Experimental Generation of Reconfigurable Ultra-Wideband Signal Based on the Pulse Difference by Using Optical Polarization Synthesis

A reconfigurable optical ultra-wideband (UWB) pulse generation scheme based on the pulse difference by using polarization synthesis is proposed and demonstrated experimentally. The Gaussian pulses with an appropriate time delay are modulated via the two sub-MZMs of the DPol-MZM, and the optically carried signals are combined via orthogonal polarization multiplexing, avoiding the optical interference. By setting the DC bias voltages of the sub-MZMs and the time delay, a pair of polarity-reversed Gaussian monocycle, doublet, triplet and quadruplet signals are generated with the central frequency of 3.37, 4.79, 4.5, 5.05 GHz and the fractional bandwidths of 166%, 129%, 124%, 89% accordingly. The signals of pulse shape modulation, binary phase shift keying modulation and on-off keying modulation are generated with the experimental system.

SECTION I. Introduction

The ultra-wideband (UWB) pulse signal has recently drawn extensive attention in the radar, precision navigation, sensor networks and wireless communication, due to its unique characteristics, such as low power consumption, large bandwidth, immunity to multi-path fading and so on. The US Federal Communications Commission (FCC) has approved the commercial use of UWB technology and defined the signal with −10 dB bandwidth over 500 MHz or fractional bandwidth (−10 dB bandwidth divided by central frequency) over 20% as the UWB signal [1]. For UWB system, high-order Gaussian pulse exhibits good performance for UWB signal generation, because the low-frequency component can be restrained much with the increase of the Gaussian pulse order. It has a higher spectral efficiency, and can avoid the interference to other low-frequency communication systems, such as the mobile communication, the GPS system and so on [2], [3].

The UWB signal generation by optical techniques can overcome the bandwidth restriction of electrical techniques and has the advantage of integration with the passive optical network and wireless access network [4], [5]. In recent years, plenty of optical schemes of the UWB signal generation have been proposed and developed. One of the common methods is based on the phase to intensity modulation conversion. The principle is to generate the UWB monocycle or doublet signals through locating the optical carrier modulated by Gaussian pulse, at the linear or quadrature region of the optical filter response spectrum. For example, the fiber Bragg grating (FBG) [6], [7], Sagnac interferometer comb filter [8] and optical bandpass filter [9]–​[11] have been demonstrated to generate the UWB signal. The UWB signal can also be obtained through optical spectrum shaper and frequency-to-time conversion. The FCC-compliant power efficient pulse shape is written in the frequency domain by using the spectral shaper such as FBG [12]–​[14] or tunable optical filters [15], [16], and the frequency-to-time conversion is completed by the single-mode fiber or other dispersion devices. Moreover, non-linear optical effects in some optical devices such as the cross gain modulation [17], [18], cross phase modulation [19], gain saturation effect in a semiconductor optical amplifier [20], and stimulated Brillouin scattering [21] have been explored to generate UWB signals. In addition, many schemes based on the delayed combination of pulses with different waveforms to generate the UWB signals have been proposed. The process of the delayed combination of pulses is equivalent to an operation of a first- or multi-order difference. The dual-drive Mach-Zehnder modulator (DD-MZM) [22] and dual-parallel Mach-Zehnder modulator (DP-MZM) [23], [24] have been proposed to generate the UWB signals. But the optical interference may affect the performance of the generated UWB signals.

In this paper, a reconfigurable optical UWB pulse generation scheme based on the pulse difference with polarization synthesis is proposed and demonstrated by simulation and experiment. Herein, only two branches of Gaussian pulses are needed, which is more concise and integrated. And the distinguished feature of this scheme is that the orthogonal polarization multiplexing is utilized to combine the optically carried signals from the two sub-MZMs in DPol-MZM to avoid the optical interference. The polarity-reversed Gaussian monocycle, doublet, triplet, and quadruplet signals are generated by setting the DC bias voltage of DPol-MZM and the time delay. In addition, the signals of pulse shape modulation (PSM), binary phase shift keying modulation (BPSK) and on-off keying (OOK) modulation are also measured by encoding the input electrical Gaussian pulses.

