Optical Distribution of UWB: Low Complexity Pulse Generation

Optical Distribution of UWB: Low Complexity Pulse Generation Supporting OOK and PSK Mehrdad Mirshafiei, Mansour Dastmalchi, Mohammad Abtahi, Sophie LaRochelle, and Leslie A. Rusch Center for Optics, Photonics, and Lasers (COPL), Department of Electrical and Computer Engineering Laval University, Quebec, Canada, G1V 0A6 Email: [email protected]

Abstract—Optical transport of UWB signals extends the reach of these power-limited signals. We propose and experimentally demonstrate a simple, low-cost method to generate UWB pulses in optics. Unlike other methods, we use the minimal hardware configuration (source/modulator/photodetector) without requiring RF pulse shaping. A novel combination of data and a sinusoidal signal modulates the intensity of a continuous wave laser to create various UWB pulses. For impulse radio UWB, onoff keying (OOK) or phase-shift-keying is accomplished simply by adjusting the data and the sinusoidal signal amplitudes. Multiband UWB signals with OOK can also be realized by this technique.

I. I NTRODUCTION Ultra-wideband (UWB) radio transmission has attracted much attention since its allocation of an unlicensed frequency band by the US Federal Communications Commission (FCC) in 2002 [1]. This wide but extremely power limited allocation, mainly from 3.1 to 10.6 GHz, allows very high data rate communications for close range applications. The wireless transmission range of UWB systems is limited to a few meters [2] due to power restrictions and high intersymbol interference above 100 Mb/s. Optical fiber distribution of UWB signals extends the reach of such systems to several kilometers. By generating UWB pulses in the optical domain, we avoid extra electrical to optical signal conversion. Optical transport has been proposed for both impulse radio (IR) UWB and multiband (MB) UWB. We briefly review optical generation of IR-UWB and MB-UWB signals. Several approaches have been proposed for optical generation of IR-UWB waveforms. In [3], an intensity modulator is biased in the nonlinear regime to generate a Gaussian doublet pulse. The pulse, however, does not respect the FCC mask. In another work, self-phase modulation in highly nonlinear fiber was used to spectrally broaden Gaussian pulses. Gaussian doublet pulses were generated by optical balanced detection [4]. FCC-compliant pulses were generated in [5] by spectral shaping of a mode-locked laser using fiber Bragg grating. These methods all require an external Mach-Zehnder modulator (MZM) for on-off-keying (OOK) data modulation. Multiband UWB minimizes interference to existing narrowband systems by flexible band selection [6]. In MB-UWB, separate temporal signals occupy bands with bandwidths greater than or equal to 500 MHz. Optical generation of MB-UWB signals was shown in [7] using two phase modulators and

978-1-4244-7825-5/10/$26.00 ©2010 IEEE

polarization-maintaining fiber (PMF). No data modulation of the optically generated signals was demonstrated. The minimal hardware configuration for optical generation of UWB signals is an optical source, an external modulator and a photodetector. For IR-UWB some pulse shaping must also be used, whether RF or optical. Most UWB-over-fiber methods use several additional optical components to generate the UWB pulses and an external modulator for data modulation. In this paper, we show a very cost-effective method of UWB pulse generation using minimum optical components and no RF pulse shaping. The data signal is added to a sinusoidal waveform and the summation modulates the light intensity using a MZM. We will show that with proper choice of data amplitude, sinusoidal signal amplitude and MZM bias point, pulses generated fall in the UWB frequency band. We demonstrate the flexibility of our solution by experimentally generating FCC-compliant pulses with OOK and PSK modulation formats, as well as MB-UWB pulses. The experimental setup is identical for all the generated pulses; the only tuning is to voltages applied to the MZM. Optical signals are converted to RF and transmitted wirelessly via two UWB antennas. We investigate the equivalent isotropic radiated power (EIRP) of transmitted signals. While our method resembles setups for upconversion of UWB signals using an MZM, our results are quite different. In [8], a UWB monocycle pulse was upconverted using the summation of the pulse and a local oscillator to modulate a laser diode via an MZM. Similarly, upconversion of orthogonal frequency-division multiplexing OFDM-UWB signals was demonstrated in [9]. In our approach we generate UWB pulses rather than upconverting previously generated UWB waveforms. This paper is organized as follows. Section II presents the experimental setup. In Section III, we find mathematical formulas describing the necessary conditions for the signals applied to the MZM. Section IV presents experimental results and discussion, and finally a conclusion is given in Section V. II. E XPERIMENTAL S ETUP Fig. 1 shows the block diagram of the UWB pulse generator and the experimental setup for validation. A continuous wave (CW) laser biased above threshold is used as the source. A polarization controller (PC) followed by a 10 GHz MachZehnder modulator (JDSU OC-192) perform the data modula-

Clock

BERT

Data

LPF

independent so that it can be removed using a DC block. Thus, we require ( ) 𝜋 (𝑉𝑏 + 𝐷(𝑡)) = 0. (3) 𝐽0 (𝜋𝑉𝜋 /𝑉𝑚 ) cos 𝑉𝜋

Signal Gen.

