Multiple Access Techniques for Bipolar Optical Code Division in Wireless Optical Communications

In this paper, multiple access techniques for bipolar optical code division are proposed for application in wireless optical communications. First, several fiber Bragg gratings (FBGs) and optical circulators are used to encode optical signals. Then, the horizontal or vertical polarization state can be controlled using bipolar data. Experiments were conducted using two devices to determine the feasibility of using multiple access techniques for bipolar optical code division in free-space optical communication. The first experiment tested the first architecture, in which each user operates a mechanical optical switch by switching data bits to generate optical signals with specific polarization states through a collimator in a free-space channel. As the decoding components, several FBGs and optical circulators are used as primary devices. Decoded signals are sent to a balanced photodetector, which can reconstruct original data from the encoder. The second experiment involved a second architecture, in which an encoder with an erbium-doped fiber amplifier and a decoder with an attenuator are used to improve system stability. The third experiment was designed to test whether multiple access interference can be alleviated with the proposed architectures. In the final experiment, an ultrafast optical switch was used instead of the original optical switch for selecting optical signals with a specific polarization state to improve the overall transmission rate. The experimental results indicate that the proposed scheme could be successfully implemented in free-space optical communication systems. The architectures of the proposed scheme mitigate multiple access interference with a simple and cost-effective design.


I. INTRODUCTION
Because of the limitations of wired optical networks, such as the inability to satisfy requirements for application in certain environments, wireless optical communication (WOC) networks have become an important research topic in recent years. However, wireless communication also has certain limitations. Wireless communication is strictly regulated in particular scenarios, such as communication between airport towers and aircraft. Furthermore, precision instruments used in clinical settings are subject to electromagnetic wave inter-The associate editor coordinating the review of this manuscript and approving it for publication was Tariq Umer . ference (EMI). The development of WOC systems is crucial for solving real-world problems. However, to develop such systems, some problems must be overcome in free-space optical (FSO) communication. For example, long-distance transmission involves considerable power attenuation, and outdoor wireless transmission is susceptible to influenced by atmospheric and environmental conditions (e.g., rain and fog).
Research on multiplex processing in FSO communication systems is crucial. The first multiplex processing scheme originated from time-division multiplexing techniques [1]. According to its operating principle, input signals are multiplexed into different time slots. One advantage of multiplex VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ processing is that it allows all users to optical signals at the same wavelength. However, adjacent wavelengths still require a time delay for accessing the correct wavelength signal. Thus, if the number of users is large, a long fiber must be used. Another multiplex technique, which is termed wavelength-division multiplexing, involves using multiple wavelengths as the transmission signals of multiple users. However, in this approach, the number of users is limited by the light-source bandwidth [2], [3]. Many researchers have used the novel multiplexing technique of optical code-division multiple access (OCDMA) [4], [5]. In this approach, code-division multiple access is applied for optical fiber communication. Thus, in OCDMA communication systems, an optical coding technique is employed in which each user is assigned a codeword to avoid mutual interference in the same channel. This technique enables the simultaneous transmission of unsynchronized data from multiple users of the same channel and bandwidth [6]. The multiplex technique provides favorable antijamming properties, a moderate security level, and highcapacity processing. Spectral amplitude coding (SAC) is an attractive OCDMA scheme because of its ability to mitigate multiple access interference (MAI) [5].
For the demodulation of optical signals, OCDMA schemes can be divided into coherent and incoherent schemes. Incoherent OCDMA schemes harness optical field intensity for encoding optical signals. The main encoding method used is unipolar encoding (0, 1). Unipolar encoding architectures have simple and low-cost designs; however, the number of codes that can be provided to users is considerably smaller in unipolar encoding architectures than in bipolar encoding architectures. An increase in the number of users necessitates an increase in code length. Coherent systems function by encoding the input light phase through a matching filter to change the phase of the optical signal [7].
