With the advent of the 5G era, the continuous development of emerging applications such as cloud computing, short videos, big data, and the Internet of Things has led to a substantial increase in mobile users, resulting in growing demands for network bandwidth and transmission capabilities. However, the next generation passive optical network (NG-PON) system cannot effectively meet these demands, making it imperative to enhance NG-PON’s bandwidth supply and transmission capacity. In addition, applying direct detection (DD) technology to NG-PON can reduce system costs and implementation complexity. Traditional PON systems do not support dynamic service modes and require predefined physical connections, significantly limiting flexibility and failing to meet the demands of future broadband access networks for high dynamism, reconfigurability, and adaptability. Digital filtering multiple access (DFMA PON), on the other hand, can fulfill these requirements and effectively support 5G fronthaul services, playing a crucial role in 5G bearer networks. However, DD DFMA-PON, as a short-reach optical communication system, is highly sensitive to periodic power fading caused by fiber dispersion. As a result, dispersion becomes the primary factor limiting the transmission performance of DD DFMA-PON systems.

To address the issue of periodic power fading induced by dispersion and further improve transmission performance, we propose a parallel intensity modulation/phase modulation (IM/PM) scheme. This approach effectively mitigates the frequency-selective fading caused by dispersion. First, the complementary characteristics between IM/DD and PM/DD are analyzed theoretically. The normalized frequency responses of IM/DD and PM/DD are approximately complementary. When IM/DD experiences frequency fading, the PM response reaches its peak. Leveraging this property, the transmitter allocates subcarriers such that some carry signals modulated using IM, and others use PM modulation in the optical domain. This ensures that after direct detection via a photodetector (PD), both signals exhibit flat frequency responses. Next, the principles of LMS and Volterra adaptive filters are discussed. The Volterra series model, known for its strong adaptability, can be effectively combined with classical adaptive algorithms like LMS. This allows the model to accurately describe nonlinear systems with memory and dynamic behaviors. At the receiver, a second-order Volterra adaptive filter based on the LMS algorithm is employed to equalize and recover the received signals, further improving the system’s transmission performance.

Theoretical analysis demonstrates that IM/DD and PM/DD exhibit complementary effects (Fig. 1). Based on this property, the proposed parallel modulation achieves a flattened signal power spectrum (Fig. 4), thus mitigating frequency-selective fading. This improves signal reception sensitivity (Fig. 5). Furthermore, the Volterra filter combined with the LMS algorithm further enhances overall system performance (Fig. 6).

In this paper, we first theoretically analyze the complementary relationship between the frequency responses of IM and PM signals, and propose a dispersion compensation scheme for DD DFMA-PON systems based on a parallel IM/PM transmitter structure. Leveraging this property, the transmitter divides subchannels into two groups and allocates them based on the frequency response characteristics of IM and PM signals, applying different optical modulation schemes to each group to mitigate signal power fading caused by fiber dispersion. On the receiver side, a second-order Volterra adaptive filter based on the LMS algorithm is employed for equalization and recovery of each subband, further enhancing the system’s transmission performance. Simulation results demonstrate that, by exploiting the property that the sum of the frequency responses of IM and PM signals approximates one, the proposed scheme effectively compensates for dispersion-induced signal degradation. Under the conditions of 40 filter taps and a received optical power of -10 dBm, the system achieves stable transmission over 25 km, with a bit error rate below 3.8×10^-3 and a receiver sensitivity improvement exceeding 1.5 dB. This approach offers a theoretical foundation and technical support for future optical access networks.