Vm SinZt

Vb CW Laser

PC

Optical UWB pulse generator

PD DC block

MZM

LPF

Scope Fig. 1.

LNA

PA

The schematic diagram of the experimental setup.

tion. A power combiner (Marki PD-0010) combines the data coming from a bit error tester (BERT) with a sinusoidal signal generated from the signal generator. Note that the sinusoidal signal also serves as the BERT clock input. The rectangular data signal is smoothed using a lowpass filter (LPF). A photodetector (PD) performs the optical-to-electrical (O/E) conversion. A DC-block eliminates the DC value of the electrical signal and an LPF dampens the spectral sidelobes of the pulse. The resulting UWB pulses are amplified and transmitted using commercial SkyCross SMT-3TO10MA antennas. The received waveforms are amplified using a low noise amplifier (LNA, Mini-Circuits ZVA-183-S) and captured by a sampling oscilloscope (Agilent 86100A). III. T HEORETICAL S TUDY

We will find values of bias 𝑉𝑏 , oscillator amplitude 𝑉𝑚 , and data amplitude for logical zero and one (𝐷0 and 𝐷1 , respectively) such that not only is (3) respected, but we also generate OOK or PSK signals. A. OOK Modulation Format A solution to (3) is 𝐽0 (𝜋𝑉𝜋 /𝑉𝑚 ) = 0, resulting in the sinusoidal signal amplitude 𝑉𝑚 = 𝑉𝜋 /𝜋𝐽00 , 𝐽00 being the first zero of the Bessel function of the first kind. In OOK modulation, the signal should be zero for data zero, i.e., ( logical ( )) 𝜋 0 0 = 0, 𝐷(𝑡) = 𝐷 . Therefore, from (2), sin 𝑉𝜋 𝑉𝑏 + 𝐷 leading to 𝐷0 = 𝑉𝜋 /2 − 𝑉𝑏 . With these conditions fulfilled, (2) simplifies to { −𝑃𝑖𝑛 𝐽1 (𝜋𝑉𝜋 /𝑉𝑚 ) sin 𝜔𝑡 𝐷 (𝑡) = 𝐷1 (4) 𝑃𝑜𝑢𝑡 = 0 𝐷 (𝑡) = 𝐷0 The result is an OOK modulation format. B. PSK Modulation Format

( ) Another solution to (3) is cos 𝑉𝜋𝜋 (𝑉𝑏 + 𝐷) = 0. Therefore, 𝐷0 = −𝑉𝜋 /2 − 𝑉𝑏 , and 𝐷1 = 𝑉𝜋 /2 − 𝑉𝑏 . In this case, the DC signal is zero and (2) can be expressed as { −𝑃𝑖𝑛 𝐽1 (𝜋𝑉𝜋 /𝑉𝑚 ) sin 𝜔𝑡 𝐷 (𝑡) = 𝐷1 (5) 𝑃𝑜𝑢𝑡 = 𝑃𝑖𝑛 𝐽1 (𝜋𝑉𝜋 /𝑉𝑚 ) sin 𝜔𝑡 𝐷 (𝑡) = 𝐷0 We can see that a PSK modulation format is achieved. The sinusoidal signal amplitude can be adjusted to maximize the output power. IV. E XPERIMENTAL R ESULTS AND D ISCUSSION

The transfer function of the MZM, configured as shown in Fig. 1, is expressed as ) ( 𝜋 (𝑉𝑏 + 𝐷 (𝑡) + 𝑉𝑚 sin 𝜔𝑡) (1) 𝑃𝑜𝑢𝑡 = 𝑃𝑖𝑛 cos2 2𝑉𝜋

In this section, we use the setup shown in Fig. 1 along with the mathematical expressions developed in Section III to generate UWB pulses with OOK and PSK modulation formats, and also multiband-UWB signals.

where 𝑃𝑖𝑛 and 𝑃𝑜𝑢𝑡 are the input and output optical powers, 𝑉𝜋 is the MZM halfwave voltage, 𝑉𝑏 is the bias voltage, 𝐷(𝑡) is the data signal, and 𝑉𝑚 is the amplitude of the sinusoidal signal. This transfer function can be expanded using the Bessel functions ( ) 𝜋 1 1 𝑃𝑜𝑢𝑡 = 𝑃𝑖𝑛 { + 𝐽0 (𝜋𝑉𝜋 /𝑉𝑚 ) cos (𝑉𝑏 + 𝐷 (𝑡)) 2 2 𝑉𝜋 ( ) 𝜋 − 𝐽1 (𝜋𝑉𝜋 /𝑉𝑚 ) sin (𝑉𝑏 + 𝐷 (𝑡)) sin 𝜔𝑡 𝑉𝜋 ( ) 𝜋 (𝑉𝑏 + 𝐷 (𝑡)) cos 2𝜔𝑡 + ...}. (2) + 𝐽2 (𝜋𝑉𝜋 /𝑉𝑚 ) cos 𝑉𝜋

