1. Introduction

The ocean is a huge and mostly unexplored place on our planet, and an incredible 95% of all the oceans in the world still remains unseen by humans. For this reason, oceanic research and exploration have attracted the attention of all over the world. Underwater sensors, or even underwater wireless sensor networks (UWSNs), as well as a variety of underwater vehicles are widely deployed for monitoring underwater environment, submarine life and seafloor activities [1]. With the growing quantity of underwater sensors and vehicles, the demand of collecting data from underwater platforms is increasing. The data can be collected via sending out either a mothership or an airborne vehicle to the location near the underwater platforms. The later solution is more attractive in terms of cost, flexibility and time consumption, especially when an unmanned aerial vehicle is employed. Conventionally, radio is used to communicate between the underwater vehicles and the airborne terminals [2]. However, because radio waves suffer heavy attenuation in water, the underwater vehicles need to be very close to the surface to communicate with the airborne terminals [2–4] which is very time-consuming and labor-intensive. For the underwater fixed sensor nodes, buoys floating at certain depth in the sea can be employed to relay the acquired information to the surface via acoustic communications [5]. Although the absorption of sound is about three orders of magnitude lower than radio in sea water [6], the underwater acoustic communications also face considerable challenges, including limited bandwidth, high propagation delay and low propagation speed [5,7]. Alternatively, green-blue laser communication has the advantages of high modulation bandwidth, high security, high flexibility, low power consumption and low latency [8–12]. In particular, the green-blue laser light can penetrate into water much deeper than the radio wave. For this reason, the green-blue light can be used as the information carrier for the communication between the underwater platforms and the airborne vehicles, as shown in Fig. 1. In [13], the authors theoretically studied the property of air-water communication channel, including the influence of the air-water surface on the channel. In [14], the authors designed and developed an air-water optical communication system based on a red laser diode (LD). However, the achieved bit rate was less than 110 Kbps and the demonstrated transmission distance was less than 1 m.

 

Fig. 1 The conception of laser communication between underwater and airborne terminals.

Download Full Size | PDF

In this work, we experimentally investigate the feasibility of high-speed laser communication through an air-water channel, which is the first step for the future study on laser communications between the underwater platforms and the airborne terminals. The downlink and uplink experiments are conducted separately. Orthogonal frequency division multiplexing (OFDM) is adopted in the demonstration, due to its high spectral efficiency and potential resistance to channel instability [10]. A single-mode fiber-pigtailed green LD and an avalanche photodetector (APD) are employed in the experiment. Over a 5-m air channel and a 21-m underwater channel, we have achieved a gross bit rate of 5.5 Gbps at a bit error rate (BER) of 2.47×10⁻³ in the downlink by using power loading (PL). The same bit rate is also achieved in the uplink with a BER of 2.92×10⁻³. To our knowledge, this is the first demonstration of 5.5-Gbps laser communication in both the downlink and the uplink through a 26-m air-water channel. As reported in [15], the authors achieved a data rate of 2 Gbps over a 12-meter-long, and 1.5 Gbps over a 20-meter-long underwater channel by using a 450-nm LD. A 10.2-m and a 1.7-m underwater laser transmission (UWLT) at data rates of 5.2 Gbps and 12.4 Gbps respectively in tap water have also been successfully demonstrated [16]. Until very recently, a 405-nm LD with light injection and optoelectronic feedback techniques was used as an underwater optical transmitter [17], enabling a 10-meter UWLT of 10-Gbps 16-QAM-OFDM signals. Compared to these works (single-direction underwater transmission), we not only have achieved a larger bit rate-distance product in both the downlink and the uplink, but also added a 5-m transmission link in the air.

2. Experimental setup

We have conducted a proof-of-concept experiment to demonstrate both the downlink and uplink transmission of the proposed air-water optical wireless communication scheme, based on the setup in Fig. 2. Note that only the downlink setup is shown here as the uplink one is similar except that the transmitting end and the receiving end are exchanged. The OFDM transmitter module, the OFDM receiver module and the water tank are shown in the insets of Fig. 2.