In 2007, Zeng et al. [8] proposed a unipolar encoding/bipolar decoding OCDMA scheme. They employed an electro-optic phase modulator and two fiber Bragg gratings (FBGs) to construct a sequence-inversion-keyed system. This system has low complexity because each unipolar encoder or decoder uses only two FBG array. However, the corresponding bipolar decoders are implemented using a series of FBGs, which limits the decoding process and system transmission rate.
In 2006, Chang and Huang [9] designed a spectral polarization coding scheme for an OCDMA network. A corresponding codec is established using several FBGs and polarization beam splitters. This coding scheme is employed to modulate the wavelengths of an optical signal with unique polarization components. The results of Chang et al. indicated that MAI could be reduced using the aforementioned scheme, which is consistent with the results obtained in previous studies using SAC schemes. However, the polarization state of the optical signal in this system is unstable during long-distance fiber transmission. Moreover, the complexity of its codec is high.
In this paper, a novel bipolar scheme is presented. A Walsh-Hadamard (WH) code was used to construct the scheme. Then, a simple architecture was developed using WH codes and several FBGs. The system is simple and provides advantages such as a straightforward design, low cost, and moderate security level. Because the proposed approach involves using an FSO communication technique to eliminate the necessity of using a star coupler, the topology is suitable for application in mountain-to-mountain communication. We conducted four experiments to verify the feasibility of the proposed system. The first experiment was performed to determine whether the proposed bipolar scheme functions properly. Certain atmospheric parameters can attenuate the intensity of optical signals transmitted in free space and were measured. In the second experiment, an erbium-doped fiber amplifier (EDFA) was used as a preamplifier for the encoder. Moreover, an attenuator was employed in the decoder to compensate for the different power levels of the components. In the third experiment, the another decoder was tested to determine if MAI could be completely eliminated with the proposed scheme. From the first experimental results, the limitation of the switch caused performance degradation during decoding. Therefore, in the final experiment, an ultrafast optical switch was used to improve the system performance.
The remainder of this paper is organized as follows. Section II describes the proposed bipolar OCDMA scheme and the corresponding system design. Section III presents the experimental results for FSO communication systems using bipolar OCDMA schemes, such as systems using a mismatched decoding process, and describes the system stability improvements achieved using an attenuator and EDFA. Finally, the conclusions are provided in Section IV.

II. PROPOSED BIPOLAR OCDMA TECHNIQUE
The proposed bipolar OCDMA technique is used to transmit bipolar data signals, and the optical signals of each user are transmitted over a common FSO path. In the proposed system 83512 VOLUME 8, 2020 design, FBGs are used as encoder and decoder devices. For data processing, an optical switch is used to encode the data bit of each user into an optical signal, which is transmitted to the decoder end through an FSO channel.
A WH code is used in the FSO OCDMA communication system because of its orthogonality. The WH matrix design is adopted from the matrix design in [8].
