Radio over fiber

Radio over fiber (ROF) is the optical carrier which is modulated by wireless radio frequency signal and then transmits in the fiber [3]. In this way, we can take the advantageous of the low loss and low cost of the fiber to extend the transmission region of the radio frequency as in Fig. 1.1 . The traditional access network is to connect base station and the antenna through the cooper pair. However, when transmitting the high frequency signal, the severe loss will happen which cause the transmission length usually only under hundred meters. If we use the fiber, we transfer the radio signal into the optical signal through the head end in the base station and transmit to the remote antenna unit (RAU). In this way, we can set the base station and the RAUs in longer distance, moreover, the ROF also can be applied in the certain poor wireless place, e.g. in the tunnel [2].

6

ROF transmission systems are usually classified into two main

categories (RF-over-Fiber ; IF-over-Fiber) [2] depending on the frequency range of the radio signal to be transported.

a) In RF-over-Fiber architecture, a data-carrying RF (Radio Frequency) signal with a high frequency (usually greater than 10 GHz) is imposed on a light-wave signal before being transported over the optical link. Therefore, wireless signals are optically distributed to base stations directly at high frequencies and converted from the optical to electrical domain at the base stations before being amplified and radiated by an antenna. As a result, no frequency up/down conversion is required at the various base stations, thereby resulting in simple and rather cost-effective implementation is enabled at the base stations.

b) In IF-over-Fiber architecture, an IF (Intermediate Frequency) radio signal with a lower frequency (less than 10 GHz) is used for modulating light before being transported over the optical link. Therefore, before radiation through the air, the signal must be up-converted to RF at the base station.

7

Figure 1.1 Traditional wireless access network and Radio over Fiber network

8

Chapter Two

Protection passive optical network system

2.1 Study motivation

With the trend of the high bandwidth demand and exponential increasing internet traffic, the wavelength division multiplexing passive optical network (WDM PON) or the hybrid time division multiplexing (TDM/WDM PON) has been a good candidate for the future data traffic solution while the fiber transmission can provide the extreme high bandwidth and the WDM can support the high split ratio for large number of end user. However, when the distance of the transmission increasing and the number of the user in the access network ascending, there is more possibility that the fault in the feeder and distributed fiber will happen which is difficult and expensive to repair. Recently there are some protection architectures have been proposed [11-16] e.g. reference [6] and [7] need the electrical active component while [8] places the tunable laser in each optical network unit (ONU).

In this chapter, we propose and analyze two protection PON architectures [9] [10] which can support C band and the L band channels in the same time. Passive component 1XN AWG is equipped, and the high reliability is also calculated and compared with other NXN AWG scheme in our work. The feeder fiber and the distributed fiber are both in protection.

In section 2.2, we will propose our two protection schemes and explain the principle. In section 2.3, the result of the experiment are performed and

9

evaluated then in next section 2.4, we will calculate the reliability for these two systems and compare with other AWG WDM protection PON. Finally it’s our conclusion. circulator (OC) and then connect to the 2x2 coupler. One working fiber and another backup fiber are connected with the coupler and there is a wavelength coupler and a 1XN AWG before the ONU. The components in the ONU are a optical switch, coupler and c Band coupler and a pair of the C band Tx and Rx. As in the Fig. 2.2, due to the spectral periodicity property of the cyclic AWG, the upstream signal Ui will be assigned to the same output port of the downstream signal Di if the frequency space between the signal is the multiple of the FSR (Free spectrum region) of AWG.

In the normal working state, the two OSs in each ONU pair are “OFF.”

The OS is controlled by medium access control (MAC) layer. The downstream and upstream signal go through the fiber paths 1 and 2 while the signals in fiber paths 1’ and 2’ are blocked by the OS in each ONU.

When a fiber fault occurs on the distributed fiber (between the RN and ONU) or the in feeder fiber (between the remote node and the ONU) as

10

shown in Fig. 2.4, the signal can’t pass to the subscriber. Since the Rx of the ONU cannot receive the downstream signal, the OS switched “ON” by the media access control (MAC). Hence the ONUL reconnect both the downstream signal and upstream signals through the neighbor ONUC

protecting fiber on path 1’, as shown in Fig. 2.4 with N pairs of ONUs. In the CO part, optical circulator (OC) is used to separate the upstream and downstream signals. A 1×2 optical switch (OS) is used to connect the RN with two feeder fiber for working and backup. The RN consists of a 1×2 coupler, a 1×N WDM (AWG) and N 1×2 couplers. Every two adjacent ONUs are assigned to a pair group which is connected to the corresponding output port of the AWG.

