Open Access
Add to BibSonomyAbstract
Two ∆Σ-modulated digital radio-over-fiber (DRoF) transmission systems that employ a multi-pulse Manchester encoder are proposed and experimentally evaluated. With a two-step modulation process comprised of ∆Σ modulation and multi-pulse Manchester encoding, a high frequency replica or image of a ∆Σ-digitized analog communication signal can be transmitted without significant power loss. This is achieved by exploiting the spectral characteristics of the modified Manchester code. For comparative analysis, a conventional ∆Σ-modulation-based DRoF system is also evaluated. Based on the evaluation results, the proposed DRoF systems more significantly improve the reliability and flexibility of the RoF system by providing higher power margins or by making the DRoF system implementation more cost-effective and easier to perform on account of the low-frequency requirement for electronics and optical transceivers.
© 2017 Optical Society of America
1. Introduction
In recent years, a radio-over-fiber (RoF) transmission system, in which the radio frequency (RF) signal is modulated on a lightwave signal at a central unit (CU) and transmitted over a fiber optic cable to a remote unit (RU), has attracted widespread interest [1–5]. This is because the installation and maintenance cost of the link can be considerably reduced with the simple RU structure. In addition, the RoF system is transparent to the modulation format of the RF signal, which is beneficial to future upgrade. A typical RoF system is shown in Fig. 1. However, the quality of the transmitted RF signal over the conventional analog RoF link deteriorates in proportion to the link length. Moreover, it is considerably affected by the nonlinearity of both electrical and optical components in the transmission system [4,5]. Therefore, transmission systems that prevent the signal quality deterioration and thereby enable more robust and reliable operation need to be developed.
A solution for this problem is to digitize the analog RF signal before transmission over the RoF system [6–11]. This solution is promising because the impairments of the analog RoF link are mainly due to the analog nature of the signal. Among several candidates for digitizing analog communication signals, ∆Σ modulation has recently drawn attention [8–11]. As depicted in Fig. 2, the ∆Σ modulation-based DRoF system does not require a digital-to-analog converter (DAC) in RUs because the signal reconstruction is easily performed by a filter, which in turn simplifies the RU structure. This advantage is considerably enhanced as the number of RUs increases.
However, the main drawback of the ∆Σ modulator-based link is the high sampling rate of the ∆Σ modulator. It is typically four times higher than the carrier frequency of the input signal and is thus known as the fs/4 band-pass ∆Σ modulator [12]. For example, for a 1 GHz RF signal, a sampling rate of 4 GHz is required for the fs/4 ∆Σ modulator, which is quite high, even with current CMOS technologies for implementation. Moreover, it becomes severe when the carrier frequency increases.
Consequently, numerous studies have aimed to decrease the high sampling frequency [13,14]. In [14], the sampling frequency can be reduced from 4 to 2.3 times the RF carrier frequency at the expense of the signal-to-noise ratio. However, although the required sampling frequency decreases down to almost half, it remains challenging. In addition, according to the Nyquist sampling theory, considerable improvement in further reducing the sampling frequency is limited and may not be viable.
Owing to this fundamental limitation, approaches have been developed that use a high frequency replica as a transmit signal rather than the fundamental signal [8]. Accordingly, the sampling frequency can be lower and even occur below the RF carrier frequency. However, because the magnitude of the replica signal is typically much smaller than that of the fundamental signal, the relative power ratio of the replica signal to the encoded pulse train significantly decreases, which leads to a low system power budget. This low budget directly affects not only the transmission distance, but also the number of supported RUs.
In this paper, two DRoF transmission systems based on a Manchester encoding-assisted ∆Σ modulator for sampling an analog RF signal are proposed and experimentally demonstrated. With the Manchester encoder, one of the high frequency replicas or images of the ∆Σ modulator output signal can be transmitted over the link with a high coding gain provided by the encoder. This is possible because, unlike a non-return-to-zero (NRZ) signaling scheme, the Manchester-encoded pulse has spectral maxima at high frequencies rather than direct current (DC). The advantage of adopting the modified Manchester code—termed N-pulse Manchester encoding in this paper—instead of the conventional Manchester code is significant.
To verify the effectiveness of the proposed DRoF transmission systems, a comparative demonstration is performed. The results of the conventional band-pass ∆Σ modulation-based DRoF links and those of the proposed N-pulse Manchester-encoding-based DRoF links are given based on a long-term evolution (LTE) 10 MHz signal at an 879 MHz carrier frequency (LTE frequency band 5 used by SK Telecom, the major mobile network service provider in South Korea).