where N is equal to 2 n and denotes the length of the WH code and the kth row of the matrix represents the kth optical codeword of the WH code assigned to the kth user. For example, when N = 4, the matrix of the proposed WH code can be expressed as follows: In (2), h k denotes the kth row in the WH matrix to be decomposed into two unipolar codes, namely c kV and c kH . The codes c kV and c kH represent the positive and negative elements of h k , respectively, and form a complementary pair in the set ∈ {0, 1}. With the above approach, h k is transmitted for a data bit ''1'' for the kth user, whereas the complement row of h k (h k ) is transmitted for a data bit ''0'' of the kth user. The notation (C V − C H ) represents the matrix formed when all users send a data bit ''1.'' The WH matrix can be mapped to the wavelength domain and characterized by one of two orthogonal polarization states: vertical or horizontal. In the second equality of (2), the first and second matrices are assigned to the vertical (C V ) and horizontal (C H ) polarization states, respectively. Similarly, the complement matrix of H 2 can be decomposed as follows: Therefore, the optical codeword R k , which is sent from the encoder of the kth user, can be expressed as follows: where b k ∈ {0,1} and c kV and c kH are the kth code encoded in the vertical and horizontal polarization states, respectively. Thus, if b k = 1, the spectrum of the coded signal corresponds to (c kV −c kH ). Similarly, if b k = 0, the spectrum of the coded signal is (c kV − c kH ). To apply WH codes with length N in the proposed schemes, the results of the following equations must be obtained: and With the results of (5) and (6), the WH codes can be implemented in an FSO OCDMA communication system to mitigate MAI by using mathematical subtraction.   Output spectra at the output ports of the MEMS fiber optical switch operating with different data bits. (a) Optical spectra output from port 1 in the optical switch for a data bit ''1. '' (b) Optical spectra output from port 1 in the optical switch for a data bit ''0. '' (c) Optical spectra output from port 2 in the optical switch for a data bit ''1. '' (d) Optical spectra output from port 2 in the optical switch for a data bit ''0. '' Through the aforementioned derivations, an FSO communication system can be constructed using bipolar OCDMA schemes. Fig. 1 displays the design of the proposed encoder with a WH code length of 4. It includes a superluminescent diode (SLD) light source, a 1 × 2 splitter, two optical circulators, two series of FBGs, a polarization beam splitter (PBS), and a 2 × 2 optical switch for selecting optical signals with the appropriate polarization components.
First, the optical signal of the SLD output is input to two optical circulators through the 1 × 2 splitter. Then, the two FIGURE 6. Output spectra at the output ports of the PBS in Decoder 2 for different data bits. (a) Optical spectra from output port 1 in the PBS for data bit ''1. '' (b) Optical spectra from output port 2 in the PBS for data bit ''1. '' (c) Optical spectra from output port 1 in the PBS for data bit ''0. '' (d) Optical spectra from output port 2 in the PBS for data bit ''0. '' .
ports of two optical circulators are connected to two series of FBGs according to the assigned optical codewords derived from the WH code (h k ). For example, when N = 4, the encoder of the second user uses a series of upper FBGs to reflect the optical signals with wavelengths λ 1 and λ 3 , which are transmitted to the first input port of the 2 × 2 optical switch through the circulator. Similarly, the optical signals with wavelengths λ 2 and λ 4 are transmitted to the second input port of the 2 × 2 optical switch. Then, the bipolar OCDMA scheme can be implemented using an optical switch according to the data bit of each user. If the data bit of the user is ''1,'' the input signals pass parallel to each other into the PBS. Conversely, when the data bit of the user is ''0,'' the input optical signals cross each other to pass through the optical switch before arriving at the input ports of the PBS. Therefore, optical signals with appropriate wavelengths and polarization states arrive at the output port of the PBS. Finally, the signals at the output port of each encoder are coupled to a fiber collimator and then transmitted into free space to implement the FSO communication system.
The design of the proposed decoder is illustrated in Fig. 2. The decoder comprises a PBS, two optical circulators, two series of FBGs, two 2 × 1 couplers, and a balanced photodetector (BPD) for mitigating the influence of MAI. First, the received FSO signal is input into a fiber collimator before being converted into an optical signal. Subsequently, the optical signal (R) from the output port of the fiber collimator is depolarized into two optical circulators through the PBS. Therefore, the vertical polarization (90 • ) component of the incident signals appears only at the first port of the upper optical circulator, whereas the horizontal polarization (0 • ) component of the incident signals appears only at the first port of the lower optical circulator.