Scheme 2

We utilize the spectral periodic property of the AWG In the same way to transmit the upload and download signal in the same AWG port.

The downstream wavelengths (CDi , LDi) and the upstream wavelengths ( CUi , LUi ) in the ith ONU group (for i=1, 2, …, N) are spaced by one free spectral range (FSR) of the AWG. Initially, two neighboring downstream signals (CDi , LDi) are coupled by a 1×2 blue/red-band WDM coupler (B/R CP) into the i-input port of AWG. Downstream signals also can be filtered by the 1×2 red and blue couplers. Here, atailor-made 2×2 optical switch (OS, produced by Lightwave Link, Taiwan) is added for self-protection.

The 2×2 OS in the same ONU group is connected to port “2” and “4” by two protecting fiber. A 1×2 re coupler is placed inside ONU1 to filter the downstream CDi and the upstream CUi wavelength channels, and a 1×2 blue coupler in ONU1’ is used to filter the downstream LDi and upstream LUi

11

wavelengths.

Next, we discuss the two scenarios of fiber fault. In normal operation, the 1×2 OS in the CO connects the ports “1” and “2” for the working path and the ports “1” and “3” for protecting the path. And, for the 2×2 OS in each ONU, the ports “3” and “4” both connect to port “1” in normal status for upstream transmission. When there are two occurrences of fiber faults, one in the feeder fiber and the other in the distribution fiber. Because all the upstream signals cannot be detected in the CO, hence the monitor circuit (MC) in the CO will detect a power drop and the MAC will trigger the OS to switch the ports (1→2) in the protecting fiber path for reconnecting the signal, as shown in Fig. 2.6. The C band signal will go through the 2X2 switch port ” 4 ” in ONU1’ to the ONU1.

2.3 Experiment and system performance

In this section, we will discuss about our architecture and the experiment result. First let have a look at the scheme 1. In our experiment, the wavelengths of CD1, CU1, LD1, and LU1 are 1546.0, 1548.0, 1562.0, and 1564.0 nm and the total length of the feeder and distributed fiber is 20 km.

The fiber between the ONU pair is 2 km. All the fiber are standard single mode fiber (SSMF). Both the downstream and up- stream signals are modulated at 10 Gb/s non-return-to-zero(NRZ) format with 2³¹ pseudorandom binary sequence data via a LiNbO3 Mach–Zehnder modulator (MZM). Optical pre amplifier is erbium-doped fiber amplifier (EDFA). In the [9], it shows the BER performances of downstream and upstream signals between the CO and the ONUC1 back-to-back

12

transmission and the 20 km single mode fiber (SMF) transmission and the protection state (22 km SMF transmission). The measured C band signals power penalties are less than 0.5 dB at a BER of 10⁻⁹ without and with fault cut and the sensitivity is about -26.5dBm. We also can see in [9], the transmission performance for the L band signals and the power penalties are about 0.8 dB at a BER of 10⁻⁹ for L band signals. By considering the insertion losses of the OS(~ 0.5 dB), an AWG (~5 dB), two CPs (~6 dB), a WC( < 1 dB), an OC ( < 0.5 dB), and the 22 km SSMF (0.2 dB/km), the total loss budget is about 18 dB. In practice, the proposed scheme can support 80 WDM wavelengths using both C + L bands (40 wavelengths in the C-band and 40 wavelengths in the L-band). The number of supported wavelengths can be further increased if dense WDM (DWDM) is used. The insets are measured corresponding eye diagrams for downstream and upstream signals with and without protection, respectively.

Now we are talking about the scheme 2 [10]. In order to investigate the transmission performance of our proposed self-protection system [10], we select four wavelengths to simulate the downstream and upstream channels for the CO and a pair of ONU1 and ONU1’. In this experiment, the CD and

13 modulation (IM). And the four signals were transmitted through the 25 km and 27 km single mode fiber (SMF), respectively, for working and protection statuses. The bit error rate (BER) performances of downstream and upstream signals between CO and ONU1; and CO and ONU1’ are shown in [10]. It also shows the BER curves between CO and ONU1

without and with fault protection. The measured power penalty was less than 0.5 dB at BER of 10⁻⁹. Since the wavelengths (close to L-band) located at one side of the EDFA gain profile, the power penalties between CO and ONU_1’ are larger than that between CO and ONU1. In addition, we also measured the restoration time of the proposed WDM-PON system.

The restoration time of OS was measured within 7 ms, as shown in the inset [10].

We compare our proposed scheme with other recently reported schemes in the literatures as shown in Table 2.1. The experimental results show that our proposed scheme has a higher power penalty between working and protection states (~0.8 dB), while other schemes in Refs. [7] and [8]. In Ref.

