Adaptive Optical Self-Interference Cancellation ... - Semantic Scholar
1018
IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 27, NO. 9, MAY 1, 2015
Adaptive Optical Self-Interference Cancellation Using a Semiconductor Optical Amplifier Matthew P. Chang, Member, IEEE, Chia-Lo Lee, Ben Wu, and Paul R. Prucnal, Fellow, IEEE
Abstract— We experimentally demonstrate an optical system that uses a semiconductor optical amplifier (SOA) to perform adaptive, analog self-interference cancellation for radiofrequency signals. The system subtracts a known interference signal from a corrupted received signal to recover a weak signal of interest. The SOA uses a combination of slow and fast light and cross-gain modulation to perform precise amplitude and phase matching to cancel the interference. The system achieves 38 dB of cancellation across 60-MHz instantaneous bandwidth and 56 dB of narrowband cancellation, limited by noise. The Nelder–Mead simplex algorithm is used to adaptively minimize the interference power through the control of the semiconductor’s bias current and input optical power. Index Terms— Self-interference cancellation, co-site interference cancellation, microwave photonics, slow and fast light, semiconductor optical amplifier (SOA).
I. I NTRODUCTION
S
ELF-INTERFERENCE, or co-site interference, is an increasingly common problem as wireless communication platforms grow in complexity but shrink in size [1], [2]. Self-interference occurs when a colocated transmitter and receiver operate simultaneously; the high-power transmitter signal leaks into the receiver, desensitizing the receiver or overwhelming weaker signals of interest from distant antennas. Vehicles with multiple communication systems [1] and shared base stations [2] all suffer from self-interference. Nearly all radios are half-duplex because of self-interference [3]–[5]. Self-interference cancellation (SIC) is difficult even with knowledge of the transmitted signal due to the sheer power of the interferer [5]. As a result, SIC is performed in the antenna, analog, and digital domains of a communication system [3], [4]. In analog cancellation, an inverted copy of the known interferer is combined with the corrupted signal. With precise amplitude and phase matching across the interferer bandwidth, the interference copy cancels the self-interference signal upon combination, recovering the signal of interest. For broadband SIC, the requirement of precise amplitude and phase matching across a wide bandwidth places a heavy burden on radio-frequency (RF) circuits, which have limited bandwidth because of the poor frequency flatness
Manuscript received December 8, 2014; accepted February 16, 2015. Date of publication February 19, 2015; date of current version April 10, 2015. The authors are with the Lightwave Communications Laboratory, Department of Electrical Engineering, Princeton University, Princeton, NJ 08544 USA (e-mail: [email protected]; [email protected]; [email protected]; [email protected]). Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LPT.2015.2405498
of RF components [4]. In particular, RF delay lines and phase shifters are inherently narrowband and lack the precision to perform high levels of SIC [6]. Optics is a natural remedy to the limited bandwidth of electronics [6]. In optical SIC, the interference copy and corrupted signal are both modulated onto optical carriers and processed in the optical domain. By leveraging the tiny fractional bandwidth of an RF-modulated optical signal, optical SIC can demonstrate extremely broadband cancellation. Previous optical SIC systems have achieved 40 dB cancellation over 50 MHz bandwidth [7] and broadband multipath compensation [8]. However, optical delay lines have prevented optical SIC from being used in practical RF environments. Motorized and thermal delays are too slow, requiring 10’s of milliseconds to adjust. In comparison, indoor WiFi has a coherence time of ∼100 ms, making millisecond tuning a costly overhead to pay [3]. A fast, tunable, and high resolution optical delay is needed to realize practical optical SIC. In this letter, we report an optical SIC system, which uses a single semiconductor optical amplifier (SOA) to perform both amplitude and phase matching across a wide bandwidth. The SOA combines slow and fast light with cross-gain modulation (XGM) to implement electronic control of its propagation delay/phase shift and weight, or transmission [9]. The SOA not only enables a fast, precise solid state optical delay to perform SIC, but also moves the system closer towards full photonic integration. With this technique, we demonstrate 38 dB of broadband cancellation over 60 MHz and adaptive interference cancellation. II. E XPERIMENTAL S ETUP The optical SIC system is a two-tap microwave photonic filter, which accepts two inputs: the corrupted received signal, xr , which consists of the signal of interest (SOI) and self-interference, and a copy of the known self-interference, x t xr = s(t) + nr (t)
(1)
x t = n t (t)
(2)
where n t is the original transmitted signal, s is the SOI, and nr is the self-interference signal after propagating through the transmitter-to-receiver wireless channel. To first order, nr will be a delayed and attenuated version of n t . The optical SIC system is designed to use the SOA’s tunable weight and delay to match the wireless channel’s attenuation and delay as closely as possible. In reality, the wireless channel is more complex than an attenuation and delay, and includes amplifier
1041-1135 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
CHANG et al.: ADAPTIVE OPTICAL SIC USING AN SOA
1019
The two signals are combined by a 50/50 optical coupler, and the combined signal is detected, producing the RF output y = s(t) + [nr (t) − αn t (t − τ )]
Fig. 1. Experimental setup. λ1 = 1550.12 nm, λ2 = 1553.33 nm. PC = Polarization Controller, EDFA = Erbium-Doped Fiber Amplifier, VOA = Variable Optical Attenuator BPF = Bandpass Filter.
