult for these fibers than for erbium-doped LNAs because the additional interaction between the erbium and ytterbium ions creates more pathways for the flow of energy, Rose said. Benchmarking of the models is also difficult because the exact composition and doping profiles of the fibers are not always known, either due to processing uncertainties or proprietary issues. Models developed by Aerospace have incorporated new processes to explain performance results and other measurements observed in the laboratory.

A third area of the Aerospace research has focused on developing better pumping schemes for fiber amplifiers. Typically, HPAs are pumped with many individually packaged laser-diode devices. According to Rose, the use of high-power laser-diode arrays could significantly simplify packaging requirements. To date, however, these devices have not been effectively used in this application because of the difficulty in coupling the optical output from these multiple-emitter structures into the fibers. Rose and his team are seeking efficient and reliable methods for coupling these arrays involving novel optical structures and waveguides.

Laser Beam Acquisition and Tracking

Laser-based communications offer greater capabilities than RF systems, but present greater challenges in implementation. One such challenge involves the difficulty of acquiring and tracking a concentrated beam of laser light arriving from another platform across vast reaches of space.

Aerospace has been investigating ways to optimize and automate the laser acquisition and tracking process. Research began a couple of years ago with the development of a two-terminal laser communications test bed, designed to simulate a separation of 80,000 kilometers between two satellites in a laser communications link. This effort combines the talents of personnel from both the Electromechanical Controls Department (ECD) and the Digital and Integrated Circuit Electronics Department (DICED). Implementation will require both field-programmable gate array (FPGA) and digital signal processing (DSP) hardware that will comprise two different, parallel implementations for a "front end" or camera head for the test bed. This test bed will be instrumental in evaluating algorithms, characterizing hardware, and verifying design concepts, whether derived by Aerospace or by contractors.

Schematic of the two-terminal laser communications test bed. The light lines do not end in the center, but pass through a far-field simulator that greatly reduces their diameter and power, thus simulating travel through 80,000 kilometers of space. Terminal 2 is an exact reflection of terminal 1, transmitting through its own series of optics and far-field simulator to the detector at terminal 1. Thus, the system is bidirectional, transmitting and receiving signals along the same path.

For the FPGA effort, John Maksymowicz of DICED has been able to design a control and image-processing algorithm whose timing and sequencing are governed solely by the control signals fed into and out of the optical sensor, a CMOS-based focal-plane array. Similarly, Kenny Conner of DICED has been developing the same processing functions on more generic DSP hardware through his development of embedded code.

Recently, Maksymowicz took his research a step further, extending the algorithm to work with the most advanced commercially available visible-wavelength focal-plane array. As Maksymowicz explains, the initial research used a sensor rated at 30 output frames per second, with a clock cycle of 10 megahertz and a single 10-bit pixel output per cycle. The newer work employs a focal-plane array with a minimum output of 500 frames per second with a clock cycle of 66 megahertz and ten parallel 10-bit pixel outputs per clock cycle. Thus, at its largest frame size, the sensor would generate pixel data 100 bits wide at 6.55 gigabits per second (1024 × 1280 pixels per frame × 10 bits per pixel × 500 frames per second). Further complicating matters, the sensor required external commands for its internal functions, such as the start/stop of photodetector integration and the resetting of the photodetectors and analog-to-digital converters. Thus, Maksymowicz had to develop "handshaking" logic—a communications protocol that would ensure proper two-way operations as well as help the FPGA establish the correct timing and sequencing of all the functions in the image-processing portion of the algorithm.

Maksymowicz generated a function that enabled the FPGA to map every pixel that was output by the focal-plane array to a specific coordinate on the array while it was actively scanning. The code allowed all of the sensor's 100 parallel output bits to be input to the FPGA in a single clock cycle. Thus, the FPGA was able to register every scanned pixel with only a single clock period delay. Each pixel value could then be compared with a predetermined threshold and, if validated, sent to a pair of divider circuits, which would in turn calculate the Cartesian coordinates of the laser beam's centroid. These x-y coordinates could be output for further processing, or sent as stand-alone commands directly to the sensor's pointing mechanism, enabling it to slew the sensor to maintain continuous, real-time tracking of the laser beam.

An FPGA-enabled imaging sensor similar to this one (the vertical printed wiring board with attached ribbon cable) will be integrated with the test bed for hardware demonstration.

For the DSP effort in integrating with the new focal-plane array, Conner attempted a similar implementation. This effort proved even more challenging. The goal was to process a single image frame from the focal-plane array within 2.0 milliseconds; however, the DSP development kit was limited in several important aspects. The main issue revolved around the DSP memory interface: Only half of the 64-bit memory bus was available for custom interface circuit design. Nonetheless, software for pointing, acquisition, and tracking was developed using both Matlab and C programming languages. Maksymowicz was able to identify another DSP architecture with a multiprocessor interface and higher throughput capability.

Both the FPGA and DSP design efforts are continuing. The next step will be to integrate them with the test bed for hardware demonstration and validation. The team also intends to modify the algorithm to enable autonomous switching of the focal-plane array between partial and full array scans.

To Spring 2006 Table of Contents