A. OOK modulation

The second and higher order harmonics in (2) are out of the UWB band and are filtered out by the bandpass response of the UWB antennas. The DC term in (2) should be data

The MZM we use has a 𝑉𝜋 = 3.4 V, resulting in 𝑉𝑚 = 𝑉𝜋 /𝜋𝐽00 = 2.6 V. We generate the sinusoidal signal with frequency of 6.85 GHz to center the UWB pulses in the middle of the FCC spectral mask. The BERT generates bits at 6.85 Gbs with a pattern of ‘1100 0000’ for logical one and ‘0000 0000’ for logical zero. The UWB bit rate is therefore 850 Mbps. Fig. 2a shows the generated UWB pulse after adjusting the MZM bias and data amplitude. The pulse has a duration of about 400 ps which is more that the 300 ps duration of ’11’, due to filtering the data with a 7.3 GHz LPF. We observe a rapidly oscillating signal when the data is zero. This is the residual second harmonic in (2). This oscillation, however, is insignificant at the receiver as it is effectively filtered by the bandpass behavior of the UWB antenna. Fig. 2b shows the

0

0.5 0

-0.5

-0.6 0.4 0.6 0.8 Time (ns)

0.2

1

1.2

0

0.2

Normalized EIRP (dBm/MHz)

-1

0

0.2

0.4 0.6 Time (ns)

0.8

1

0.4 0.6 Time (ns)

0.8

1

0.4 0.6 Time (ns)

0.8

1

(a)

Transmit PSD

-20

1 0 -1

0

0.2

EIRP

-30 -40

0

1.2

(b) FCC mask

-10

1

1

Time (ns)

(a) 0

0.4 0.6 0.8

6.8 GHz Centered Pulse

0

-1

(b)

0

5

10 Frequency (GHz)

15

(c)

Fig. 2. (a) The normalized transmit UWB signal, (b) the normalized received waveform, and (c) the power spectral density of the transmit pulse (red) and the EIRP (blue).

8 GHz Centered Pulse

-1

1

5 GHz Centered Pulse

Normalized Received Pulse

Normalized Transmit Pulse

0.6

1 0 -1

0

0.2

normalized measured received pulse after antenna transmission over a distance of about 70 cm in the lab environment. The EIRP is calculated by measuring the antenna gain response using a vector network analyzer as explained in [5]. The normalized EIRP is calculated and shown in Fig. 2.c (blue line). Comparing the EIRP with the transmit pulse spectrum (Fig. 2.c, red line), the second harmonic has been eliminated by the antenna and the EIRP respects the FCC spectral mask. Note that wider spectrum could be expected if the signal and data generators worked at different frequencies, allowing for more precise adjustment of the pulse duration.

Normalized EIRP (dBm/MHz)

(c) 0

(a)

(c)

-10 -20 -30 -40

0

2

4

6 8 Frequency (GHz)

10

12

(d)

B. Multiband-UWB The MB-UWB can be generated similarly to the OOK pulses. The data pattern is set to ‘1111 1100’ to generate pulses with longer time durations. The center frequency of the band is set by the frequency of the sinusoidal signal. A 933 MHz LPF is used to narrow the spectral width of the pulses. Fig. 3a, Fig. 3b, and Fig. 3c show pulses centered at 5 GHz, 6.8 GHz, and 8.5 GHz, respectively. We can see that the pulses have different durations because the duration of the pattern changes depending on the BERT bit rate. The solution would be using an independent clock signal for the BERT. Fig. 3d plots the calculated EIRP for each of the multiband-UWB signals superimposed in one figure. It is observed that the pulses do not have the same spectral bandwidth, which can be attributed to their difference in the time duration. The second harmonic for the pulse centered around 5 GHz falls at 10 GHz, which makes it difficult to filter as it is in the UWB band.

(b)

Fig. 3. The mutiband-UWB signals centered around (a) 5 GHz, (b) 6.8 GHz, and (c) 8.5 GHz. (d) The corresponding EIRP plots.