SECTION II. Operation Principle

Fig. 1 shows the schematic diagram of the UWB signal generation system. The modulation unit is located in the central station. Two electrical Gaussian pulses generated by the arbitrary waveform generator (AWG) with a relative time delay of τ₁ are amplified to suitable amplitude by the electronic amplifiers (EA₁ and EA₂) and input to the Dual Polarization Mach–Zehnder Modulator (DPol-MZM). In the DPol-MZM, the Gaussian pulses are modulated to the optical carrier from the laser diode (LD) via MZM₁ and MZM₂ with the adjustable states by DC bias voltages (V_DC1 and V_DC2), respectively. The optically carried signal from MZM₂ is rotated of 90° by the polarization rotator (PR) and combined with the optically carried signal from MZM₁ via the polarization beam combiner (PBC), forming a pair of orthogonal signals. The orthogonal polarization multiplexing avoids the interference between the two optical signals with the same frequency. The orthogonally multiplexed signals are transmitted to the base station through the optical fiber. In the base station, the optical carried signals are divided into two branches through the adjustable optical coupler (AOC), and the power coupling coefficients of the upper and lower branch are α and 1-α (0<α≤1). A time delay τ₂ is introduced to the lower branch by a tunable optical time delay line (TODL). After the differential detection upon the BPD, the UWB signal is obtained in the electrical domain.

Fig. 1.

The schematic diagram of the reconfigurable UWB signal generation system.

In the upper and lower branches of DPol-MZM, the optically carried signals before the AOC can be expressed as $$\begin{align*} {E_x}{\rm{(t)}} =& \frac{{{\rm{ }}\sqrt {2} }}{{4}}{E_{0}}\exp (j{\omega _0}t)\left\{ {\exp \left[ {j\frac{\pi }{{{2}{V_\pi }}}{g_1}(t)} \right]} \right.\\ &\left. { + \exp \left[ { - j\frac{\pi }{{{2}{V_\pi }}}{g_1}(t) - j\frac{\pi }{{{V_\pi }}}{V_{DC1}}} \right]} \right\} \tag{1}\\ {E_y}{\rm{(t)}} =& \frac{{{\rm{ }}\sqrt {2} }}{{4}}{E_{0}}\exp (j{\omega _0}t)\left\{ {\exp \left[ {j\frac{\pi }{{{2}{V_\pi }}}{g_{2}}(t - {\tau _1})} \right]} \right.\\ &\left. {{\rm{ }} + \exp \left[ { - j\frac{\pi }{{{2}{V_\pi }}}{g_{2}}(t - {\tau _1}) - j\frac{\pi }{{{V_\pi }}}{V_{DC2}}} \right]} \right\} \tag{2} \end{align*}$$ where the subscript of y denotes the 90° rotation of polarization state for the signal from MZM₂. E₀ and ω₀ are the amplitude and angular frequency of the light wave, g₁(t) and g₂(t-τ₁) are the Gaussian signals driving MZM₁ and MZM₂ respectively; V_DC1 and V_DC2 are the bias voltages of MZM₁ and MZM₂; V_π represents the half wave voltage of the two MZMs; and τ₁ is the time delay between the two Gaussian pulses. Then the output signal from DPol-MZM is $$\begin{equation*} E(t) = {E_x}(t) + {E_y}(t) \tag{3} \end{equation*}$$

The optical carried signal is divided into upper and lower branches by the AOC, the output of PD₁ can be written as $$\begin{equation*} {i_{PD1}}(t) = \alpha r\left[ {|{E_x}(t){|^2} + |{E_y}(t){|^2}} \right] \tag{4} \end{equation*}$$