Let the intensity of the received optical signal (R) be proportional to R V + R H .
where K is the number of active users, R V is the sum of the intensities of the optical signals with only vertical polarization components, and R H is the sum of the intensities of the optical signals with only horizontal polarization components. VOLUME 8, 2020 According to (7), the optical signals (R V ) with wavelengths corresponding to the ''1s'' position of c kV are reflected by a series of FBGs in the decoder for the kth user. These reflected signals then arrive at the first input port of the lower 2 × 1 coupler. Simultaneously, the optical signals (R V ) with residual wavelengths corresponding to the ''1s'' position of c kV are transmitted to the first input port of the upper 2 × 1 coupler. Similarly, the optical signals (R H ) with wavelengths corresponding to the position of the ''1s'' of c kH arrive at the second input port of the upper 2 × 1 coupler after being FIGURE 9. Decoded spectra at the output ports of the attenuator and the lower coupler at different data bits. (a) Optical spectra at the output port of the attenuator in decoder 2 for sending data bit ''1. '' (b) Optical spectra at the output port of the lower coupler in decoder 2 for sending data bit ''1. '' (c) Optical spectra at the output port of the attenuator in decoder 2 for sending data bit ''0. '' (d) Optical spectra at the output port of the lower coupler in decoder 2 for sending data bit ''0. '' reflected by a series of lower FBGs. Concurrently, the optical signals (R H ) with reflected wavelengths are transmitted to the second input port of the lower 2 × 1 coupler. Then, the optical signals at the output ports of the upper and lower 2 × 1 coupler are transmitted to the input ports of the BPD to respectively conduct correlation subtraction and eliminate MAI. The aforementioned approach is detailed in the following equation.
where h l and h l denote the assigned and complementary optical codewords of the desired user. Finally, the decision current (I ) is used to determine the data bit of the desired user in the FSO channel.

III. EXPERIMENTAL SETUP AND RESULTS
The objective of the experiments was to test the feasibility of FSO communication systems using the proposed bipolar OCDMA schemes. The experimental FSO communication system using WH codes with N = 4 is displayed in Fig. 3. The NXTAR SLD-2000 was adopted as an SLD light source with stable power. Furthermore, wideband couplers (Fiber Optic Communications, Inc., Taiwan) were used as 2 × 1 couplers and the 1 × 2 splitter. An Anritsu MS9170C optical spectrum analyzer was used to determine the accuracy of the spectral codec output. The BPD model used for decoding was Model 1817 (New Focus Inc., CA, USA), which was used to perform signal subtraction in the optical domain and then convert the results into electrical signals. A ±15-V currentlimited power supply was used to power the BPD. Moreover, a Tektronix TDS2102B digital oscilloscope was used for monitoring the signal output of the BPD. Finally, a Twintx TFG 3510 signal generator was used to obtain square waves with different duty cycles to simulate the data transmission by multiple users. The FBG resonance wavelengths used for the FBG codecs used in this study were 1543, 1546, 1549, and 1552 nm, represented by λ 1 , λ 2 , λ 3 , and λ 4 , respectively. The aforementioned wavelengths correspond to the binary sequences of −1s and 1s. We made the following preliminary assumptions: (1) The WH codes (h) for the four users are h 1 = (1, 1, 1, 1), VOLUME 8, 2020   After propagation through the PBS into a free-space channel, the encoded spectra with specific polarization states from FBG encoder 2 were sent to FBG decoder 2 for decoding. The received optical signals were then depolarized through the PBS of decoder 2. Subsequently, they arrived at port 2 of the upper and lower optical circulators. Fig. 6(a) displays the depolarized spectra (λ 1 , λ 3 ) with a 90 • component arriving at the first output port of the PBS in decoder 2 when user 2 sent data bit ''1.'' For these spectra, the central wavelengths of λ 1 and λ 3 were 1543 and 1549 nm, respectively, and the corresponding light intensities were −45.29 and −44.2 dBm. Fig. 6(b) presents the depolarized spectra (λ 2 , λ 4 ) with a 0 • component appearing at the second output port of the PBS in decoder 2 when user 2 transmitted data bit ''1.'' For these spectra, the central wavelengths of λ 2 and λ 4 were 1546 and 1552 nm, respectively, and the corresponding light intensities were −39.92 and −38.96 dBm. Fig. 6(c) displays the depolarized spectra (λ 2 , λ 4 ) with a 90 • component appearing at the first output port of the PBS in decoder 2 when user 2 sent data bit ''0.'' For these spectra, the central wavelengths of λ 2 and λ 4 were 1546 and 1552 nm, respectively, and the corresponding light intensities were −46.58 and −45.23 dBm. Fig. 6(d) displays the depolarized spectra (λ 1 , λ 3 ) with a 90 • component appearing at the first output port of the PBS in FBG decoder 2 when user 2 transmitted data bit ''0.'' For these spectra, the central wavelengths of λ 1 and λ 3 were 1543 and 1549 nm, respectively, and the corresponding light intensities were −36.66 and −36.11 dBm. Therefore, the light output through the free-space channel had an intensity loss of approximately 10-15 dB for the reflected spectra. This loss was due to the effects of atmospheric disturbances and divergence.
