Nowadays, passive optical networks (PONs) are the ultimate solution for optical access networks and have been deployed worldwide. To meet the rapidly increasing bandwidth demand, high-speed next-generation PONs are expected. PONs are cost-sensitive, and the main cost of a PON comes from constructing the optical distribution network (ODN), which can be up to 80%. Therefore, a smooth PON upgrade is highly desirable as it reuses the deployed ODN instead of rebuilding a new one. Since the new PON link shares the ODN with the legacy PON link, the coexistence of the two PON links with low crosstalk must be ensured. In the upstream direction, the new PON optical network units (ONUs) use an upstream wavelength different from that of the legacy PON ONUs, and the new PON upstream signal and the legacy PON upstream signal can be separated by a wavelength division multiplexer/demultiplexer (WDM) in the optical line terminator (OLT). The cost of adding one WDM in the OLT is acceptable. In the downstream direction, the separation (or filtering) should be in ONUs. The new ONUs have optical filters to filter out the legacy PON signal. However, the legacy ONUs have no optical filters to protect them from the new PON signal because they are designed with the coexistence in mind when they were installed. The retrofitting cost of adding one WDM to each legacy ONU is too high due to the large number and wide distribution of legacy ONUs. Thus, the crosstalk from the new PON downstream signal to the legacy PON downstream signal should be eliminated or reduced.

We utilize several wavelengths to construct a novel mark ratio modulation (MRM) based on wavelength coding for single-wavelength coexistence. The pulse position modulation (PPM) is transferred from the time domain to the wavelength domain, meaning that the pulse position is the “wavelength position” instead of the “time position”. The new PON signal is mapped to four sequences through the 4PPM coding method. Then, the legacy PON signal modulates the mark ratio of the four sequences by inverting them. When the legacy PON signal is 1, there will be three marks among the four corresponding bits, and when it is 0, there will be one mark. The four sequences are loaded to four transmitters of different wavelengths. The four optical signals are combined by a wavelength division multiplexer/demultiplexer (WDM) and then fed into a feeder fiber. In the remote node, the four signals are split into ONUs. Each legacy ONU receives all optical signals because there is no optical filter or WDM. The total amplitude depends on the sum of the four signals, and the mark ratio among the four bits is modulated by the legacy PON signal, so the total amplitude corresponds to the legacy PON signal. Each new ONU separates the four optical signals by a WDM and then receives them separately. The received signals are then decoded to recover the origin new PON signal.

The tested eye diagrams are shown in Fig. 4. The 10 Gbit/s legacy PON signal shows worse performance than the 10 Gbit/s individual signal. This is because the extinction ratio (ER) of the legacy PON signal is limited, and the delay difference and amplitude difference of the four optical signals cannot be completely compensated. To further demonstrate the signals’ performance, bit error rates (BERs) are measured and shown in Fig. 5. The legacy PON signal shows worse performance compared with S₁. The 2.5 Gbit/s legacy PON signal shows BER performance close to that of the 10 Gbit/s S₁. To reach the same BER level, the 10 Gbit/s legacy PON signal requires 3 dB?5 dB higher received power than S₁. The receiving sensitivities of the 1.25, 2.5, and 10 Gbit/s legacy PON signals are -30.5 dBm, -28.8 dBm, and -25.5 dBm, respectively. The legacy PON signal is a combination of four optical signals. Assuming the output power of each transmitter in the OLT is 3 dBm, the total power of the transmitted four optical signals reaches 9 dBm in the OLT. So the power budgets are 39.5 dB, 37.8 dB, and 34.5 dB, respectively. To test mismatch induced signal degradation, the BERs of the legacy PON signals are measured again, as shown in Fig. 6. For the 1.25 Gbit/s signal when the mismatch distance does not exceed 1 km, there is minimal variation in BER with almost no power loss; at 2.5 km, the power loss is approximately 0.5 dB. For the 2.5 Gbit/s signal, the BER difference widens, especially reaching the error limit at a 2.5 km mismatch; at 1 km, there is still no significant power loss. The 10 Gbit/s signal displays more pronounced attenuation, with the 2.5 km mismatch resulting in BER measurement failure, and only a small power loss of about 0.5 dB at 0.5 km. The mismatch compensation does not affect the separately received signal S₁. A smooth PON upgrade is achieved by the proposed wavelength coding based MRM. The new PON link is added without changing the legacy ONUs and the deployed ODN. The new PON signal and the legacy PON signal use the same wavelengths. The upgrade is “traceless” because all wavelengths can be switched to carry independent signals individually after the upgrade without hardware retrofitting.

In our present study, a novel wavelength coding based MRM is proposed for PON upgrade. The proposed method uses several wavelengths to carry both the new PON signal and the legacy PON signal. The new PON signal applies PPM over wavelengths. The signal is modulated by selecting one specific wavelength. The legacy PON signal is modulated by inversing the coded new PON signal, which changes the mark ratio of the coded signals. At the receiving end, the legacy PON signal can be received by a traditional amplitude-shift-keying (ASK) receiver directly and the new PON signal is recovered from the separately received optical signals. After the upgrade, the coding can be switched off, and each wavelength can carry an independent signal, so the upgrade is “traceless”. In the upgrade based on the proposed method, orthogonal modulation is not used, and the bit rate of the new PON signal is not reduced. The proposed upgrade is demonstrated by simulation. In the simulation, four wavelengths are used, and the new PON signal applies 4PPM. 1.25, 2.5, and 10 Gbit/s legacy PON signals are tested when the transmitting rate of each wavelength is set as 10 Gbit/s. The transmission length is 25 km. The time misalignment and the power difference among the optical signals are pre-compensated in the OLT. In the simulation, the legacy PON signal and the new PON signal show good performance. The power budget is sufficient, satisfying most current PONs’ specifications. For the 1.25 Gbit/s legacy PON signal and 2.5 Gbit/s legacy PON signal, the power penalty from the degradation induced by the compensation mismatch is negligible when the compensation mismatch is not beyond 1 km. The measured results verify the feasibility of the proposed wavelength coding MRM.