Research Overview

The Optical Communications Group, led by Professor Kahn, applies principles of photonics, signal processing, communications and information theory to advance the performance of communication or imaging systems. We have demonstrated sustained technical leadership in several optical communication media, including single-mode fiber (SMF), multi-mode fiber (MMF), and free-space optical (FSO) links, setting world records for speed and efficiency in each medium. We have also made seminal, highly cited contributions to multi-input, multi-output (MIMO) wireless communications and sensor networks.

Spatial multiplexing in MMF, a form of MIMO transmission, can dramatically enhance capacity in long-haul core networks or data-center networks, and will be needed to accommodate exponentially growing traffic demands. Long-haul systems already use multiplexing in the two polarizations of SMF, which is enabled by coherent detection and digital MIMO processing. Capacity can be increased about tenfold by multiplexing in the spatial modes of MMF. Long-haul MMF systems will require multi-mode optical amplifiers, but their mode-dependent gain will cause "fading" that fundamentally limits transmission capacity and system reliability. Over the past year, in three seminal papers, we described channel models, fading statistics and noise statistics for MIMO systems in MMF, quantified the impact of fading on capacity and outage probability, and analyzed frequency diversity for reducing outage probability. We formulated a rigorous definition of diversity order based on principal component analysis, which is equally applicable to wireless MIMO systems. We are now designing multi-mode transmission fibers and amplifiers, as well as MIMO coding techniques and digital MIMO processing algorithms. To address capacity demands in data centers, where only direct-detection links are considered practical, we are experimenting with spatial light modulators (SLMs) for adaptive optical MIMO signal processing. This builds on our previous work using SLMs to compensate for modal dispersion, which enabled 10 Gbit/s and 100 Gbit/s data transmission far beyond traditional bandwidth-distance limits.

Modern wireless and wireline communication systems use rate-adaptive coding and modulation, whereas all optical systems to date have used fixed rates. We are pioneering methods for rate-adaptive and bandwidth-scalable coding, modulation and multiplexing in optical networks. We have developed rate-adaptive coding and modulation techniques for long-haul or metropolitan networks, employing a family of low-density parity-check (LDPC) codes with different orders of quadrature amplitude modulation (QAM). This scheme enhances network flexibility and reliability, enabling very high bit rates over relatively shorter distances (200 Gbit/s up to 2000 km), while enabling lower-rate transmission over distances far longer than conventional methods (100 Gbit/s to 5000 km, 50 Gbit/s to 8000 km). Currently, we are adding shaping codes with higher-order QAM to enhance the performance at high bit rates (200 to 400 Gbit/s) at shorter distances. We are also developing bandwidth-scalable modulation and multiplexing techniques for long-haul or metropolitan networks, which use channels of variable bandwidth (nominally 50 to 400 GHz) to convey variable bit rates (100 Gbit/s to beyond 1 Tbit/s), depending on traffic demands. Our scheme increases spectral efficiency substantially. Unlike existing approaches for terabit-per-second channels, ours uses colorless (tunable) transceivers and (de)multiplexers, a key requirement for reconfigurable optical networks. Hence, our work has attracted considerable attention for future 1 Tbit/s Ethernet systems. One key innovation is a variable-bandwidth, (de)multiplexer with colorless add/drop ports, which is based on a liquid crystal-on-silicon (LCOS) wavelength-selective switch (WSS), and we are working with an industrial partner to demonstrate this device. A second key innovation is cooperative transmission of wideband channels by multiple synchronized colorless (tunable) transceivers. This is more flexible and economical than previous wideband transceivers, which address a fixed bandwidth at a fixed wavelength. We have designed transceivers using orthogonal frequency-division multiplexing (OFDM), and are now designing transceivers based on single-carrier modulation.

Accurate and efficient metrology of optical networks is a key to their successful design and deployment. In long-haul networks, polarization-mode dispersion and polarization-dependent losses are of particular importance. We are working to demonstrate a coherent network analyzer intended to measure these important fiber characteristics with unprecedented accuracy and speed. In the future, we plan to develop coherent network and signal analysis methods for spatially multiplexed systems.

Our deep expertise in coherent detection and associated digital signal processing algorithms are keys to the research described above. In recent years, we have performed seminal work on equalization techniques enabling reduced sampling rates, carrier phase recovery techniques enabling soft-decision forward error-correction codes, and digital methods for compensating fiber nonlinearity enabling data transmission far beyond previously assumed "fundamental limits".

For many years, we have been leaders in FSO communications research, having authored some of the most highly cited papers on both long-range outdoor systems and short-range indoor systems. In the past year, we have studied adaptive coherent array receivers to mitigate turbulence-induced fading in outdoor links, and have analyzed the impact of fog on such links. We have also completed the first systematic comparison of OFDM and single-carrier modulation for mitigating multipath distortion in indoor links, concluding that single-carrier modulation performs better in most applications.

Imaging and targeted light delivery via MMFs represent a new research thrust for our group. We use an SLM to control the field profile input to a MMF, and thus to control the field profile at the output end. Our work may enable a single-fiber scanning microscope for minimally invasive in vivo imaging, and may enable adaptive or reconfigurable optogenetics, adaptive phototherapy, or addressing of multiple photonic sensors. Our initial approach was to form a concentrated spot of light that can be scanned in the space near the fiber output. We have developed provably optimal algorithms for adapting the SLM, using a camera placed at the fiber output to provide a feedback signal. Once the camera is removed, this apparatus can be used as a single-fiber scanning microscope. More recently, we demonstrated an imaging method using random SLM patterns to generate random intensity patterns for sampling an object. Remarkably, this method can resolve a number of image features equal to four times the number of spatial modes propagating in the MMF.


Back to Joseph M. Kahn's home page

Last modified: November 20, 2012.