noise, nonlinearities, and multipath; however, the goal of the current system is to cancel the strongest component of self-inteference, the line-of-sight signal. A schematic of the fiber setup is shown in Fig. 1. The output from a λ1 = 1550.12 nm distributed feedback laser is split in half to provide the optical carrier for each filter tap. The corrupted received signal, xr , is modulated onto the optical carrier of the top tap, called the receiver tap, by a Mach-Zehnder modulator (MZM). The copy of the self-interference, x t , is modulated onto the optical carrier of the bottom tap, called the transmitter tap, by a second MZM. The self-interference and SOI are generated by two distinct signal generators. xr is created by coupling the self-interference and the SOI using an RF combiner. x t is tapped from the self-interference signal generator. The signal in the receiver tap propagates unaltered to a 50/50 optical coupler. The signal in the transmitter tap is optically inverted, delayed, and weighted to match it in amplitude and phase, except with a 180° offset, to the self-interference in the corrupted received signal. The key component to perform this matching is the SOA, which is highlighted in the call-out in Fig. 1. The SOA performs three essential functions. First, it produces an electronically tunable time delay or microwave phase shift through the slow and fast light effect. In an SOA, the slow and fast light effect generates a tunable delay or phase shift by changing the propagating group velocity in the semiconductor using carrier dynamics [10]. Second, the SOA modifies its weight, or transmission, to amplitude-match the interference copy with the actual self-interference signal. The time delay/phase shift and weight of the SOA are controlled through two independent parameters: the SOA bias current and the input optical power, which is controlled through a variable optical attenuator. The technique used to simultaneously control the SOA’s weight and delay is further described in [9]. Third, the SOA performs the necessary inversion of the transmitter tap signal through XGM. Note that XGM is also used to convert the tap wavelength from λ1 = 1550.12 nm to λ2 = 1553.33 nm to prevent coherent beat noise when the transmitter tap signal is combined with the receiver tap signal.
(3)
where α and τ are the SOA’s weight and delay, respectively. If the self-interference and the interference copy are matched exactly in amplitude and phase, they will destructively interfere upon detection. Any amplitude and phase mismatch across the bandwidth of the self-interference signal will result in residual interference, given by the term in brackets in Eqn 3. The output RF signal is characterized by a network or spectrum analyzer to determine how well the system cancelled the self-interference. The output signal is also used as feedback to adaptively adjust the SOA’s bias current and input optical power to minimize the interference. Both broadband and narrowband cancellation were measured by varying the bandwidth of the self-interference signal and examining how well the system could adaptively cancel the self-interference and recover the SOI. III. E XPERIMENTAL R ESULTS AND A NALYSIS A. Broadband and Narrowband Cancellation Figure 2 shows the experimentally measured cancellation after optimizing the SOA’s weight and delay. The output of the system was measured without (red dashed curve) and then with (blue solid curve) the transmitter tap to observe the effect of interference cancellation. The SOI was a weak 915 MHz single-tone signal. The 900 MHz band was the chosen testbed because it contains the crowded industrial, scientific, and medical (ISM) radio bands within the American continents. Other frequency bands are compatible with the photonic hardware; however, the slow and fast light effect decreases with increasing RF frequency [9]. Broadband cancellation was characterized by sweeping a 0 dBm single-tone interferer across a 60 MHz bandwidth centered at 915 MHz and measuring the system output using an RF spectrum analyzer; 60 MHz is greater than most channel bandwidths used in common protocols like WiFi and LTE. The results in Fig. 2a show that at least 38 dB of cancellation was achieved across the 60 MHz bandwidth. More importantly, the SOI, which had been overwhelmed without cancellation, was recoverable with cancellation. Note that the filter removes in-band interference while leaving the SOI unharmed, making this system more effective than simple spectral filtering. The interference was reduced to the spectrum analyzer noise floor at the center frequency in Fig. 2a, suggesting that the maximum measurable cancellation is greater than what is observed. To determine the maximum achievable cancellation, narrowband measurements were performed using a 10 kHz bandwidth interferer and the same SOI as before. The results are plotted in Fig. 2b, showing the system output with and without cancellation. The maximum level cancellation was 56 dB, and the SOI was recovered. We now examine how cancellation changes with bandwidth and what factors affect cancellation bandwidth. Cancellation degrades with increasing bandwidth (see Fig. 2a) because of the inability of the transmitter tap to exactly model
1020
Fig. 2.
IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 27, NO. 9, MAY 1, 2015
(a) Broadband and (b) narrowband cancellation results. The signal of interest is recovered after cancelling the in-band interference.
Fig. 3. (Top) The measured amplitude and phase mismatch between the transmitter-receiver channel and the transmitter tap from 700 MHz to 1100 MHz. (Bottom) Measured cancellation and simulated cancellation based on the amplitude and phase mismatch. More negative cancellation represents greater interference cancellation.
the transmitter-receiver channel across the entire bandwidth. The transmitter tap, a mostly linear system, can only compensate for the propagation loss and delay of the channel. However, frequency response mismatches between the transmitter tap and the channel that cannot be modeled by a simple delay and weight result in residual interference. Much of this mismatch originates from the remaining RF components in the system, which have non-flat frequency responses (e.g. RF splitters), but there is also a contribution from nonlinearities and amplifier noise. We measure the frequency response mismatch, both amplitude and phase, between the transmitter-receiver channel and the transmitter tap across a 400 MHz bandwidth using a network analyzer and plot the mismatch in Fig. 3 (top). Zero mismatch is indicated by the line representing zero amplitude offset and 180 degrees phase offset. The optimal cancellation is achieved when both amplitude and phase mismatch
approach this line, as seen in the bottom half of Fig. 3, which shows experimentally measured cancellation (solid curve) as well as simulated cancellation (dotted curve). The simulatation is performed by treating the system as a two-tap finite-impulse response filter with frequency-dependent weights and delays that correspond to the mismatch measurements in Fig. 3. The simulation corresponds very well to experiment, showing that minimizing mismatch is central to cancellation performance. Figure 3 shows that there are two qualitatively different contributions to amplitude and phase mismatch in this experiment: a gradually changing mismatch over the total bandwidth and more abruptly changing mismatch over frequency ranges much smaller than the total bandwidth. The gradually changing mismatch originates from the limited bandwidth of the slow and fast light effect, which is determined by the resonance used to generate the refractive index dispersion [9], [10]. Because the SOA only exists in one of the two filter taps, the gradual change in the SOA’s delay and transmission with RF frequency manifests as a gradual change in mismatch. This will eventually limit the cancellation bandwidth and directly contributes to the degradation in cancellation at the bandwidth edges in Fig. 3. The more abruptly changing mismatch comes from the non-flat frequency response of RF components, which exhibit ripples in amplitude and phase. Despite being only 0.4 dB and 1 degree in amplitude and phase mismatch, respectively, this source of mismatch is what is currently limiting the cancellation bandwidth. B. Adaptive Cancellation Using the SOA The system was made adaptive by using feedback from the output to iteratively update the SOA bias current and input optical power. By treating the SOA bias current and input optical power as two independent dimensions, minimizing the interference power becomes an optimization problem. We used the Nelder-Mead simplex algorithm [11] to minimize the interference power because it is a simple direct search method that converges quickly. For a 2-D space, the algorithm analyzes the system output power at three points in the space, forming a triangle (or simplex). During each iteration of the algorithm, the triangle vertex with the highest output power is replaced
CHANG et al.: ADAPTIVE OPTICAL SIC USING AN SOA
1021
The SOA’s range, precision, and speed are suited to track delay variations such as antenna oscillations, which are too fast or small to be compensated by standard delays, yet just as important to achieve high levels of cancellation. In a practical system, the SOA will be used in tandem with longer delays, which would perform coarse adjustments to bring the system close to the optimal delay. The SOA would then be used to lock onto the optimal delay with high precision and speed. Future work involves designing SOAs with a larger delay range and bandwidth. An extended delay range would enable the SOA to track the interference minimum over a wider range of environmental variations, opening doors to more wireless applications. In contrast to prior optical SIC systems, this system can be built only using semiconductors. An integrated SIC system would have high utility and enable more complex filters, such as multipath cancellers. IV. C ONCLUSION
Fig. 4. (a) Progression of the Nelder-Mead simplex algorithm to adaptively cancel interference using the SOA bias current and input optical power. (b) The system is able to locate the interference minimum through 83 ps of delay range. Each curve is averaged over 5 algorithm runs. Insets show receiver output spectra after 0, 40, and 60 algorithm iterations.