The r is the responsivity of PD₁. To illustrate the generated waveform, i_PD1(t) is calculated as $$\begin{align*} {i_{P{D_1}}}(t) \propto & \alpha \left\{ \exp \left[ {j\frac{\pi }{{{V_\pi }}}{g_1}(t) + j\frac{\pi }{{{V_\pi }}}{V_{DC1}}} \right] \right.\\ &\left. + \exp \left[ { - j\frac{\pi }{{{V_\pi }}}{g_1}(t) - j\frac{\pi }{{{V_\pi }}}{V_{DC1}}} \right] \right\}\\ &+ \alpha \left\{ \exp \left[ {j\frac{\pi }{{{V_\pi }}}{g_{2}}(t{\rm{ - }}{\tau _{1}}) + j\frac{\pi }{{{V_\pi }}}{V_{DC{2}}}} \right] \right.\\ &\left.+ \exp \left[ { - j\frac{\pi }{{{V_\pi }}}{g_{2}}(t{\rm{ - }}{\tau _{1}}) - j\frac{\pi }{{{V_\pi }}}{V_{DC{2}}}} \right] \right\}\\ \propto & \alpha \left\{ \cos \left[ {\frac{\pi }{{{V_\pi }}}{g_{1}}(t) + \frac{\pi }{{{V_\pi }}}{V_{DC1}}} \right] \right.\\ &\left.+\cos \left[ {\frac{\pi }{{{V_\pi }}}{g_{2}}(t{\rm{ - }}{\tau _{1}}) + \frac{\pi }{{{V_\pi }}}{V_{\mathrm{DC2}}}} \right] \right\} \tag{5} \end{align*}$$

The time delay τ₂ is introduced before PD₂. Therefore, the output of the PD₂ can be expressed as $$\begin{align*} {i_{P{D_2}}}(t) \propto & {\rm{(1 - }}\alpha {\rm{)}} \cdot \left\{ {\cos \left[ {\frac{\pi }{{{V_\pi }}}{g_{1}}(t{\rm{ - }}{\tau _2}) + \frac{\pi }{{{V_\pi }}}{V_{DC1}}} \right]} \right.\\ &\left. + \cos \left[ {\frac{\pi }{{{V_\pi }}}{g_{2}}(t{\rm{ - }}{\tau _1}{\rm{ - }}{\tau _2}) + \frac{\pi }{{{V_\pi }}}{V_{\mathrm{DC2}}}} \right] \right\} \tag{6} \end{align*}$$

After the differential combination, the output of the BPD can be written as $$\begin{equation*} {i_{BPD}}(t) = {i_{P{D_{1}}}}(t) - {i_{P{D_{2}}}}(t - {\tau _2}) \tag{7} \end{equation*}$$

Therefore, by properly setting the DC bias voltages, V_DC1, V_DC2, the time delay τ₁, τ₂, and the AOC, the polarity-reversed Gaussian monocycle, doublet, triplet and quadruplet signal can be generated as shown in Fig. 2. Firstly, the Gaussian pulse is modulated on the optical carrier via MZM₁ and V_DC1 is set to the positive linear region of MZM response function. The power coupling coefficient α is set to 1. Then the positive Gaussian pulse is generated from PD₁ as shown in Fig. 2(a). If MZM₁ is set to the positive second order differential region by V_DC1, the positive Gaussian doublet pulse is generated as shown in Fig. 2(b). Similarly, the negative Gaussian pulse and Gaussian doublet pulse can be obtained in the lower branch by adjusting V_DC2.

Fig. 2.

Principle of the reconfigurable UWB puls generator.

Then an appropriate time delay τ₁ is introduced between the two Gaussian pulses with opposite polarity, the Gaussian monocycle pulse is generated as shown in Fig. 2(c), which is equivalent to the operation of a first-order difference. Similarly, an appropriate time delay τ₁ is introduced between the two Gaussian doublet pulses with opposite polarity, and the Gaussian triplet pulse is generated as shown in Fig. 2(d). After that, the power coupling coefficient α is set to 0.5, the Gaussian triplet pulse is split into two optical paths and launched into the two ports of the BPD. The time delay τ₂ is introduced between the two branches. After the differential combination, the Gaussian quadruplet signal is generated as shown in Fig. 2(e).

