Qingqing Meng

Self-interference (SI) influenced the recovery of the signal of interest (SOI) in In Band Full Duplex (IBFD) radio-over-fiber (ROF) links. The multipath effect and nonlinear distortion aggravate the difficulty of the SI signal elimination. A fiber-transmission-assisted digital self-interference cancellation (DSIC) scheme is proposed to solve the problem at a low time complexity. The key to removing the impact of harmful factors is combined with the advantages of photonics architecture and digital filter algorithm. A fiber-transmission-assisted scheme achieves the transmission of reference signals with all the linear and nonlinear features which transform the nonlinear problem into a linear question for DSIC subsequently. In DSIC, a fast transversal recursive least-squares (FTRLS) algorithm realizes a fast-channel estimating process when the SOI is involved. For 10km fiber transmission, exceeding 38 dB and 34 dB cancellation depth over 50 MHz bandwidth in the X band and Ku band are experimentally demonstrated, and 25 Mbaud 16-quadrature amplitude modulation (16-QAM) SOI is successfully recovered after SIC. In addition, FTRLS merely consumed 7N+14 multiplication times at each iteration (filter order N) in the channel estimating process, which saves multiplications amount above 86% than RLS, DNN, Volterra-RLS algorithm, when N is equal to 80.

In this paper, a photonic-enabled image rejection mixer (IRM) that features an ultrawideband self-interference cancellation (SIC) function and a compact configuration is proposed. The parameter tuning of SIC is realized in an optical domain, which avoids the use of electrically tuned devices with limited bandwidth and precision, so that high-precision parameter matching can be realized in the optical domain to realize deep and ultrawideband SIC. The key point of image rejection (IR) is to construct a pair of orthogonal local oscillation (LO) signals through DC-bias-induced phase shift. This not only avoids a high-frequency electrical 90-degree hybrid coupler (HC) applied in the traditional Hartley structure, but also compensates the phase deviation in the electrical intermediate frequency (IF) 90-degree HC flexibly, ensuring wideband and deep IR operation. The simulation results show that the proposed IRM can achieve ultrawideband SIC and IR with the simultaneous high-efficiency recovery of useful signals. They also verify that the scheme has good resistance to strong interference, and can cope with the phase imbalance of the IF 90-degree electrical HC, ensuring the good performance of the system, which has a wide application prospect in various in-band full-duplex (IBFD) systems.

A novel scheme that can simultaneously measure the Doppler frequency shift (DFS) and angle of arrival (AOA) of microwave signals based on a single photonic system is proposed. At the signal receiving unit (SRU), two echo signals and the reference signal are modulated by a Sagnac loop structure and sent to the central station (CS) for processing. At the CS, two low-frequency electrical signals are generated after polarization control and photoelectric conversion. The DFS without direction ambiguity and wide AOA measurement can be real-time acquired by monitoring the frequency and power of the two low-frequency electrical signals. In the simulation, an unambiguous DFS measurement with errors of ±3 × 10⁻³ Hz and a −90° to 90° AOA measurement range with errors of less than ±0.5° are successfully realized simultaneously. It is compact and cost-effective, as well as has enhanced system stability and improved robustness for modern electronic warfare systems.

A photonic approach to the cancellation of self-interference in the optical domain with fiber dispersion immunity and harmonic frequency down-conversion function is proposed based on an integrated, dual-parallel, dual-drive Mach–Zehnder modulator (DP-DMZM). A dual-drive Mach–Zehnder modulator (DMZM) is used as an optical interference canceller, which cancels the self-interference from the impaired signal before fiber transmission to avoid the effect of fiber transmission on the cancellation performance. Another DMZM is used to provide carrier-suppressed, local-oscillation (LO)-modulated, high-order double optical sidebands for harmonic frequency down-conversion to release the strict demand for high-frequency LO sources. By regulating the DC bias of the main modulator, the signal of interest (SOI) can be down-converted to the intermediated frequency (IF) band after photoelectric conversion with improved frequency-conversion efficiency, immunity to the fiber-dispersion-induced power-fading (DIPF) effect, and effective signal recovery. Theoretical analyses and simulation results show that the desired SOI in the X and K bands with a bandwidth of 500 MHz and different modulation formats can be down-converted to the IF frequency. The self-interference noise with the 2 GHz bandwidth is canceled, and successful signal recovery is achieved after a 10 km fiber transmission. The recovery performance of down-converted signals and the self-interference cancellation depth under different interference-to-signal ratios (ISRs) is also investigated. In addition, the compensation performance of DIPF is verified, and a 6 dB improvement in frequency conversion gain is obtained compared with previous work. The proposed scheme is compact, cost-effective, and thus superior in wideband self-interference cancellation, long-range signal transmission, and effective recovery of weak desired signals.