The loss difference between the horizontal and vertical paths through the PBS in encoder 2 was estimated to be 5 dB.
The depolarized spectra were then decoded by two series of FBGs, which were coupled in parallel for signal arrival at the input ports of the BPD. Fig. 7(a) displays the decoded spectra (λ 1 , λ 2 , λ 3 , λ 4 ) measured at the output port of the upper coupler in FBG decoder 2 when user 2 had data bit ''1.'' For these spectra, the central wavelengths of λ 1 , λ 2 , λ 3 , and λ 4 were 1543, 1546, 1549, and 1552 nm, respectively, and the corresponding light intensities were −45.2, −45.7, −45.46, and −46.43 dBm. Fig. 7(b) illustrates the decoded spectra (λ 1 , λ 2 , λ 3 , λ 4 ) measured at the output port of the lower coupler in FBG decoder 2 when user 2 had data bit ''1.'' For these spectra, the central wavelengths of λ 1 , λ 2 , λ 3 , and λ 4 were 1543, 1546, 1549, and 1552 nm, respectively, and all the corresponding light intensities were lower than −53 dBm. Fig. 7(c) displays the decoded spectra (λ 1 , λ 2 , λ 3 , λ 4 ) measured at the output port of the upper coupler in FBG decoder 2 when user 2 had data bit ''0.'' For these spectra, the central wavelengths of λ 1 , λ 2 , λ 3 , and λ 4 were 1543, 1546, 1549, and 1552 nm, respectively, and all the corresponding light intensities were lower than −50 dBm. Fig. 7(d) displays the decoded spectra (λ 1 , λ 2 , λ 3 , λ 4 ) measured at the output port of the lower coupler in FBG decoder 2 when user 2 had data bit ''0.'' For these spectra, the central wavelengths of λ 1 , λ 2 , λ 3 , and λ 4 were 1543, 1546, 1549, and 1552 nm, respectively, and the corresponding light intensities were −43.94, −45.29, −45, and −46.32 dBm, respectively. The results displayed in Fig. 7 indicate that the performance of the decoded spectra in converting the results into accumulated VOLUME 8, 2020   To test the stability of the entire system, a second experiment was conducted. This experiment involved using an EDFA in encoder 2 to enhance the transmission of light and an attenuator in decoder 2 to mitigate noise (Fig. 8). Fig. 9(a)-(d) displays the decoded spectra at the output ports of the attenuator and lower coupler for system operating with different data bits and using an EDFA and attenuator for FBG codec 2. Fig. 10 displays the decoding results for switching between frequencies during transmission through encoder 2. The BPD was used to convert the decoded spectra into electrical signals. Then, the digital oscilloscope was used to determine the accuracy of data transmission from FBG encoder 2. Fig. 10(a)-(e) displays the decoding results obtained when switching the 2 × 2 optical switch of FBG FIGURE 15. Decoded spectra at the input ports of the BPD for encoder 2 obtained using a Nanona ultrafast optical switch and different data bits. (a) Optical spectra at the output port of the attenuator in decoder 2 for data bit ''1'' input to encoder 2. (b) Optical spectra at the output port of the lower coupler in decoder 2 for data bit ''1'' input to encoder 2. (c) Optical spectra at the output port of the attenuator in decoder 2 for data bit ''0'' input to encoder 2. (d) Optical spectra at the output port of the lower coupler in decoder 2 for data bit ''0'' input to encoder 2. encoder 2 to frequencies of 0.1, 0.5, 1, 1.5, and 2 kHz. Fig. 10(e) reveals that when the signal generator transmitted a signal with a frequency of approximately 2 kHz to the optical switch, the decoding result for FBG decoder 2 exhibited severe waveform distortion. This result was obtained because the OSW22-1310E MEMS fiber optical switch kit (Thorlabs, NJ, USA) only supports switching frequencies lower than 2 kHz.