The remainder of this paper is organized as follows. In Section 2, the N-pulse Manchester encoding scheme is introduced. Additionally, two different DRoF systems—herein termed type-I and type-II—are proposed with the modified Manchester encoder. In Section 3, the experimental setups and evaluation results of the proposed DRoF transmission systems are outlined and discussed. The conclusions are presented in Section 4.
2. Proposed ∆Σ modulated DRoF systems with multi-pulse Manchester encoding
Before introducing the proposed DRoF systems, N-pulse Manchester encoding is introduced first in Section 2.1. The focus here is on the spectral characteristics of the encoding scheme. Then, the proposed ∆Σ-digitized DRoF systems assisted by the N-pulse Manchester encoder are given in Section 2.2.
2.1 N-pulse Manchester encoding
In Manchester coding, input bits are encoded into directional transitions, such as a low to high level, or a high to low level, at the middle of each bit period. Two conventions exist for the directions of the transitions. In this paper, the input bit of ‘0’ and ‘1’ is expressed by a high-to-low transition and a low-to-high transition, respectively, according to the IEEE 802.3 convention.
Owing to the fact that a transition always exists for each bit period regardless of the input bit, the Manchester-coded signal has no DC components; rather, it contains spectral humps at certain frequencies above DC. Because the spectrum of the digital pulse train is significantly affected by the power spectral density (PSD) shapes, replica signals at higher frequencies, rather than the fundamental in the pulse train, can experience larger amplitude gains with the Manchester coding technique. They can then be used as a transmit signal.
To exploit an even higher-order replica signal for transmission, an integer multiple, N, of the Manchester pulse can be used to modulate each bit of the input pulse train, called N-pulse Manchester encoding in this paper. It enables the spectral peaks of the N-pulse Manchester encoded signal to transition to higher frequencies.
Figure 3 shows the encoding principle of the N-pulse Manchester code. fDATA is defined as the output rate of a digital data with NRZ signaling scheme, and fMAN ( = 2·fDATA) is the rate of a conventional Manchester code determined by the minimum pulse width. For the first data bit of ‘1’ in two-pulse Manchester encoding, two consecutive Manchester ‘1’ pulse waveforms are used for the bit encoding. In the same manner, two consecutive Manchester ‘0’ pulses are modulated in the case of the encoding for the input bit of ‘0.’ For the N-pulse case, N consecutive Manchester pulses are used in the encoding.
According to [15], the spectral shaping of the N-pulse Manchester pulse waveform is described, as follows:
where and N is the number of Manchester pulses for encoding one bit. The results of the equation are plotted with values of N = 1, 2, and 3 in Fig. 4, which depicts the spectral characteristics of the N-pulse Manchester encoding. Compared with the one-pulse Manchester waveform, which has the highest spectral peak at , the spectral maxima of the two-pulse and three-pulse Manchester waveforms occur at nearly two and three times the, respectively.
Fig. 4 Normalized power spectral densities of N-pulse Manchester encoding with N values of 1, 2, and 3.
2.2 Proposed Manchester-encoded DRoF transmission systems
Figure 5 shows one possible implementation of the proposed DRoF system based on the N-pulse Manchester code-assisted ∆Σ modulation. The signal modulation process in the proposed transmission system consists of two steps: ∆Σ modulation and Manchester encoding. At the CU in Fig. 5, the analog communication signal at an intermediate frequency (IF), fIF, is converted into a digital pulse train by the low-speed ∆Σ modulator.