C. PSK Modulation To have equal duration for the one and zero pulses in the PSK modulation format we generate the data with the pattern ’1111 0000’, corresponding to a symbol rate of 1.7 Gbps. The patterns ‘1111’ and ‘0000’ generate out-of-phase pulses. The sinusoidal signal amplitude is adjusted to maximize the output signal. The MZM bias point and data amplitude are adjusted according to relations developed in Section III. Fig. 4a shows the generated PSK UWB pulses. The pulse duration is about 0.6 ns. Fig. 4b plots the measured received signals after antenna transmission. The autocorrelation function of the received signal (Fig. 4c) shows that a correlator or

1

0

1

0

1

0 -1

0

0.5

1

1.5

2

2.5

3

3.5

Time (ns) (a) 1

0

1

1

0

0

1 Normalized Amplitude

0

One Zero

0.5 0 -0.5 -1

1

-1 0

Time (ns)

(d) 0.5

1

1.5

2

2.5

3

3.5

0

Time (ns) (b) 1

1

0

1

0

1

0

0 -1 0

0.6

0.4

0.2

0

0

1

2

3

4

Time (ns) (c)

Normalized EIRP (dBm/MHz)

Normalized Transmit Pulse Amplitude Normalized Received Pulse Amplitude Normalized Received Pulse Autocorrelation

1

-10 -20

One Zero

-30 -40

0

2

4 6 8 Frequency (GHz)

10

12

(e)

Fig. 4. The PSK scheme. (a) The transmit UWB pulses, (b) the received waveforms, (c) the autocorrelation function of the received waveform, (d) superposition of a 1 and a 0, and (e) the EIRP of the 1 and 0 pulses.

matched filter receiver can easily decode the PSK signal by employing a zero threshold. The sampling times are pointed by green circles in Fig. 4c. Fig. 4e shows the EIRP for the zero and one signals. The spectral bandwidth can be tuned by changing the data pattern. V. C ONCLUSION We experimentally demonstrated a simple, low-cost optical UWB pulse generation method. A CW laser,a MachZehnder modulator, and a photodetector were the only optical components in this technique. A combination of data and a sinusoidal signal was used to generate various UWB pulses. On-off keying and phase-shift keying modulation formats were generated by adjusting the data and the sinusoidal signal amplitudes. Multiband-UWB signals can also be supported using the same setup, however there remain challenges for dealing with undesired second harmonics. Antenna transmission measurements and EIRP calculations were reported. The pulses generated by this method have a narrow optical spectral width and are chirp free. We expect the optical fiber distribution of such pulses would impose little distortion. Future work includes studying the effects of fiber transmission on pulses, measuring bit-error rates under realistic wireless channel conditions, and investigating the possibility of integration using a silicon photonics modulator. ACKNOWLEDGMENT The authors would like to thank A. Ghazisaeidi for worthwhile discussions.

R EFERENCES [1] “First report and order, (Revision of part 15 of the commission’s rules regarding ultra-wideband transmission systems), US. Fed. Comm. Commission, adopted Feb. 14, 2002, released Apr. 22, 2002.” Tech. Rep. [2] D. Porcino and W. Hirt, “Ultra-wideband radio technology: potential and challenges ahead,” IEEE Communications Magazine, vol. 41, no. 7, pp. 66–74, 2003. [3] Q. Wang and J. Yao, “UWB doublet generation using nonlinearly-biased electro-optic intensity modulator,” Electronics Letters, vol. 42, no. 22, pp. 1304–1305, 2006. [4] A. Zadok, X. Wu, J. Sendowski, A. Yariv, and A. Willner, “Photonic Generation of Ultra-Wideband Signals via Pulse Compression in a Highly Nonlinear Fiber,” IEEE Photonics Technology Letters, vol. 22, no. 4, pp. 239–241, 2010. [5] M. Abtahi, M. Mirshafiei, S. LaRochelle, and L. Rusch, “All-Optical 500-Mb/s UWB Transceiver:An Experimental Demonstration,” Journal of Lightwave Technology, vol. 26, no. 15, pp. 2795 –2802, Aug.1, 2008. [6] X. Wu, Z. Tian, T. Davidson, and G. Giannakis, “Optimal waveform design for UWB radios,” IEEE Transactions on Signal Processing, vol. 54, no. 6 Part 1, pp. 2009–2021, 2006. [7] H. Chen, T. Wang, M. Li, M. Chen, and S. Xie, “Optically tunable multiband UWB pulse generation,” Opt. Express, vol. 16, pp. 7447–7452, 2008. [8] Y. Le Guennec and R. Gary, “Optical frequency conversion for millimeterwave ultra-wideband-over-fiber systems,” IEEE Photonics Technology Letters, vol. 19, no. 13, pp. 996–998, 2007. [9] P. Lombard, Y. Le Guennec, G. Maury, E. Novakov, and B. Cabon, “Optical Distribution and Upconversion of MB-OFDM in UltrawideBand-Over-Fiber Systems,” J. Lightwave Technol, vol. 27, pp. 1072–1078, 2009.