SECTION III. Simulation Results

In this part, a simulation of the UWB signal generation is carried out to verify the proposed scheme. The DPol-MZM in the simulation system is composed of two parallel MZMs, a 90° PR and a PBC. The half-wave voltage of MZM is set to 8 V and the DC bias voltages are adjusted according to the operation principle shown in Fig. 2. The bit sequence is “00100” and the bit rate is 5 Gbit/s. The Gaussian pulse width is 0.5 bit and the amplitude of pulse is 4 V.

Firstly, the Gaussian pulse is modulated on the optical carrier by the MZM₁ and MZM₂ with V_DC1 = 10 V and V_DC2 = 2 V, which are respectively the positive and negative linear region of MZM response function. A time delay τ₁ of 0.05 ns is introduced between the two branches. The power coupling coefficient α is set to 1. As a result, the positive Gaussian monocycle pulse is generated from PD₁ as shown in Fig. 3(a). Secondly, the Gaussian pulse is only modulated on the upper branch by MZM₁ with V_DC1 of 0.65 V, which is the positive second order differential region of MZM response function. So, the positive Gaussian doublet pulse is generated as shown in Fig. 3(c). As for the triplet pulse, the Gaussian pulse is modulated by both MZM₁ and MZM₂ with V_DC1 = 0.65 V and V_DC2 = 8.65 V, which are respectively the positive and negative second order differential region of MZM response function. By introducing the time delay τ₁ of 0.05 ns, the Gaussian triplet pulse is generated as shown in Fig. 3(e). In order to obtain the Gaussian quadruplet pulse, the power coupling coefficient α is set to 0.5. The positive Gaussian triplet signal is split into two optical paths by the AOC. And the time delay τ₂ between the two ports of BPD is 0.05 ns. Finally, the Gaussian quadruplet pulse is generated as shown in Fig. 3(g).

Fig. 3.

Simulation waveforms of (a) Positive Gaussian monocycle pulse and (b) Its spectrum, (c) Positive Gaussian doublet pulse and (d) Its spectrum, (e) Positive Gaussian triplet pulse and (f) Its spectrum, (g) Positive Gaussian quadruplet pulse and (h) Its spectrum, respectively.

Fig. 3(b), (d), (f) and (h) are the spectrums corresponding to Fig. 3(a), (c), (e) and (g), respectively. It can be seen that, the low frequency component (<2 GHz) of the spectrum is much restrained with the increase of UWB signal order. The measured results of the central frequency, −10 dB bandwidths and fractional bandwidths are summarized in Table I. The central frequency increases gradually from 3.75 GHz to 4.5 GHz, 5.5 GHz and 6 GHz, while the −10 dB bandwidths keep almost the same, which means that the power spectrum moves to higher frequency.

TABLE I Simulation Results of the Generated UWB Signals

SECTION IV. Experimental Results and Discussion

Base on the above analysis and the simulation, the experiment is carried out to verify the feasibility of the reconfigurable UWB signal generation. The established experimental system is shown in Fig. 4. The 1550 nm optical carrier with a power of 17.35 dBm is emitted from a distributed feedback (DFB 1782A-NM-100-33-FC-PM) laser and sent to the DPol-MZM (FTM 7980EDA). The two branches of Gaussian pulses with the full width half maximum (FWHM) of 0.1 ns are generated from the AWG (KWYSIGHT M8196A) and amplified to 4 V (peak to peak value) by two electrical amplifiers. Then the amplified Gaussian pulse is modulated on the optical carrier via MZM₁ and MZM₂, with the half wave voltage of 8 V. The optical carried signal is divided into two branches through the AOC (PMFC15501EB1112) and a time delay τ₂ is introduced by a tunable optical time delay line (General Photonics MDL-002) in the lower branch. After the differential detection upon the BPD (BPD V2150R), the UWB signal is obtained. The temporal output is measured by the real-time oscilloscope (KEYSIGHT DSOZ592A) and the spectrum is analyzed by the spectrum analyzer (Agilent E4440A).