The IF frequency, fIF, and the sampling rate of the ∆Σ modulator, fDSM, for a certain N value are chosen to enable one of the replica or image components in the ∆Σ modulator output to be located at the desired carrier frequency, fRF. At this point, the RF signal at fRF obtains a high coding gain by leveraging the spectral characteristics of the N-pulse Manchester encoder explained in Section 2.1. It is then fed into the electrical-to-optical (E/O) converter for transporting over the fiber, and transmitted from the antenna at the RU. The relationship among fIF, fDSM, fRF and fMAN can be described as the following equations. The relationship is expressed as follows for the case in which one of the replicas of the fundamental signal is chosen for transmission:
where . If the image component of one of the replicas is selected as a transmit signal, fRF isTherefore, as N increases, fDSM proportionally decreases for a desired fRF, thereby causing the Σ∆ modulator to operate at a lower sampling rate. This is a key advantage of the proposed DRoF system. However, the reduced fDSM deteriorates the signal to the quantization noise ratio (SQNR) of a ∆Σ modulator on account of the low oversampling ratio (OSR) which is defined as follows:where B is the bandwidth of the analog signal applied to the ∆Σ modulator. Therefore, for a minimum required OSR, OSRreq, the maximum N value, NMAX, can be formulated asandfor a high-frequency replica and the image of a high-frequency replica, respectively.At the RU, the digital pulse train carrying the RF signal information is reconstructed by the optical-to-electrical (O/E) converter and the clock and data recovery circuit. Then, the original analog communication signal at the RF frequency, fRF, is easily reconstructed by filtering the received digital pulse train.
In this way, the proposed DRoF system has several advantages over the conventional ∆Σ modulation-based DRoF system shown in Fig. 2. Firstly, because the ∆Σ modulator in the proposed system digitizes the input signal at a frequency that is much lower than the carrier frequency, a low sampling rate is required for the signal digitation. Therefore, a low-cost implementation of the modulator is possible with an inexpensive semiconductor process.
Secondly, a high-frequency mixer—where the IF signal, fIF, is frequency up-converted to the RF frequency, fRF—is not needed because the N-pulse Manchester performs the frequency up-convert operation with the multi-pulse codes. Finally, the optical transmission rate of the proposed DRoF link is likely to be lower than that of the conventional scheme. This is because the required optical transmission rate of the proposed DRoF system,, is roughly 2·fRF according to Eqs. (2) and (3), whereas the conventional fs/4 band-pass ∆Σ-digitized DRoF system requires 4·fRF.
However, disadvantages of Manchester encoding exist. The Manchester code has two well-known drawbacks. First, it requires a complex decoder circuit to reconstruct the original data from a Manchester-encoded signal. Second, the signal bandwidth after the encoding is extended. Fortunately, the first issue is not applicable to the proposed DRoF architecture because the desired RF signal at fRF is extracted not by the Manchester decoder, but by the band-pass filter. Thus, the complex decoder can be avoided in the proposed system. On the other hand, the second issue still remains, and high-speed E/O and O/E converters must be installed at both CU and RU to properly handle the encoded signal.
To mitigate the bandwidth problem, the proposed DRoF structure is slightly modified, as shown in Fig. 6. Here, the N-pulse Manchester encoder is located at RU, rather than CU. Because Manchester encoding is performed at RU, the optical transmission bandwidth decreases. Thus, the E/O and O/E converters with the clock and data recovery circuit operate at the rate of the Σ∆ modulator, which can be considerably reduced with a high N value according to Eqs. (2) and (3). This enables the high-frequency RF signal to be transmitted over the proposed DRoF system at a far lower rate than the RF carrier frequency. However, all these advantages are obtained at the expense of the local N-pulse Manchester encoder at RU. Therefore, disadvantages due to the additional block must be taken into account for the design of the proposed type-II transmission system. As previously noted, the former and latter DRoF systems are herein termed type-I and II. Their respective advantages and disadvantages are summarized in Table 1.
Table 1. Pros and cons of the conventional DRoF and the proposed type-I/II DRoF systems.
3. Experimental setups and results
The experimental evaluation results of the proposed type-I and type-II DRoF schemes are given in this section. For a realistic system evaluation, the LTE downlink signal using a frequency range of 874 to 884 MHz (LTE band 5), which is employed by SK Telecom’s network in South Korea, was used as an RF signal to be transmitted. For comparison purposes, the conventional band-pass ∆Σ modulator-based DRoF system, in which the RF signal is directly ∆Σ-digitized at CU and transferred to RU, was also developed and compared to the proposed DRoF systems.
3.1 Experimental setups
Figure 7 shows the experimental setup for the proposed DRoF systems and the conventional band-pass ∆Σ modulation-based DRoF system. An LTE 10 MHz signal with a modulation format of 64 quadrature amplitude modulator (QAM) was generated at an IF frequency of 70 MHz in the LTE signal generator. This is the most frequently used frequency for IF stages in communication systems. The peak-to-average power ratio (PAPR) of the modulation signal is 8.5 dB owing to the crest factor reduction (CFR) algorithm and at the expense of the error-vector magnitude (EVM) of 2.4%.