 

Fig. 2 The downlink experimental setup of the proposed air-water laser communication scheme. Inset: (a) the transmitter module, (b) the receiver module and (c) the water tank.

Download Full Size | PDF

In this experiment, we first translated a pseudorandom binary sequence (PRBS) into parallel binary code via serial-to-parallel (S/P) translation. Then we assigned these binary data to different OFDM subcarriers after they were encoded into 32 quadrature amplitude modulation (32-QAM) symbols. Each OFDM frame was allocated 150 OFDM symbols, in which 4 symbols were for channel equalization. Additional 2 training symbols (TS) were added for timing synchronization. Hermitian symmetry was imposed in the inverse fast Fourier transform (IFFT) to generate the real-valued 32-QAM OFDM signals. A cyclic prefix (CP) of 100 samples was added to mitigate inter-symbol interference (ISI) in the transmission link and a frequency gap of 1 subcarrier near zero frequency was used to avoid the low-frequency noise. Finally, the generated OFDM signals were loaded into a Tektronix AWG70002A arbitrary waveform generator (AWG) via a local area network (LAN). The sampling rate of the AWG was set at 5 GSamples/s and the amplitude of the AWG output was clipped within 0.5 Vpp. After being electrically pre-amplified by a 25-dB broadband amplifier (Mini-Circuits ZHL-6A-S+) (AMP), the driving voltage of the baseband OFDM signals were further adjusted by a key-press variable electrical attenuator (VEA). A low cost, commercially available and single-mode fiber-pigtailed green-light LD (Thorlabs LP520-SF15) with the emission peak wavelength at around 520 nm is mounted on a thermoelectric cooler module (Thorlabs TED 200 C) and driven by a LD controller (Thorlabs LDC205C). The baseband OFDM signals were superimposed on the driven current via a bias-tee (Thorlabs LDM9LP). The pigtail fiber was followed by a collimation lens to achieve parallel emitting light. A water tank (length: 25 m, width: 0.6 m, height: 0.5 m) was filled with tap water to simulate the underwater channel. Note that a pair of mirrors were placed in the water tank to realize the air-water transmission by means of mirror reflection, as shown in Fig. 2. After transmitting through a 5-meter air channel and a 21-meter water channel, the optical OFDM signals were focused into a 1-GHz APD (Menlo Systems, APD210) by a plano-convex lens. The detected signals were captured by a mixed signal oscilloscope (MSO) with a sampling rate of 100 GSamples/s. Then the captured signals were sent to a computer via a LAN for demodulation.

3. Experimental results

We first investigated the impact of the distance of the water channel on the BER performance in the downlink, where the laser light first went through a 5-m air channel and then went through a water channel. We set the LD bias current at 105.07 mA, and set the VEA value at 0 dB, to achieve the optimized results. Figure 3(a) shows the measured BERs of the 5-Gbps OFDM signal (using 205 subcarriers) at different underwater channel distances. The constellation maps of the 32-QAM OFDM signal at underwater channel distance of 10 m, 11 m, 12 m, 13 m, 20 m and 21 m are shown in the insets of Fig. 3(a). As the underwater channel distance increases from 10 m to 21 m, the measured BERs are approaching the forward error correction (FEC) limit. The BER fluctuation is due to the small detection area of the APD that requires good pointing accuracy between the transmitter and receiver. In the experiment, 21 m was the maximum distance that our tank could support. So we then fixed the underwater transmission distance at 21 m (with a 5-m air channel) and investigated the maximum data rate that could be achieved by gradually increasing the data rate to 6 Gbps, with the results shown in Fig. 3(b). The constellation maps of the 32-QAM OFDM signal at data rates of 4.8 Gbps, 5.3 Gbps, 5.5 Gbps and 6 Gbps are shown in the insets of Fig. 3(b). For the 32-QAM OFDM signals with 218 subcarriers, we achieved a gross bit rate of 5.3 Gbps at a BER of 2.64×10⁻³ which is below the FEC threshold of 3.8×10⁻³, and the net bit rate was 4.4 Gbps. The waveform of the captured 5.3-Gbps OFDM signal with an average amplitude of 249.8 mV is shown in Fig. 4(a) and the corresponding spectrum is shown in Fig. 4(b). After extending the signal bandwidth up to 1.1 GHz (using 226 subcarriers) to achieve a gross bit rate of 5.5 Gbps, the BER was increased to 9.09×10⁻³ which is beyond the FEC limit. In order to lower the BER, we further adopted PL. For subcarriers from 201 th to 226 th, the allocated power was linearly emphasized from 2 dB to 7 dB. As a result, we achieved a lower BER of 2.47×10⁻³ at the same data rate. The net bit rate was 4.5 Gbps. The average amplitude of the captured 5.5-Gbps OFDM signal with PL was 224.0 mV in the downlink. Figures 4(c) and 4(d) present the corresponding spectrum of the 32-QAM OFDM signal without and with PL, respectively. The effect of PL of enhancing SNR in high frequency can be verified by comparing Figs. 4(c) and 4(d).