Fig. 4.

The experimental setup of the reconfigurable UWB pulse generation system.

Firstly, by setting the following parameters of V_DC1 = 10 V, V_DC2 = 12 V, τ₁ = 0.1 ns and α = 1, the positive Gaussian monocycle pulse is generated from the established setup as shown in Fig. 5(a). Then V_DC1 and V_DC2 are set to 17.5 V and 19 V, corresponding to the negative and positive linear region of MZM response function and the negative Gaussian monocycle pulse is obtained as shown in Fig. 5(c). In Fig. 5(b) and (d), the central frequency is at 3.37 GHz and the −10 dB bandwidth is 5.6 GHz, resulting in the fractional band width of 166%.

Fig. 5.

Measured waveforms of (a) Positive Gaussian monocycle pulse and (b) Its spectrum, (c) Negative Gaussian monocycle pulse and (d) Its spectrum, respectively.

Then the Gaussian pulse is only modulated on the optical carrier via the MZM₁ with V_DC1 = 15.2 V. The generated positive Gaussian doublet pulse is shown in Fig. 6(a). Similarly, when V_DC1 is 7.2 V, the negative Gaussian doublet pulse is obtained as shown in Fig. 6(c). From Fig. 6(b) and (d), the central frequency is at 4.79 GHz and the −10 dB bandwidth is 6.17 GHz, resulting in the fractional band width of 129%.

Fig. 6.

Measured waveforms of (a) Positive Gaussian doublet pulse and (b) Its spectrum, (c) Negative Gaussian doublet pulse and (d) Its spectrum, respectively.

As for the Gaussian triplet signal, the two Gaussian pulses are modulated on the optical carrier via MZM₁ and MZM₂, respectively. When V_DC1 is 15 V, V_DC2 is 16.4 V and τ₁ is 0.1 ns, the positive Gaussian triplet pulse is obtained as shown in Fig. 7(a). Similarly, when V_DC1 is 6.5 V and V_DC2 is 7.8 V, the negative Gaussian triplet pulse is generated as shown in Fig. 7(c). In the spectrums of Fig. 7(b) and (d), the central frequency is at 4.5 GHz and the −10 dB bandwidth is 5.6 GHz, resulting in the fractional band width of 124%.

Fig. 7.

Measured waveforms of (a) Positive Gaussian triplet pulses and (b) Its spectrum, (c) Negative Gaussian triplet pulses and (d) Its spectrum, respectively.

The Gaussian quadruplet signal is obtained by synthesizing the triplet signal. When the power coupling coefficient α is set to 0.5, the positive Gaussian triplet signal is split into two branches and sent to PD₁ and PD₂ in the BPD. The time delay τ₂ is set to 0.2 ns. Then, the positive Gaussian quadruplet pulse is generated as shown in Fig. 8(a). Similarly, the negative Gaussian quadruplet pulse is obtained as shown in Fig. 8(c) when the Gaussian triplet pulses is negative. And from Fig. 8(b) and (d), the central frequency is at 5.05 GHz and the −10 dB bandwidth is 4.5 GHz, resulting in the fractional band width is 89%.

Fig. 8.

Measured waveforms of (a) Positive Gaussian quadruplet pulse and (b) Its spectrum, (c) Negative Gaussian quadruplet pulse and (d) Its spectrum, respectively.