Fig. 7 Experimental setup for the conventional band-pass ∆Σ-based DRoF system and type-I/II DRoF systems.
Fourth-order band-pass ∆Σ modulators were designed in MATLAB to convert the LTE signals at fRF or fIF into digital bit streams at the rate of fDSM. For the type-I system, the output digital bits were encoded by the N-pulse Manchester encoder and then delivered to the pulse pattern generator (PPG) (Agilent E4862B), PPG #1, for optical transmission at the rate of. In the cases of the type-II and conventional links, the ∆Σ-digitized bit patterns were directly downloaded to the PPG #1 to be transported over the fiber at the rate of by the 2.5 Gbps E/O converter operating at a 1,310 nm wavelength. For the conventional system, the LTE IF signal was up-converted to 879 MHz before being applied to the high-frequency ∆Σ modulator.
The lightwave signals passing through the 10 km single-mode fiber, the variable optical attenuator and the coupler were translated into electrical bit streams by the 2.5 Gbps O/E converter. The received signals were then captured by an oscilloscope (Keysight DSO-S 204A) at a sampling rate of 10 GHz for the bit error rate (BER) evaluation of the optical transmission link and the operations of the clock-and-data recovery (CDR) and the N-pulse Manchester encoding. The spectra of the received signals were also measured by a spectrum analyzer (Keysight N9030A) for both spectral characteristic and signal-to-noise ratio (SNR) measurements.
The captured waveforms were then reconstructed as digital pulse trains by another PPG (Agilent E4862B), PPG #2, through the CDR for the conventional and the type-I schemes or the CDR/N-pulse Manchester encoder for type-II systems. With the CDR and the DAC (PPG #2), most of the distortions and noise present in the received digital signals at the O/E converter output can be removed out. As a result, clear rectangular pulse waveforms can be obtained at the PPG #2 output. This operation is important because the bit streams from the ∆Σ modulators at CU were generated based on an ideal rectangular pulse shape (more strictly, the pulse shape of the feedback DACs in the ∆Σ modulators), and any discrepancies between current pulse waveform and the ideal rectangular waveform decreased the SNR.
The output signals of the PPG #2 were then measured by the spectrum analyzer with vector signal analyzer (VSA) software filtering and down-converting the LTE 10 MHz signal at 897 MHz to baseband at a sampling rate of 30.72 MHz. The IQ signals sampled by the VSA were finally analyzed in terms of the error-vector magnitude (EVM) and constellation in the LTE signal analyzer in MATLAB.
Table 2 summarizes the important parameters used in the experiments for N = 1, 2, and 3 for the N-pulse Manchester encoder to transmit the LTE signal at 879 MHz. It should be noted that, in the experiments, the image signal of the Nth replica was transmitted instead of the Nth replica signal because it experiences a higher magnitude gain than the Nth replica signal through the N-pulse Manchester encoding process. In addition, as shown in Table 2, fIF, fDSM andfor N = 3 are selected as 70.5, 316.5, and 1,899 MHz, respectively, for simplicity and to relieve the resolution requirement of the clock generator for fDSM.
Table 2. Important parameters of the conventional band-pass ∆Σ-based DRoF and type-I/II DRoF transmission systems in the experimental setup.
3.2 Experimental results and discussions
The experimental results of the proposed type-I and type-II systems, and the conventional ∆Σ-based DRoF links with the setups described in Section 3.1, are given in this subsection. For a fair comparison, the band-pass ∆Σ-modulated signal in the conventional DRoF system was generated for the LTE signal at 879 MHz with the sampling clock of 1,898 MHz, which was only 2.16 times higher than the carrier frequency (close to the Nyquist rate). It was obtained by modifying the NTF of a conventional fs/4 band-pass ∆Σ modulator, whereas the conventional ∆Σ modulator required the clock frequency of 3,516 MHz ( = 4 × 879 MHz). Therefore, the transmission rate of the band-pass ∆Σ-modulated DRoF system was also 1,898 MHz, the same as that of the type-I DRoF link.
