Digital Filter Multiplexing-enabled Advanced Networking Devices and PON Architectures for 5G Network Convergence

Electronic versions

Documents

  • Yixian Dong

    Research areas

  • PhD, School of Electrical Engineering, 5G network, passive optical network, digital filter multiple access

Abstract

To meet the stringent 5G network requirements, passive optical networks (PONs) are considered as one of the most promising strategies for seamlessly converging independently developed optical access/metro networks and mobile fronthauls (MFHs)/backhauls (MBHs). To enable PON-based converged 5G networks to offer various on-demand services, apart from the implementation of software defined networking (SDN), it is greatly advantageous if the PONs can provide sufficiently high flexibility, elasticity, reconfigurability and adaptability, as well as transparency to major network design characteristics. In addition, advanced reconfigurable optical add/drop multiplexers (ROADMs) are also envisaged to play a vital role in delivering dynamic and transparent connectivity between an expanded number of individual network nodes over various networks with diversified design/traffic characteristics. To address all the aforementioned challenges, in this thesis, four advanced techniques targeting the future converged 5G networks are proposed and explored, including: i) multiple channel interference cancellation of digital filter multiple access (DFMA) PONs based on intensity modulation and direct detection (IMDD), ii) hybrid orthogonal frequency division multiplexing (OFDM)-DFMA PONs, iii) hybrid discrete Fourier transform (DFT)-spread OFDM-DFMA PONs, and iv) digital signal processing (DSP)-enabled optical-electricaloptical (O-E-O) conversion-free ROADMs without incorporating optical filters. The previously reported DFMA PONs utilise SDN controller-managed, DSP-based digital orthogonal filters to perform channel multiplexing/demultiplexing. For the cost-sensitive 5G application scenarios, the degradation of digital filtering-based channel orthogonality introduces significant cross-channel interferences. To effectively reduce the cross-channel interference effect, a DFMA channel interference cancellation (DCIC) technique is proposed and investigated, and a comprehensive DCIC theoretical model is developed. Through extensive numerical fitting with experimental measurements, the developed theoretical model is rigorously verified, and a set of accurate transceiver/system parameters are also identified. Detailed numerical simulations show that the DCIC technique increases the aggregated upstream signal transmission capacity by a factor of >2 and extends the differential optical network unit (ONU) launch power dynamic range by >14 dB. The aforementioned performance improvements are ONU-count independent and just require one DSP iteration stage. Other salient DCIC advantages include low DSP complexity, negligible latency and excellent transparency to signal modulation format, signal bit rate and initial system operation conditions. In the IMDD DFMA PONs, the required parallel digital filters implemented in the optical line terminal (OLT) is proportional to ONU count, thus the corresponding DSP complexity increases with ONU count. To further improve the PON performance and simultaneously reduce its hardware/software complexity and installation/operation expenditure, a hybrid OFDM-DFMA PON is proposed and investigated, where multiple independent OFDM channels are multiplexed using ONU-embedded dynamically reconfigurable and adaptive digital filters, and simultaneously recovered in the OLT by a single fast Fourier transform (FFT) operation. The proposed hybrid OFDM-DFMA PONs have all the desired advantages of the DFMA PONs including: i) DSP-enabled dynamic network reconfigurability, flexibility and elasticity, ii) inherent transparency to signal modulation format, signal bit rate and network topologies, and iii) capability of SDN-based network abstraction and virtualization. An analytical theoretical hybrid OFDM-DFMA PON model is developed, based on which extensive numerical simulations of upstream performances are undertaken. It is shown that the proposed PON offers additional unique features including greatly relaxed digital filter DSP complexity and further improved performance flexibility and robustness. Moreover, numerical results also show that the hybrid PONs can increase the differential ONU launch power dynamic range by 16 dB compared to the DFMA PONs. To overcome high peak-to-average power ratios (PAPRs) of OFDM signals, an advanced variant of hybrid OFDM-DFMA PON, termed hybrid DFT-spread OFDM-DFMA PON, is proposed, which offers all the salient features associated with the hybrid OFDM-DFMA PONs. More importantly, it additionally reduces the upstream signal PAPRs by ≥2 dB. As a direct result, in comparison with the hybrid OFDM-DFMA PON, the hybrid DFT-spread OFDM-DFMA PON can increase the upstream system power budget by ≥3 dB and reduce the minimum required digital-to-analogue converter (DAC)/analogue-to-digital converter (ADC) quantization bits by at least 1 bit. To cost-effectively provide dynamic and flexible network connectivity with reduced latency, optical filter- and optical-electrical-optical (O-E-O)-free ROADMs with excellent flexibility, adaptability and transparency to physical-layer network characteristics are investigated, which perform DSP-enabled dynamic add/drop operations at wavelength, subwavelength and orthogonal sub-band levels. In IMDD-based network nodes, detailed investigations show that: i) the add/drop operation is digital filtering space locationindependent; ii) the add and drop operations introduce around 3 dB power penalties, and iii) for distances of <40 km, the ROADM drop operation is robust to transmission system impairments. In summary, this PhD dissertation research presents a wide range of major elements required for realising the converged 5G networks. The work paves a solid platform leading to practical implementation of the converged 5G networks in a cost-effective manner.

Details

Original languageEnglish
Awarding Institution
Supervisors/Advisors
Thesis sponsors
  • Sêr Cymru National Research Network in Advanced Engineering and Materials
Award date2 Oct 2019