Table II presents the measured results of the generated UWB signals. The experimental results are consistent basically with the simulation ones. The low frequency component of the spectrum is much restrained and the power spectrum moves to higher frequency with the increase of UWB signal order. There is a slight difference between the measured central frequency and the simulated one. In the simulation, the central frequency of Gaussian doublet pulse is lower than that of the Gaussian triplet pulse. In the experiment, the central frequency of Gaussian doublet pulse is higher than the one of the Gaussian triple pulse. The reason is that the center frequency of UWB signal is related to the Gaussian pulse order and the pulse width. The Gaussian triplet pulse is generated by the combination of polarity-reversed Gaussian doublet pulses. Therefore, the central frequency component of the triplet pulse is related to τ₁, and the central frequency will move to higher frequency with a smaller time delay τ₁. Besides, the measured −10 dB bandwidth and the fractional bandwidth in experiment are a little different from the simulation. The reason is that the width of the input Gaussian pulse in the experiment is not completely same as the simulation. The pulse width is set by the bit rate in the simulation, but it is set by the time of the rising edge and falling edge in the experiment. The width of generated UWB pulse is determined by the width of input Gaussian pulses and the time delay of τ1 and τ2. According to the relationship between frequency domain and time domain, the narrower pulse corresponds to the wider spectrum. So, the measured central frequency and the −10 dB bandwidth are not quit the same as the simulation.

TABLE II Experimental Results of the Generated UWB Signals

Generally, pulse shape modulation (PSM), binary phase shift keying modulation (BPSK) and on-off keying modulation (OOK) can be utilized in the ultra-wideband communication system, for example the high-speed wireless personal-area networks (WPAN) communications, distance measurement and positioning functions [25], [26]. Here, these typical modulation formats are experimentally generated by encoding the input electrical Gaussian pulses. For the PSM generation, the power coupling coefficient α is set to 1. V_DC1 and V_DC2 are set to the positive and negative second order differential region of MZM response function respectively. The upper branch Gaussian pulse is encoded as “11111”, and the lower branch Gaussian pulse is encoded as “10101”. A time delay τ₁ of 0.1 ns is introduced between the two branches. After the combination upon the BPD, the Gaussian triplet pulse is encoded for data “1” and the Gaussian doublet pulse is encoded for data “0”, with the PSM signal generation of “10101” as shown in Fig. 9(a). For the BPSK generation, the upper branch Gaussian pulse is encoded as “10101”, and the lower branch Gaussian pulse is encoded as “01010”. Then the positive Gaussian doublet pulse is encoded for data “1” and the negative Gaussian doublet pulse is encoded for data “0”, with the BPSK signal generation of “10101” as shown in Fig. 9(b). For the OOK generation, the power coupling coefficient α is changed to 0.5, and a time delay τ₂ of 0.2 ns is introduced between the two branches. The upper branch Gaussian pulse is encoded as “10101”, and the lower branch Gaussian pulse is encoded as “10101”. The Gaussian quadruplet pulse is encoded for data “1” and transmitting nothing is encoded for data “0”, with the OOK signal generation of “10101” as shown in Fig. 9(c). For the proposed UWB signal generation scheme is applied in the ultra-wideband communication system. As shown in Fig. 1, the modulation unit is located in the central station and a certain length fiber is utilized to connect the central station and the base station where the UWB signal is generated upon the photodetector. The chromatic dispersion of the optical fiber may influence the performance of the generated UWB signal due to its relatively wide bandwidth. To overcome this problem, the non-zero dispersion-shifted fiber (NZDF) can be utilized to distribute the UWB signal to the base station.

Fig. 9.

UWB pulses encoded with “10101” for (a) PSM, (b) BPSK and (c) OOK.

SECTION V. Conclusion

In this paper, a scheme of reconfigurable optical UWB pulse generation is proposed and demonstrated by simulation and experiment. In this scheme, the two sub-MZMs in DPol-MZM are driven by the Gaussian pulses with an appropriate time delay, and the optically carried signals are combined via orthogonal polarization multiplexing, avoiding the optical interference. By properly setting the DC bias voltage of DPol-MZM and the time delay of τ₁ and τ₂, a pair of polarity-reversed Gaussian monocycle, doublet, triplet, and quadruplet signals with the central frequency of 3.37, 4.79, 4.5, 5.05 GHz and fractional bandwidths of 166%, 129%, 124%, 89% are successfully generated in the experiment. In addition, three typical modulation formats, the PSM, BPSK and OOK, are achieved by encoding the input electrical Gaussian pulses. The UWB signal is generated in the optical domain and distributed with low loss to the base station via optical fiber, so it can be integrated into the passive optical network and wireless access network.