LTE 10 MHz signals (red) applied to the ∆Σ modulators and the corresponding ∆Σ-modulated bit patterns (blue) are shown in Fig. 8. Theses bit patterns are then downloaded to the PPG #1, modulated onto a lightwave by the E/O converter, and transmitted over the 10 km fiber to the O/E converter. It is commented that different time periods for Figs. 8(a)-8(d) and 8(e)-8(g) are used for better illustration. As shown in Fig. 8, on account of the fact that the LTE signal carrier frequencies applied to the ∆Σ modulators in the type-I and II DRoF systems are 70 MHz (N = 1 and 2) and 70.5 MHz (N = 3), the input signal waveforms to the ∆Σ modulators are much slower than that of the conventional DRoF using the LTE signal at 879 MHz. This makes the circuit implementation of the ∆Σ modulators much easier or possibly fully digital, which is a desirable feature for the next-generation fully-digital DRoF systems.
Fig. 8 LTE signal waveforms applied to ∆Σ modulators and their corresponding ∆Σ-digitized bit streams at CU in the conventional ∆Σ modulation-based DRoF and proposed type-I and II DRoF systems (N = 1, 2, and 3).
Measured BER curves of the DRoF systems versus the received optical power are depicted in Fig. 9(a). It should be noted that the BER values were calculated by comparing the received signal bit patterns after the optical receiver with the bit streams applied to the optical transmitter, not the data encoded in the LTE signal with QAM, which represents the quality of the optical transmission.
Fig. 9 Measured BER curves and EVM values of the proposed type-I/II DRoF systems and the conventional ∆Σ modulation-based DRoF link.
Because the optical transmission rates of the type-I system and the conventional ∆Σ-based DRoF system were all the same as 1,898 Mbps (1,899 Mbps for N = 3), the BER performances of the four systems were similar, as expected. However, the sampling clock frequency for the ∆Σ modulator in the type-I DRoF system proportionally decreased from the rate of 1,898 MHz by a factor of 2N on account of the N-pulse Manchester encoding. For example, in the case of N = 2, the clock frequency for the modulator was 474.5 MHz, which was four times lower than the rate of 1,898 MHz. This lower sampling frequency enables the ∆Σ modulator implementation to be considerably easier and more cost-effective.
Unlike the type-I systems, the optical transmission rates of the type-II systems are directly determined by the sampling frequencies of the ∆Σ modulators. Thus, better BER performance can be obtained with the lower transmission rates. As can be seen, the best BER performance was achieved with N = 3 for the optical transmission rate of 316.5 MHz, which was the lowest transmission rate in the experiments. Compared to the BER curve of the conventional ∆Σ-based DRoF link, the type-II system shows 3.3 dB, 5.3 dB, and 6.4 dB higher optic power margins for N = 1, 2, and 3, respectively.
The measured EVM results of the transmissions with respect to the received optical power levels and BER values of the optical link are also shown in Figs. 9(b) and 9(c), respectively. The minimum EVM value is limited to approximately 2.4% with the CFR algorithm, which is clearly observed in the measurement results. All EVM curves in the experiments satisfy the 3GPP signal quality requirement of EVM of less than 8% for 64 QAM. However, for EVM of 8%, the type-II systems show better power margins up to 2.7 dB, 4.4 dB, and 4.9 dB for N of 1, 2, and 3, respectively, than that of the conventional ∆Σ modulation-based link.
It was expected that the conventional DRoF system would demonstrate better signal quality performance because the OSR value for the ∆Σ modulator in the conventional DRoF link is 94.9. This is at least two times higher than the OSR values in the proposed type-I/II systems. However, as shown in Fig. 9(b), the resulted EVM values of the received LTE signals are almost identical for the same Prec of –35 dBm. This indicates that the major signal-quality limiting factor is not the quantization noise from the ∆Σ modulators. Furthermore, the OSR value of 94.9 for the conventional system is too high for the minimum EVM of 2.4%. Therefore, OSR for the ∆Σ modulation can be reduced without significantly deteriorating the signal quality in the EVM range.
To elucidate the spectral behavior of the proposed DRoF systems, the spectra over the systems were measured by a Keysight N9030A spectrum analyzer. Figure 10 shows the measured signal spectra at the O/E converter and the hybrid CDR (PPG #2) outputs at RU in the conventional ∆Σ-digitized and proposed type-I DRoF links, all at Prec of –35 dBm. It is evident that the spectral maxima of the signals are observed around, as expected according to Eq. (3). Therefore, the LTE image signals at 879 MHz in the type-I systems for N of 1 to 3 have the maximum gain among other LTE signals at fundamental, replica, or image frequencies, as designed.