A novel scheme that can simultaneously measure the Doppler frequency shift (DFS) and angle of arrival (AOA) of microwave signals is proposed. At the signal receiving unit (SRU), two echo signals and the reference signal are modulated by a Sagnac loop structure and sent to the central station (CS) for processing. At the CS, two low-frequency electrical signals are generated after polarization control and photoelectric conversion. The DFS without direction ambiguity and wide AOA measurement can be real-time acquired by monitoring the frequency and power of the two low-frequency electrical signals. In the simulation, an unambiguous DFS measurement with errors of ±3×10-3 Hz and a -90° to 90° AOA measurement range with errors of less than ±0.5° are realized. The safety and robustness of the system to environmental disturbance are improved, and it is more suitable for the modern electronic warfare system.

A novel photonic-assisted system for realizing the simultaneous measurement of Doppler frequency shift (DFS) and angle of arrival (AOA) is proposed. It is a simplified structure that is based on a dual-polarization dual-drive Mach–Zehnder modulator (DPol-DDMZM). The DFS direction can be accurately determined by comparing the phase relationship between the upper and lower low-frequency signal waveforms. An amplitude comparison function (ACF), which constructs a one-to-one mapping between the power ratio and the phase difference of echo microwave signals, can reduce the AOA measurement errors. The simulation results show that the simulated ACFs agree well with the theoretical ACFs, and the measurement errors of AOA are less than ±2.8° in a range of 0° to 78°. Moreover, the DFS can be realized by analyzing the spectrum of the low-frequency electrical signal with a measurement error of less than ±0.05 Hz. The system structure is compact and cost-effective, which provides an alternative solution for modern radar and electronic warfare receivers.

In this paper, a novel and efficient photonic-assisted remote frequency measurement (RFM) system with a significantly simplified structure and flexible operation range is proposed. By simply changing the dispersion coefficient or length of the dispersion medium in the central station (CS), the microwave frequency measurement range in the remote antenna unit (RAU) can be tuned. In this system, the RAU and the CS is separated to ensure the concealment and safety of the signal processing unit. The measurement range of the RFM system can be tuned easily during the measurement process without system reconstruction, and different RAUs located at different places can be controlled to work at the same measurement range. The simulation results show that a frequency measurement over the high frequency range (>18 GHz) can be achieved with a measurement error better than ±0.2 GHz. Noteworthy, the impact of the non-ideal factors such as bias drift, intensity noise, phase noise, the equivalent deviation of the polarization beam splitter (PBS), and the dispersion value of the single mode fiber (SMF) is also discussed. It has been proved that they have little influence on the system performance over the high frequency range.

Human action recognition has received much research attention in recent years. Discriminative features learning is important for modeling the spatial and temporal evolutions of different actions. Recent works attempted to use RNN/LSTM to learn spatio-temporal features of joints. However, the features of joints are not fully explored in these conventional LSTMs, where each joint only receives context information from itself from previous frames. In this paper, we propose a LSTM network model by using aggregated feature learning strategy to extend feature representation in both spatial and temporal domains. In this model, the features of a joint are aggregated not only from all joints at previous moments, but also at current moment. Therefore, action-related features within the human skeleton sequences can be better represented and analyzed. To further optimize the network, we present a reparameterization that involves batch normalization and sparsemax classifier for the proposed LSTM network. We evaluate our approach on NTU RGB+D benchmark dataset. Experiment results demonstrate the effectiveness of the proposed approaches.