We demonstrated an optical self-interference cancellation system, which uses a single SOA to perform precise amplitude and phase matching. The SOA weight and delay are controlled through the SOA bias current and input optical power; the device latency is about 200 ns. The system achieves 38 dB in-band interference cancellation over 60 MHz, and adaptively minimizes the self-interference using the Nelder-Mead simplex algorithm. The SOA has a continuous delay range of 83 ps. The cancellation depth is limited by SOA-induced nonlinearities and noise, and cancellation bandwidth is limited by the remaining RF components in the system.
by a new point, determined by the algorithm, which forms a new triangle. As the algorithm detects improved performance, it contracts in size, converging on the interference minimum. The algorithm was run on a computer, which controlled the SOA bias current and the input optical power, and received feedback from a power meter. A +10 dBm, 10 MHz bandwidth interferer was used along with a single-tone SOI. Figure 4a shows a typical progression as the algorithm adjusts the two parameters to minimize interference power. The initial triangle was arbitrarily chosen and the algorithm was able to minimize the interference (i.e. achieve amplitude and phase matching) in nearly all cases. We measured the rise-time of the SOA to be about 200 ns, limited by the parasitics in the SOA electrodes, making the delay and weight tuning extremely fast. To determine the range of time delays accessible by the SOA, the delay of the transmitter tap was manually detuned from the optimal value by a variable delay line before initiating the algorithm. The same initial triangle was used as in Fig. 4a. The results, shown in Fig. 4b, indicate that the SOA is able to cover an 83 ps continuous delay range, or 26 mm of free space. Cancellation is noise-limited towards longer delays. The noise originates from ASE noise and nonlinearities generated by the SOA under certain biasing conditions; high bias currents and low input powers are particularly susceptible to nonlinearities and noise in this scheme [9].
[1] D. Fiedler, “Cosite interference: Force XXI communications roadblock,” Army Commun., vol. 21, no. 4, pp. 54–59, Sep. 1996. [2] M. Faulkner and S. Ahmed, “Optimized interference canceling for colocated base station transceivers,” IEEE Trans. Veh. Technol., vol. 60, no. 9, pp. 4175–4183, Nov. 2011. [3] D. Bharadia, E. McMilin, and S. Katti, “Full duplex radios,” in Proc. ACM SIGCOMM, Aug. 2013, pp. 375–386. [4] M. Jain et al., “Practical, real-time, full duplex wireless,” in Proc. 17th ACM Int. Conf. Mobile Comput. Netw. (MobiCom), Las Vegas, NV, USA, 2011, pp. 301–312. [5] S. Hong et al., “Applications of self-interference cancellation in 5G and beyond,” IEEE Commun. Mag., vol. 52, no. 2, pp. 114–121, Feb. 2014. [6] R. A. Minasian, “Photonic signal processing of microwave signals,” IEEE Trans. Microw. Theory Techn., vol. 54, no. 2, pp. 832–846, Feb. 2006. [7] M. P. Chang, M. Fok, A. Hofmaier, and P. R. Prucnal, “Optical analog self-interference cancellation using electro-absorption modulators,” IEEE Microw. Wireless Compon. Lett., vol. 23, no. 2, pp. 99–101, Feb. 2013. [8] J. Chang and P. R. Prucnal, “A novel analog photonic method for broadband multipath interference cancellation,” IEEE Microw. Wireless Compon. Lett., vol. 23, no. 7, pp. 377–379, Jul. 2013. [9] M. P. Chang, N. Wang, B. Wu, and P. R. Prucnal, “A simultaneous variable optical weight and delay in a semiconductor optical amplifier for microwave photonics,” J. Lightw. Technol., to be published. [10] S. Sales, W. Xue, J. Mørk, and I. Gasulla, “Slow and fast light effects and their applications to microwave photonics using semiconductor optical amplifiers,” IEEE Trans. Microw. Theory Techn., vol. 58, no. 11, pp. 3022–3038, Nov. 2010. [11] J. A. Nelder and R. Mead, “A simplex method for function minimization,” Comput. J., vol. 7, no. 4, pp. 308–313, 1965.