In experiment 3, a switching frequency of 1 kHz was input into FBG encoder 2. Moreover, optical signals from FBG encoder 2 were accessed using FBG decoder 4. Fig. 11 illustrates the architecture of FBG decoder 4. Fig. 12 illustrates the decoded spectra at the input ports of the BPD using encoder 4 and different data bit inputs to encoder 2. The BPD could convert the decoded spectra into electrical signals. The output results from FBG decoder 4 could then be measured using the digital oscilloscope. Fig.  13 indicates that the designed codec can alleviate the effect of MAI from FBG encoders other than that with the original signal. The alleviation of MAI does not affect the transmitted electrical signal required for access.
In experiment 4, the optical switch of encoder 2 was replaced, and an inverter was used for electrical signal processing. First, a Nanona ultrafast optical switch (Boston Applied Technologies, Inc., MA, USA) was adopted to replace the original fiber optical switch kit and improve the transmission rate of the proposed system. During switching operations, a Nanona optical switch operates differently from an MEMS fiber optical switch. If data bit ''0'' is input into the Nanona optical switch, it enters parallel operation mode. Conversely, if data bit ''1'' is input into the Nanona optical switch, it enters cross mode. Therefore, an inverter was used in experiment 4 for converting bits 1 and 0 into electrical signals. The results obtained in experiment 4 were consistent with those obtained with FBG encoders in experiments 1 to 3. Fig. 14 displays the design of the proposed FBG encoder 2 without an EDFA and with a Nanona optical switch. Fig. 15 displays the decoded spectra at the input ports of the BPD when data bits ''1'' and ''0'' were input into FBG encoder 2 by using a Nanona ultrafast optical switch. The digital oscilloscope was used to access the waveform of the signal transmitted by encoder 2, as displayed in Fig. 16. This figure indicates that when the signal generator input a frequency of 100 kHz into the optical switch, the original signal could be reconstructed through the output of FBG decoder 2. However, the results revealed distortion in decoding when a 200 kHz signal was input into FBG encoder 2. This result was obtained because the Nanona optical switch can operate only at frequencies lower than 200 kHz (response time of 5 µs).
The experimental results indicate that the proposed settings and arrangement can effectively mitigate MAI from FBG encoders other than that with the original signal. Moreover, the transmission rate of the proposed FSO system with the bipolar OCDMA scheme could be improved using an ultrafast optical switch. FBGs are used as primary components for constructing the codec; thus, the proposed FSO system is simple and inexpensive.

IV. CONCLUSION
In this study, an FSO communication system using bipolar OCDMA schemes was developed. This system has the advantages of light weight, flexible design, low cost, moderate security level, and resistance to EMI and MAI. The experimental results indicate that the proposed communication system can access the original signal from its encoder without experiencing MAI from other FBG encoders. Interference from other FBG encoders can be avoided with the proposed system, which reduces crosstalk. The experimental results confirm that system stability can be improved by using an EDFA and attenuator. The measurement results for the signal transmission rate revealed the switching limitations of the optical switch. Signals could only be detected at a frequency below 200 kHz. However, the proposed system still presents potential for application, such as in high-maintenance sensor environments and environmental monitoring. With the development of faster optical switches in the future, the proposed FSO system may achieve even higher data rates for multiple users.