Fig. 10 Signal spectra at the O/E converter and the hybrid CDR outputs (PPG #2) at RU in the conventional ∆Σ-based DRoF system and the proposed type-I DRoF systems (N = 1, 2, and 3).
It should be noted that, as depicted in Fig. 10(a) for the conventional DRoF link, spectral ringing was measured in the frequency range of DC to 500 MHz. It originated from the low stability margin for the feedback loop used in the ∆Σ modulator at the low sampling frequency (1,898 MHz) for the 879 MHz-centered LTE signal, giving the ratio of 2.16, which is too close to the Nyquist rate. Therefore, to deploy a real RoF system based on the conventional ∆Σ modulator, the clock rate and optical transmission rate must increase to relieve the potential stability problem. This is another reason why the type-I DRoF system may require a lower sampling frequency for the ∆Σ modulator than the conventional system.
To determine the amount of noise near the signal band at 879 MHz, the same signal spectra in a magnified view for the 879 MHz frequency range are shown in Fig. 10(e). As expected, the amount of noise around the signal frequency range increases with high N values for the N-pulse Manchester encoder. This is because OSR for the ∆Σ modulators in the proposed DRoF systems decreases as N increases. However, the increased noise level for N = 1 to 3 does not considerably affect the best EVM performance (~2.4%) of the LTE signal used in the experiments. Furthermore, because the current matured RF surface acoustic wave (SAW) filter provides an out-of-band rejection ratio that is larger than 30 dB, it is easy to obtain a sufficiently high rejection in the stop bands [16]. Consequently, the value of N for the multi-pulse Manchester encoder can be treated as a system design parameter; moreover, it can be chosen with consideration of the stop band attenuation required for a specific DRoF transmission system.
For the type-II DRoF links, the spectra of the received signals at the O/E converter and the hybrid CDR/N-pulse Manchester encoder (PPG #2) outputs at RU are shown in Fig. 11. It is evident that the spectra of the received signals are transformed according to the spectral shapes of the pulse waveforms for the N-pulse Manchester encoder with N of 1 to 3. Thus, among the fundamental, replica, or image components, the maximum gains are observed at the LTE signals at 879 MHz, similar to the type-I system. The measured SNR values of the LTE signals at 879 MHz for N of 1, 2, and 3 were 45.4 dB, 35.3 dB and 30.9 dB, respectively.
Fig. 11 Signal spectra at the O/E converter and the hybrid CDR outputs (PPG #2) at RU in the proposed type-II DRoF systems (N = 1, 2, and 3).
To determine the extent of SNR improvement obtained by the CDR for type-I and the CDR/N-pulse Manchester encoder for type-II, the measured SNRs of LTE signals at 879 MHz at the O/E converter and the hybrid CDR/N-pulse Manchester encoder (PPG #2) outputs at RUs in the DRoF transmission systems are depicted in Fig. 12. As shown in Figs. 12(a)-12(d), for the conventional and type-I DRoF links, a considerable amount of pulse recovery (PR) gains was achieved. The PR gain is defined here as the amount of increase in SNR by the CDR. At Prec of –35 dBm, PR gains of 21.7, 19.5, 13.0 and 8.9 dB were measured for the conventional and type-I with N of 1, 2 and 3, respectively. For the type-II systems, the improvement in the SNR of the 879 MHz-centered LTE signal is even higher because N-pulse Manchester coding gain is added to the PR gain. The measured gains in SNR by both PR and N-pulse Manchester encoder were 41.8 and 33.0 dB with N of 1 and 2, respectively. For N = 3, the SNR of the LTE signal at 879 MHz at the O/E converter output was not detectable. The measured SNR values of the LTE signal at 70 MHz at the O/E converter output were also plotted.
Fig. 12 Measured SNRs of LTE signals at 879 MHz at the O/E converter and the hybrid CDR/N-pulse Manchester encoder outputs with respect to received optical power levels.
3.3 Experimental evaluation of the jitter effects on the type-I and II DRoF systems
As illustrated in Fig. 8, the N-pulse Manchester-encoded digital signals have more transitions in a given time period than the NRZ-based conventional DRoF system. Because of the property, the SNR of an N-pulse Manchester encoded signal is more sensitive to jitter. This becomes even worse as N for the N-pulse Manchester encoder increases. Therefore, it is important to determine how much jitter noise is acceptable for the proposed type-I and II DRoF systems.