 

Fig. 3 (a) BERs of the downlink 5-Gbps OFDM signal at different underwater channel distances (with a 5-m air channel). Insets: corresponding 32-QAM constellation diagrams. (b) Downlink BERs at different data rates (21-m underwater and 5-m air channel). Insets: corresponding 32-QAM constellation diagrams.

Download Full Size | PDF

 

Fig. 4 (a) The waveform and (b) the corresponding spectrum of the captured downlink 5.3-Gbps OFDM signal. The spectrum of the captured downlink 5.5-Gbps OFDM signal (c) without PL, (d) with PL.

Download Full Size | PDF

We then examined the uplink performance at the data rate of 5.5 Gbps using 32-QAM. In the uplink experiment, the laser light first went through a 21-m water channel and then went through a 5-m air channel. Again, we set the LD bias at 105.07 mA and the VEA at 0 dB. The OFDM parameters were also kept the same as those in the downlink experiment. The measured BER was 8.24×10⁻³ without using PL. By employing PL, the BER was reduced to 2.92×10⁻³. The average amplitude of the captured 5.5-Gbps OFDM signal with PL was 207.0 mV in the uplink. After the transmission through a 21-m water channel and a 5-m air channel, error vector magnitudes (EVMs) of different subcarriers in both the downlink and uplink 5.5-Gbps OFDM signals with PL, are shown in Fig. 5. The corresponding constellation diagrams of the uplink 32-QAM OFDM signal without PL and with PL are shown in insets (a) and (b), from which the improvement resulting from PL can be observed.

 

Fig. 5 The downlink and uplink EVMs for different subcarriers. Insets: constellation maps of the captured uplink 5.5-Gbps OFDM signal (a) without PL, (b) with PL.

Download Full Size | PDF

4. Discussion

Although these experimental results are impressive in terms of both data rate and transmission distance, they are achieved in a static air-water channel under good pointing accuracy between the transmitter and receiver. In real scenario, such a channel is nonstatic but suffers from perturbations induced by random scattering, turbulence, and waves, to name a few. Future studies are required to tackle all these challenging issues. As an example, we have theoretically investigated how the BER performance is affected by the link distance, the divergence angle and the deflection angle of the light source [18], using Monte Carlo simulation.

5. Conclusion

In summary, we have experimentally investigated the feasibility of an air-water laser communication system in both the downlink and the uplink, which demonstrates great potential for the application of optical wireless communications between the underwater platforms and the airborne terminals. In the experiment, high spectral efficiency OFDM technology and a single-mode pigtailed green-light LD were employed. Over a 5-m air channel and a 21-m water channel, a 5.3-Gbps transmission without PL was achieved in the downlink. By using PL, we increased the bit rate up to 5.5 Gbps in both the downlink and the uplink. The corresponding BERs were 2.47×10⁻³ and 2.92×10⁻³, respectively.