PhD defence by Samina Rajbhandari

Abstract

Multimode fiber (MMF) offers an additional spatial dimension to address the capacity crunch driven by increasing data demands. When operated in a few-mode regime, MMF supports few-mode transmission and effectively behaves as a few-mode fiber (FMF). FMF transmission provides a cost-effective alternative to wavelengthdivision multiplexing (WDM) for increasing data throughput in short-reach intensitymodulation and direct-detection (IM/DD) networks, as a two-mode fiber can provide capacity comparable to that of a two-wavelength system while requiring fewer optical components. However, imperfect mode selectivity of the spatial multiplexer (MUX) and demultiplexer (DEMUX) introduces inter-modal crosstalk (XT), which, in shortreach links, is often more dominant than the transmission fiber itself.

While transmitter or receiver-side multiple-input multiple-output (MIMO) processing can mitigate this XT, it is typically undesirable in short-reach scenarios due to increased hardware complexity, power consumption, and cost. Therefore, low-complexity transmitter-side solutions are of particular interest. In this thesis, a multimode IM/DD link is modeled using a simplified intensitydomain MIMO framework, in which the dominant XT is assumed to originate primarily from the spatial MUX/DEMUX and to remain quasi-static over the observation window. Based on this model, low-complexity transmitter-side zero-forcing (ZF) based precoding is applied to mitigate XT. As an initial step, an emulated spacedivision multiplexing (SDM) transmission system based on a 3-dB coupler mesh is developed to experimentally validate the proposed ZF precoding approach. Using this platform, transmission experiments are conducted at a symbol rate of 10 Gbaud (hereafter denoted as 10 G) with pulse amplitude modulation (PAM)4 and PAM8 signals.

The results show that, for a fixed power budget, digital precoding improves the tolerance to XT by 1.3 dB and 3.3 dB for PAM4 and PAM8 signaling, respectively, at the KP4 forward error correction (FEC) bit error rate (BER) threshold (2.2×10−4). The approach is then demonstrated in a practical SDM system with a 1.6 km MMF. For simplicity, the experiments are restricted to two fundamental linearly polarized (LP) mode groups, LP01 and LP11, which together form an effective 2 × 2 spatial channel. This configuration allows MMF to be treated as a few-mode system. The modes are (de)multiplexed using air-clad photonic lanterns. Transmission experiments using 10 G PAM4 signaling show that precoding enables the LP01 mode to operate below the hard-decision FEC threshold, while the LP11 mode group satisfies the soft-decision-FEC requirement under average XT levels of approximately −5.95 ii Abstract dB for LP01 and −3.94 dB for LP11 mode group.

Although precoding improved the transmission performance, the initial real-link experiments also revealed practical aspects that required further investigation. Followup studies therefore examined least-squares-based channel estimation, mode-group combining of the degenerate LP11a and LP11b branches, and channel variations induced by natural drift and controlled polarization perturbations. The results show that LS-based estimation can improve precoding performance compared with powerbased estimation, while maximum-ratio combining provides an effective approach for treating the LP11 mode group. Overall, the thesis demonstrates that simplified intensity-domain channel modeling combined with low-complexity digital ZFbased precoding can improve the tolerance of short-reach IM/DD SDM systems to component-induced XT.

Supervisors

• Principal supervisor: Professor Søren Forchhammer, Department of Electrical and Photonics Engineering, DTU, Denmark
• Co-supervisor: Engineer Metodi Plamenov Yankov, Ciena, Denmark
• Co-supervisor: Professor Leif Katsuo Oxenløwe, Department of Electrical and Photonics Engineering, DTU, Denmark

Evaluation Board

• Associate Professor Michael Galili, Department of Electrical and Photonics Engineering, DTU, Denmark
• Professor Christophe Peucheret, L’Institut Foton, Université de Rennes, France
• Associate Professor Nick Volet, Department of Electrical and Computer Engineering, Aarhus University, Denmark

Master of the Ceremony

• Professor Karsten Rottwitt, Department of Electrical and Photonics Engineering, DTU, Denmark