Figure 13 shows the experimental setup for the evaluation of the jitter sensitivity. N-pulse Manchester-encoded LTE signals were downloaded the PPG, and applied to the spectrum analyzer (Keysight N9030A) and the oscilloscope-based jitter analyzer (Agilent DSO81304A) for SNR and jitter measurements, respectively. To control the amount of the jitter at the PPG, an additive white Gaussian noise generator (AWGN) (Agilent 33250A) was used to provide a delay control signal for the PPG.
Fig. 13 Experimental setup for the evaluation of the jitter effects on the proposed type-I and II DRoF systems.
The measured SNR values with respect to the RMS jitter and the measured spectra for the proposed DRoF systems are depicted in Fig. 14. As shown in Fig. 14(a), the N-pulse Manchester encoded signals were more sensitive to jitter noise, and the sensitivity increased as N value increased. Therefore, for a given RMS jitter noise, the maximum SNR of the N-pulse Manchester encoded signal decreased with a higher N parameter. For a small amount of jitter noise, the SNR was limited by the quantization noise generated from the ∆Σ modulator. From the SNR results for N of 1 to 3, it can be argued that the RMS jitter for the proposed DRoF systems at 1,898 Mbps must be less than 2.5 ps. According to previous literature on CDR circuits, the jitter requirement is acceptable for current circuit technology [17, 18].
Fig. 14 Measured SNRs with respect to the RMS jitter and the measured spectra for the proposed DRoF systems with N of 1 to 3.
4. Conclusion
In this paper, two ∆Σ-modulated DRoF transmission systems using multi-pulse Manchester encoding—termed type-I and type-II schemes—were proposed and experimentally demonstrated. The LTE 10 MHz downlink signal at 879 MHz (LTE band 5), which is used for LTE service in South Korea, was digitized and transmitted at the rate of 1,898 Mbps for the type-I system (1,899 Mbps for N = 3) and 949, 474.5, and 316.5 Mbps for the type-II systems. N parameters of 1, 2, and 3 were used for the multi-pulse Manchester encoder to provide high coding gains for image or replica signals. The transmitted signals were reconstructed and analyzed in terms of spectra, SNR, BER, and EVM at RU. For comparison purposes, a conventional ∆Σ modulator-based DRoF system was also implemented and evaluated with the same LTE signal.
Compared to the conventional DRoF system, the usage of the low-sampling-rate ∆Σ modulator for the LTE signal at 70 MHz (879 MHz for the conventional system) was possible. Moreover, no high-frequency up-converting mixer was required because the 70 MHz-centered LTE signal was directly ∆Σ-digitized in the proposed type-I and II DRoF systems. Furthermore, the type-II link provided an additional optical power margin up to 4.9 dB for N = 3 for the EVM requirement of 8% for the LTE 64-QAM signal.
Based on the above results, the proposed DRoF schemes enhance the reliability and flexibility of RoF transmission systems with higher power margins, or they make the system implementation more cost-effective and easier to conduct with low-frequency ∆Σ modulators. In particular, the lower-bandwidth requirement of electronics and optical transceivers used in the proposed DRoF systems is clearly more favorable to the next-generation all-digital DRoF systems.