R EFERENCES
IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 27, NO. 9, MAY 1, 2015
Adaptive Optical Self-Interference Cancellation Using a Semiconductor Optical Amplifier Matthew P. Chang, Member, IEEE, Chia-Lo Lee, Ben Wu, and Paul R. Prucnal, Fellow, IEEE
Abstract— We experimentally demonstrate an optical system that uses a semiconductor optical amplifier (SOA) to perform adaptive, analog self-interference cancellation for radiofrequency signals. The system subtracts a known interference signal from a corrupted received signal to recover a weak signal of interest. The SOA uses a combination of slow and fast light and cross-gain modulation to perform precise amplitude and phase matching to cancel the interference. The system achieves 38 dB of cancellation across 60-MHz instantaneous bandwidth and 56 dB of narrowband cancellation, limited by noise. The Nelder–Mead simplex algorithm is used to adaptively minimize the interference power through the control of the semiconductor’s bias current and input optical power. Index Terms— Self-interference cancellation, co-site interference cancellation, microwave photonics, slow and fast light, semiconductor optical amplifier (SOA).
I. I NTRODUCTION
S
ELF-INTERFERENCE, or co-site interference, is an increasingly common problem as wireless communication platforms grow in complexity but shrink in size [1], [2]. Self-interference occurs when a colocated transmitter and receiver operate simultaneously; the high-power transmitter signal leaks into the receiver, desensitizing the receiver or overwhelming weaker signals of interest from distant antennas. Vehicles with multiple communication systems [1] and shared base stations [2] all suffer from self-interference. Nearly all radios are half-duplex because of self-interference [3]–[5]. Self-interference cancellation (SIC) is difficult even with knowledge of the transmitted signal due to the sheer power of the interferer [5]. As a result, SIC is performed in the antenna, analog, and digital domains of a communication system [3], [4]. In analog cancellation, an inverted copy of the known interferer is combined with the corrupted signal. With precise amplitude and phase matching across the interferer bandwidth, the interference copy cancels the self-interference signal upon combination, recovering the signal of interest. For broadband SIC, the requirement of precise amplitude and phase matching across a wide bandwidth places a heavy burden on radio-frequency (RF) circuits, which have limited bandwidth because of the poor frequency flatness
Manuscript received December 8, 2014; accepted February 16, 2015. Date of publication February 19, 2015; date of current version April 10, 2015. The authors are with the Lightwave Communications Laboratory, Department of Electrical Engineering, Princeton University, Princeton, NJ 08544 USA (e-mail: [email protected]; [email protected]; [email protected]; [email protected]). Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LPT.2015.2405498
of RF components [4]. In particular, RF delay lines and phase shifters are inherently narrowband and lack the precision to perform high levels of SIC [6]. Optics is a natural remedy to the limited bandwidth of electronics [6]. In optical SIC, the interference copy and corrupted signal are both modulated onto optical carriers and processed in the optical domain. By leveraging the tiny fractional bandwidth of an RF-modulated optical signal, optical SIC can demonstrate extremely broadband cancellation. Previous optical SIC systems have achieved 40 dB cancellation over 50 MHz bandwidth [7] and broadband multipath compensation [8]. However, optical delay lines have prevented optical SIC from being used in practical RF environments. Motorized and thermal delays are too slow, requiring 10’s of milliseconds to adjust. In comparison, indoor WiFi has a coherence time of ∼100 ms, making millisecond tuning a costly overhead to pay [3]. A fast, tunable, and high resolution optical delay is needed to realize practical optical SIC. In this letter, we report an optical SIC system, which uses a single semiconductor optical amplifier (SOA) to perform both amplitude and phase matching across a wide bandwidth. The SOA combines slow and fast light with cross-gain modulation (XGM) to implement electronic control of its propagation delay/phase shift and weight, or transmission [9]. The SOA not only enables a fast, precise solid state optical delay to perform SIC, but also moves the system closer towards full photonic integration. With this technique, we demonstrate 38 dB of broadband cancellation over 60 MHz and adaptive interference cancellation. II. E XPERIMENTAL S ETUP The optical SIC system is a two-tap microwave photonic filter, which accepts two inputs: the corrupted received signal, xr , which consists of the signal of interest (SOI) and self-interference, and a copy of the known self-interference, x t xr = s(t) + nr (t)
(1)
x t = n t (t)
(2)
where n t is the original transmitted signal, s is the SOI, and nr is the self-interference signal after propagating through the transmitter-to-receiver wireless channel. To first order, nr will be a delayed and attenuated version of n t . The optical SIC system is designed to use the SOA’s tunable weight and delay to match the wireless channel’s attenuation and delay as closely as possible. In reality, the wireless channel is more complex than an attenuation and delay, and includes amplifier
1041-1135 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
CHANG et al.: ADAPTIVE OPTICAL SIC USING AN SOA
1019
The two signals are combined by a 50/50 optical coupler, and the combined signal is detected, producing the RF output y = s(t) + [nr (t) − αn t (t − τ )]
Fig. 1. Experimental setup. λ1 = 1550.12 nm, λ2 = 1553.33 nm. PC = Polarization Controller, EDFA = Erbium-Doped Fiber Amplifier, VOA = Variable Optical Attenuator BPF = Bandpass Filter.