[6], there is about 0.5 dB and negligible power penalty, respectively.

Our scheme can protect both feeder and distribution fiber without active control between the CO and ONU. Moreover, other protection schemes [17-19] could only be employed in C-band for WDM operation. For

14

example, 40 wavelength channels can be used in the C-band (1530 to 1560 nm) for WDM transmission while the channel spacing is 100 GHz (0.8 nm).

Here, our proposed PON architecture can also support both C to L bands

Our availability models are illustrated by reliability block diagrams (RBDs) [21] as in [20] and are shown in Fig. 2.7, 2.8 and 2.9. The asymptotic unavailability value [22-23] for each block in the diagrams is for a component or fiber section. For these values, we refer to [22-23] and show them in Table 2.2. In the reference [20], they compare and calculate other protection PON structures [57-60] with RBD. Therefore, we also compare our schemes unavailability to these protection schemes. We transfer the protection system scheme 1 and scheme 2 into the RBD as Fig. 2.9 and using the unavailability value in Table 2.2 to calculate the reliability of the scheme 1 and 2. We do some assumption as follow, the optical coupler, C band optical coupler, wavelength coupler and the splitter are the same unavailability. And the optical circulator which is in the dash block is excluded for our calculation because in [20], the OC has been included in the OLT already. The feeder fiber plus distribution fiber length is 20 km for working path while the fiber length is 2 km for distribution fiber and the interconnection fiber length is 3km for neighbor connection. By the way

15

It has to be mentioned that our reliability calculation is based on each ONU.

After our assumption, the scheme “1” unavailability is 1.73e-6 and the scheme “2” is 6.02E-6 in the Fig. 2.10. We comparing to the Ref. [20,57-60]

which using the NxN AWG in their protection systems while the lowest unavailability of protection scheme is 2.43E-6 in [20]. From Fig. 2.10, we show that our proposed protection scheme 1 has the best reliability and our scheme 2 still is better than un-protection and Ref. [57-58] scheme. In our analysis, the key point to reduce the unavailability is the protection length in the system. As we can in the Table 2.2, most of the passive components have the similar unavailability near 1E-7 while the optical fiber has the highest unavailability. When the transmission distance up to 25km, the unavailability from fiber will dominate the system reliability and we through the protection path to make the unavailability multiplication instead of add. By doing so, we can decrease the unavailability a lot for the system even we increase our TDM split ratio to 1:64.

Comparing to our scheme to the NxN AWG protection scheme [20], both protection region include feeder fiber and distribution fiber, and the supporting ONU number both are large. In our scheme “1”, it can support both C and L band about 80 channel when the guard band is about 0.8nm which is equal to 100Ghz. Adding the time division multiplexing (TDM) split ratio 1:16, the scheme “1" can support the 1280 ONUS with the low unavailability.

2.5 Conclusion

In our work, we propose two feeder and distribution fiber protection

16

systems which both can support C and L band with the AWG periodic spatial feature. We show up the architecture and the experiment of these two schemes and the error free at received power -26.5dBm. The supporting number of ONU is 80 wavelength channels multiple the TDM ratio 1:16 which is 1280. We also compare about the reliability of our scheme with other protection systems which use NxN AWG. The result is that our schemes have the higher reliability in the same condition.

17

Figure 2.1 Schematic of the proposed C + L bands WDM-PON.

Figure 2.2 The wavelength assignments for the upstream and downstream signals.

18

Figure 2.3 The proposed distributed-controlled protection WDM-PON architecture.

19

Figure 2.3 Two fiber faults occur at the distributed fiber and the distributed fiber, and the recovery system signal path.

20

Figure 2.5 Another proposed self-protection WDM-PON architecture

21

Figure 2.6 The network protection scheme when the fiber links are broken

22

Figure 2.7 Reliability block diagram (RBD) for (a) an unprotected scheme, (b–d) the existing protection schemes in [57–59]

(a) (b) (c) (d)

23

Figure 2.8 Reliability block diagram (RBD) for (e) [60] (f-g) [20]

(e) (f) (g)

24

Figure 2.9 Our protection scheme 1 and scheme 2

(scheme1) (scheme2)

25

Figure 2.10 The unavailability of our propose two schemes and the other protection schemes

1.00E-06 1.00E-05 1.00E-04

Unavailibility

Other paper Our scheme

26

Table 2.1 Comparison of our proposed scheme with other recently reported schemes.