Acknowledgment
This work was supported by the ICT R&D program of MSIP/IITP. [No. R0101-16-244, Development of 5G Mobile Communication Technologies for Hyper-connected Smart Services]
References and links
1. Y. Xu, X. Li, J. Yu, and G. K. Chang, “Simple and reconfigured single-sideband OFDM RoF system,” Opt. Express 24(20), 22830–22835 (2016). [CrossRef] [PubMed]
2. H. Y. Wang, Y. C. Chi, and G. R. Lin, “Remote beating of parallel or orthogonally polarized dual-wavelength optical carriers for 5G millimeter-wave radio-over-fiber link,” Opt. Express 24(16), 17654–17669 (2016). [CrossRef] [PubMed]
3. B. Wu, M. Zhu, J. Zhang, J. Wang, M. Xu, F. Yan, S. Jian, and G. K. Chang, “Multi-service RoF links with colorless upstream transmission based on orthogonal phase-correlated modulation,” Opt. Express 23(14), 18323–18329 (2015). [CrossRef] [PubMed]
4. R. Zhu, X. Zhang, D. Shen, and Y. Zhang, “Ultra broadband predistortion circuit for radio-over-fiber transmission systems,” J. Lightwave Technol. 34(22), 5137–5145 (2016). [CrossRef]
5. M. Sung, C. Han, S. H. Cho, H. S. Chung, and J. H. Lee, “Improvement of the transmission performance in multi-IF-over-fiber mobile fronthaul by using tone-reservation technique,” Opt. Express 23(23), 29615–29624 (2015). [CrossRef] [PubMed]
6. P. A. Gamage, A. Nirmalathas, C. Lim, D. Novak, and R. Waterhouse, “Design and analysis of digitized RF-over-fiber links,” J. Lightwave Technol. 27(12), 2052–2061 (2009). [CrossRef]
7. A. Nirmalathas, P. A. Gamage, C. Lim, D. Novak, and R. Waterhouse, “Digitized radio-over-fiber technologies for converged optical wireless access network,” J. Lightwave Technol. 28(16), 2366–2375 (2010). [CrossRef]
8. L. M. Pessoa, J. S. Tavares, D. Coelho, and H. M. Salgado, “Experimental evaluation of a digitized fiber-wireless system employing sigma delta modulation,” Opt. Express 22(14), 17508–17523 (2014). [CrossRef] [PubMed]
9. S. Jang, G. Jo, J. Jung, B. Park, and S. Hong, “A digitized IF-over-fiber transmission based on low-pass delta-sigma modulation,” IEEE Photonics Technol. Lett. 26(24), 2484–2487 (2014). [CrossRef]
10. J. Wang, Z. Yu, K. Ying, J. Zhang, F. Lu, M. Xu, and G. K. Chang, “Delta-sigma modulation for digital mobile fronthaul enabling carrier aggregation of 32 4G-LTE/30 5G-FBMC signals in a single-λ 10-Gb/s IM-DD channel,” in Optical Fiber Communication Conference (Optical Society of America, 2016), paper W1H–2. [CrossRef]
11. J. Wang, Z. Yu, K. Ying, J. Zhang, F. Lu, M. Xu, L. Cheng, X. Ma, and G. K. Chang, “10-Gbaud OOK/PAM4 digital mobile fronthaul based on one-bit/two-bit Δ-Σ modulation supporting carrier aggregation of 32 LTE-A signals with up to 256 and 1024QAM,” in 42nd European Conference on Optical Communication (Optical Society of America, 2016), pp. 1–3.
12. B. K. Thandri and J. Silva-Martinez, “A 63 dB SNR, 75-mW bandpass RF Σ∆ ADC at 950 MHz using 3.8-GHz clock in 0.25-μm SiGe BiCMOS technology,” IEEE J. Solid-State Circuits 42(2), 269–279 (2007). [CrossRef]
13. B. Park, S. Jang, P. Ostrovskyy, and J. Jung, “A high voltage swing dual-band bandpass modulator for mobile base-station,” IEEE Microw. Wirel. Compon. Lett. 23(4), 199–201 (2013). [CrossRef]
14. P. Ostrovskyy, H. Gustat, M. Ortmanns, and J. C. Scheytt, “A 5-Gb/s 2.1–2.2-GHz bandpass modulator for switch-mode power amplifier,” IEEE Trans. Microw. Theory Tech. 60(8), 2524–2531 (2012). [CrossRef]
15. T. Johnson, R. Sobot, and S. Stapleton, “Manchester encoded bandpass sigma-delta modulation for RF class D amplifiers,” IET Circuits Dev. Syst. 1(1), 21–26 (2007). [CrossRef]
16. G. Endoh, M. Ueda, O. Kawachi, and Y. Fujiwara, “High performance balanced type SAW filters in the range of 900 MHz and 1.9 GHz,” in IEEE Ultrasonics Symposium Proceedings (IEEE, 1997), pp. 41–44. [CrossRef]
17. M. K. Raja, D. L. Yan, and A. B. Ajjikuttira, “A 1.4-psec Jitter 2.5-Gb/s CDR with wide acquisition range in 0.18-μm CMOS,” in ESSCIRC (2007), pp. 524–527.
18. Maxim Integrated, “Multirate Clock and Data Recovery with Limiting Amplifier,” MAX3872 datasheet, Rev 3, 2/07.