noise, nonlinearities, and multipath; however, the goal of the current system is to cancel the strongest component of self-inteference, the line-of-sight signal. A schematic of the fiber setup is shown in Fig. 1. The output from a λ1 = 1550.12 nm distributed feedback laser is split in half to provide the optical carrier for each filter tap. The corrupted received signal, xr , is modulated onto the optical carrier of the top tap, called the receiver tap, by a Mach-Zehnder modulator (MZM). The copy of the self-interference, x t , is modulated onto the optical carrier of the bottom tap, called the transmitter tap, by a second MZM. The self-interference and SOI are generated by two distinct signal generators. xr is created by coupling the self-interference and the SOI using an RF combiner. x t is tapped from the self-interference signal generator. The signal in the receiver tap propagates unaltered to a 50/50 optical coupler. The signal in the transmitter tap is optically inverted, delayed, and weighted to match it in amplitude and phase, except with a 180° offset, to the self-interference in the corrupted received signal. The key component to perform this matching is the SOA, which is highlighted in the call-out in Fig. 1. The SOA performs three essential functions. First, it produces an electronically tunable time delay or microwave phase shift through the slow and fast light effect. In an SOA, the slow and fast light effect generates a tunable delay or phase shift by changing the propagating group velocity in the semiconductor using carrier dynamics [10]. Second, the SOA modifies its weight, or transmission, to amplitude-match the interference copy with the actual self-interference signal. The time delay/phase shift and weight of the SOA are controlled through two independent parameters: the SOA bias current and the input optical power, which is controlled through a variable optical attenuator. The technique used to simultaneously control the SOA’s weight and delay is further described in [9]. Third, the SOA performs the necessary inversion of the transmitter tap signal through XGM. Note that XGM is also used to convert the tap wavelength from λ1 = 1550.12 nm to λ2 = 1553.33 nm to prevent coherent beat noise when the transmitter tap signal is combined with the receiver tap signal.
(3)
where α and τ are the SOA’s weight and delay, respectively. If the self-interference and the interference copy are matched exactly in amplitude and phase, they will destructively interfere upon detection. Any amplitude and phase mismatch across the bandwidth of the self-interference signal will result in residual interference, given by the term in brackets in Eqn 3. The output RF signal is characterized by a network or spectrum analyzer to determine how well the system cancelled the self-interference. The output signal is also used as feedback to adaptively adjust the SOA’s bias current and input optical power to minimize the interference. Both broadband and narrowband cancellation were measured by varying the bandwidth of the self-interference signal and examining how well the system could adaptively cancel the self-interference and recover the SOI. III. E XPERIMENTAL R ESULTS AND A NALYSIS A. Broadband and Narrowband Cancellation Figure 2 shows the experimentally measured cancellation after optimizing the SOA’s weight and delay. The output of the system was measured without (red dashed curve) and then with (blue solid curve) the transmitter tap to observe the effect of interference cancellation. The SOI was a weak 915 MHz single-tone signal. The 900 MHz band was the chosen testbed because it contains the crowded industrial, scientific, and medical (ISM) radio bands within the American continents. Other frequency bands are compatible with the photonic hardware; however, the slow and fast light effect decreases with increasing RF frequency [9]. Broadband cancellation was characterized by sweeping a 0 dBm single-tone interferer across a 60 MHz bandwidth centered at 915 MHz and measuring the system output using an RF spectrum analyzer; 60 MHz is greater than most channel bandwidths used in common protocols like WiFi and LTE. The results in Fig. 2a show that at least 38 dB of cancellation was achieved across the 60 MHz bandwidth. More importantly, the SOI, which had been overwhelmed without cancellation, was recoverable with cancellation. Note that the filter removes in-band interference while leaving the SOI unharmed, making this system more effective than simple spectral filtering. The interference was reduced to the spectrum analyzer noise floor at the center frequency in Fig. 2a, suggesting that the maximum measurable cancellation is greater than what is observed. To determine the maximum achievable cancellation, narrowband measurements were performed using a 10 kHz bandwidth interferer and the same SOI as before. The results are plotted in Fig. 2b, showing the system output with and without cancellation. The maximum level cancellation was 56 dB, and the SOI was recovered. We now examine how cancellation changes with bandwidth and what factors affect cancellation bandwidth. Cancellation degrades with increasing bandwidth (see Fig. 2a) because of the inability of the transmitter tap to exactly model
1020
Fig. 2.
IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 27, NO. 9, MAY 1, 2015
(a) Broadband and (b) narrowband cancellation results. The signal of interest is recovered after cancelling the in-band interference.
Fig. 3. (Top) The measured amplitude and phase mismatch between the transmitter-receiver channel and the transmitter tap from 700 MHz to 1100 MHz. (Bottom) Measured cancellation and simulated cancellation based on the amplitude and phase mismatch. More negative cancellation represents greater interference cancellation.
the transmitter-receiver channel across the entire bandwidth. The transmitter tap, a mostly linear system, can only compensate for the propagation loss and delay of the channel. However, frequency response mismatches between the transmitter tap and the channel that cannot be modeled by a simple delay and weight result in residual interference. Much of this mismatch originates from the remaining RF components in the system, which have non-flat frequency responses (e.g. RF splitters), but there is also a contribution from nonlinearities and amplifier noise. We measure the frequency response mismatch, both amplitude and phase, between the transmitter-receiver channel and the transmitter tap across a 400 MHz bandwidth using a network analyzer and plot the mismatch in Fig. 3 (top). Zero mismatch is indicated by the line representing zero amplitude offset and 180 degrees phase offset. The optimal cancellation is achieved when both amplitude and phase mismatch
approach this line, as seen in the bottom half of Fig. 3, which shows experimentally measured cancellation (solid curve) as well as simulated cancellation (dotted curve). The simulatation is performed by treating the system as a two-tap finite-impulse response filter with frequency-dependent weights and delays that correspond to the mismatch measurements in Fig. 3. The simulation corresponds very well to experiment, showing that minimizing mismatch is central to cancellation performance. Figure 3 shows that there are two qualitatively different contributions to amplitude and phase mismatch in this experiment: a gradually changing mismatch over the total bandwidth and more abruptly changing mismatch over frequency ranges much smaller than the total bandwidth. The gradually changing mismatch originates from the limited bandwidth of the slow and fast light effect, which is determined by the resonance used to generate the refractive index dispersion [9], [10]. Because the SOA only exists in one of the two filter taps, the gradual change in the SOA’s delay and transmission with RF frequency manifests as a gradual change in mismatch. This will eventually limit the cancellation bandwidth and directly contributes to the degradation in cancellation at the bandwidth edges in Fig. 3. The more abruptly changing mismatch comes from the non-flat frequency response of RF components, which exhibit ripples in amplitude and phase. Despite being only 0.4 dB and 1 degree in amplitude and phase mismatch, respectively, this source of mismatch is what is currently limiting the cancellation bandwidth. B. Adaptive Cancellation Using the SOA The system was made adaptive by using feedback from the output to iteratively update the SOA bias current and input optical power. By treating the SOA bias current and input optical power as two independent dimensions, minimizing the interference power becomes an optimization problem. We used the Nelder-Mead simplex algorithm [11] to minimize the interference power because it is a simple direct search method that converges quickly. For a 2-D space, the algorithm analyzes the system output power at three points in the space, forming a triangle (or simplex). During each iteration of the algorithm, the triangle vertex with the highest output power is replaced
CHANG et al.: ADAPTIVE OPTICAL SIC USING AN SOA
1021
The SOA’s range, precision, and speed are suited to track delay variations such as antenna oscillations, which are too fast or small to be compensated by standard delays, yet just as important to achieve high levels of cancellation. In a practical system, the SOA will be used in tandem with longer delays, which would perform coarse adjustments to bring the system close to the optimal delay. The SOA would then be used to lock onto the optimal delay with high precision and speed. Future work involves designing SOAs with a larger delay range and bandwidth. An extended delay range would enable the SOA to track the interference minimum over a wider range of environmental variations, opening doors to more wireless applications. In contrast to prior optical SIC systems, this system can be built only using semiconductors. An integrated SIC system would have high utility and enable more complex filters, such as multipath cancellers. IV. C ONCLUSION
Fig. 4. (a) Progression of the Nelder-Mead simplex algorithm to adaptively cancel interference using the SOA bias current and input optical power. (b) The system is able to locate the interference minimum through 83 ps of delay range. Each curve is averaged over 5 algorithm runs. Insets show receiver output spectra after 0, 40, and 60 algorithm iterations.