27

Table 2.2 Reliability data sheet

28

Chapter Three

Radio over fiber long reach PON

3.1 Study motivation

Future access networks need to provide broadband services with wired and wireless approaches. The convergent optical wired and wireless access network has been proposed to provide broadband services in a single and integrated perform. Passive optical network (PON) is promising for providing wired services, while radio-over-fiber (ROF) is an important technique for providing wireless services in the optical domain. The simultaneous generation and transmission of both signals for PON and ROF for the convergent optical wired and wireless has been proposed and demonstrated [24–33]. The scheme reported in [31] only needs a single electro-absorption modulator (EAM) to produce baseband and ROF signal simultaneously. However, the signal performance is limited by the nonlinearity, chirp of the EAM. Techniques used in the integrated ROF and access networks reported in [29] and [32] either require optical modulator at the remote node (RN) or multiple laser sources which will complicate the system. System reported in [33] requires electrical up-conversion at each base station, which is costly and complicated. Techniques used in [24]

and [30] can provide simple generation and demodulation of the baseband and ROF signal, however, the frequency separation between the two bands signal should be larger than the bandwidth of the signal to avoid cross-talk.

29

Our proposed convergent optical wired and wireless scheme is spectral efficient, in which the channel spacing equals the bit-rate per subcarrier. It does not require multiple laser sources, or electrical up-conversion at the RN or at the base station. Recently, using orthogonal multi-carrier modulations provide cost-effective and high spectral-efficient optical communication. These modulations include orthogonal frequency division multiplexing (OFDM) [34], coherent wavelength division multiplexing (WDM) [35] and orthogonal WDM [36]. And orthogonal WDM is promising since its operation speed does not limited by the electronic bottleneck caused by the digital signal processing (DSP).

Besides, long-reach (LR) access can integrate the metro access for simplifying the network architecture [37]. In this work, we propose and demonstrate a convergent optical wired and wireless LR access networks based on orthogonal WDM. High signal spectral efficiency can be obtained.

Although spectral efficiency is not the top priority issue for the present access networks, it is not true for the case of WDM LR access (since it integrates the present metro and access sections [37-38]). As the optical amplifier has a fixed and limited bandwidth of about 30 nm. Increasing the spectral efficiency of the WDM LR access is essential. Here 5 Gb/s baseband non-return-to-zero (NRZ) signal and 10 GHz double sideband ROF signal (carrying 5Gb/s data) are orthogonally wavelength-division-multiplexed. They occupy a bandwidth of 20 GHz.

Error-free de-multiplexing and down-conversion can be achieved after 60 km (long-reach) of single-mode fiber (SMF) transmission without dispersion compensation. Time-delays and power differences between the

30

baseband NRZ and the ROF signals in the transmitter (Tx) are characterized. The scalability of the system for higher bit-rate (60 GHz) is also discussed. In the section 3.2, we illustrate the experiment and the simulation about our ROF system. Unless the OOK modulation, we also apply differential phase shift keying (DPSK) modulation format to our proposed system without adding optical equipment in section 3.4.

We briefly introduce about the DPSK modulation. First we can start from the binary phase shift keying (BPSK). BPSK modulation is to modulate the optical carrier phase with the alter binary signal, e.g. the carrier phase will be modulated in 180 degrees if sequence is ‘0’; while 0 degrees when the sequence is ‘1’. DPSK is doing the difference encode before the modulation. For a example, there is a original sequence bk, after encoding, the sequence bk become dk. Then, we use the differential encoded sequence dk to do the binary phase modulation. If the dk equals ‘1’, the optical phase is 0 degrees; while the dk equals ‘0’, the optical phase is 180 degrees. The relation between the bk and the optical carrier is that when bk

equals ‘1’ the carrier

31

If we use the phase modulation in our system, we can’t avoid the coherence detection which needs the optical phase lock loop (OPLL) [39]

to lock the optical carrier. And it is very difficult to manufacture the OPLL moreover there is the loop delay. The advantageous of the DPSK is that we can use the delay-line in optical communication system to modulate the BPSK system which has the coherence optical carrier. Comparing to the OPLL, delay-line is much easier and reliable to be utilized in the system. In this way we can avoid the OPLL to implement the coherence detection in phase modulation. In section 3.4, the setups of ROF systems which use the DPSK are explained. In section 3.5 the measurement result for BER and the eye diagram are shown in this section and the power difference which influents the signals performance is also tested.

3.2 System setup and the principle

Fig. 3.1 shows the experimental setup of the wired and wireless access network. A continuous wave (CW) signal at wavelength of 1550.97 nm was divided into two paths by a 3-dB coupler. A Mach-Zehnder modulator

Fig. 3.1 shows the experimental setup of the wired and wireless access network. A continuous wave (CW) signal at wavelength of 1550.97 nm was divided into two paths by a 3-dB coupler. A Mach-Zehnder modulator