We demonstrated an optical self-interference cancellation system, which uses a single SOA to perform precise amplitude and phase matching. The SOA weight and delay are controlled through the SOA bias current and input optical power; the device latency is about 200 ns. The system achieves 38 dB in-band interference cancellation over 60 MHz, and adaptively minimizes the self-interference using the Nelder-Mead simplex algorithm. The SOA has a continuous delay range of 83 ps. The cancellation depth is limited by SOA-induced nonlinearities and noise, and cancellation bandwidth is limited by the remaining RF components in the system.
by a new point, determined by the algorithm, which forms a new triangle. As the algorithm detects improved performance, it contracts in size, converging on the interference minimum. The algorithm was run on a computer, which controlled the SOA bias current and the input optical power, and received feedback from a power meter. A +10 dBm, 10 MHz bandwidth interferer was used along with a single-tone SOI. Figure 4a shows a typical progression as the algorithm adjusts the two parameters to minimize interference power. The initial triangle was arbitrarily chosen and the algorithm was able to minimize the interference (i.e. achieve amplitude and phase matching) in nearly all cases. We measured the rise-time of the SOA to be about 200 ns, limited by the parasitics in the SOA electrodes, making the delay and weight tuning extremely fast. To determine the range of time delays accessible by the SOA, the delay of the transmitter tap was manually detuned from the optimal value by a variable delay line before initiating the algorithm. The same initial triangle was used as in Fig. 4a. The results, shown in Fig. 4b, indicate that the SOA is able to cover an 83 ps continuous delay range, or 26 mm of free space. Cancellation is noise-limited towards longer delays. The noise originates from ASE noise and nonlinearities generated by the SOA under certain biasing conditions; high bias currents and low input powers are particularly susceptible to nonlinearities and noise in this scheme [9].
[1] D. Fiedler, “Cosite interference: Force XXI communications roadblock,” Army Commun., vol. 21, no. 4, pp. 54–59, Sep. 1996. [2] M. Faulkner and S. Ahmed, “Optimized interference canceling for colocated base station transceivers,” IEEE Trans. Veh. Technol., vol. 60, no. 9, pp. 4175–4183, Nov. 2011. [3] D. Bharadia, E. McMilin, and S. Katti, “Full duplex radios,” in Proc. ACM SIGCOMM, Aug. 2013, pp. 375–386. [4] M. Jain et al., “Practical, real-time, full duplex wireless,” in Proc. 17th ACM Int. Conf. Mobile Comput. Netw. (MobiCom), Las Vegas, NV, USA, 2011, pp. 301–312. [5] S. Hong et al., “Applications of self-interference cancellation in 5G and beyond,” IEEE Commun. Mag., vol. 52, no. 2, pp. 114–121, Feb. 2014. [6] R. A. Minasian, “Photonic signal processing of microwave signals,” IEEE Trans. Microw. Theory Techn., vol. 54, no. 2, pp. 832–846, Feb. 2006. [7] M. P. Chang, M. Fok, A. Hofmaier, and P. R. Prucnal, “Optical analog self-interference cancellation using electro-absorption modulators,” IEEE Microw. Wireless Compon. Lett., vol. 23, no. 2, pp. 99–101, Feb. 2013. [8] J. Chang and P. R. Prucnal, “A novel analog photonic method for broadband multipath interference cancellation,” IEEE Microw. Wireless Compon. Lett., vol. 23, no. 7, pp. 377–379, Jul. 2013. [9] M. P. Chang, N. Wang, B. Wu, and P. R. Prucnal, “A simultaneous variable optical weight and delay in a semiconductor optical amplifier for microwave photonics,” J. Lightw. Technol., to be published. [10] S. Sales, W. Xue, J. Mørk, and I. Gasulla, “Slow and fast light effects and their applications to microwave photonics using semiconductor optical amplifiers,” IEEE Trans. Microw. Theory Techn., vol. 58, no. 11, pp. 3022–3038, Nov. 2010. [11] J. A. Nelder and R. Mead, “A simplex method for function minimization,” Comput. J., vol. 7, no. 4, pp. 308–313, 1965.
R EFERENCES
