Significance With the advancement of information technology, humans have entered the mobile Internet age. Moreover, surging communication services have resulted in an exponential growth of the capacity and rate of communication systems, spurring the research and development of an ultrahigh-speed and ultralarge-capacity optical communication system. However, high-speed optical signals are susceptible to channel dispersion and nonlinearity, increasing the bit error rate at the receiver and limiting further improvement of the communication rate. Furthermore, as the signal rate increases, the bandwidth demand for the transmitter and detector in the communication system also increases, increasing the overall system cost. To address these challenges, conventional optical communication transmission and detection methods must be improved further to enhance the efficiency of optical communication.

Ghost imaging is a novel dual-arm imaging technology, which has attracted considerable research attention. A correlation algorithm between two spatially separated light beams can be used to reconstruct the spatial intensity information of the target object. The object is illustrated by one beam called the object beam, and the transmitted or reflected light is collected using a bucket detector with no spatial resolution, yielding the total light intensity value. The other beam called the reference beam does not interact with the target object and is directly received using a detector with spatial distribution. Neither of the detectors can produce an image of the object on their own; however, the object will appear after correlation operation for these two beams. Ghost imaging has aroused great interest of researchers because of its unique nonlocality, which has a significant potential application value in noisy and turbulent scenarios where the application of conventional imaging methods is limited.

Ghost imaging has recently been extended from the spatial domain to the temporal domain by investigating the duality of light propagation in space and time, i.e., the correspondence between the diffraction of a light beam and the dispersive propagation of a short optical pulse. Temporal ghost imaging has been proposed in theory and verified in experiment. To reconstruct a temporal object by correlating operation, different temporal sequences must be illuminated and the cumulative energy of the illuminated light must be collected using a bucket detector. Temporal ghost imaging uses a slow-speed detector to detect high-speed optical signals and is insensitive to the damage between the signal and detection. Therefore, it is a promising method for optical signal detection and recovery.

Progress Since the implementation of temporal ghost imaging in an optical fiber-based system (Fig. 1), related research works on long-distance optical fiber detection, high-speed optical signal detection, information security, and camera frame rate have been proposed recently. In the practical application of optical fiber sensing systems, the distance of temporal objects to detector is usually far and the position is changeable. The dual-arm structure reduces the flexibility of the system and increases the cost. To overcome these problems, Tang et al. proposed a method for obtaining the intensity information of a random light source using intensity-only detection, thus realizing a computational time-domain ghost imaging scheme on a single optical fiber (Fig. 3). By exploiting the long-distance optical fiber to achieve time dispersion, Yao et al. proposed the use of a thermal light source to achieve imaging between two remote ends of an optical fiber link (Fig. 5). Although temporal ghost imaging is insensitive to the damage between the object and detector, which provides an opportunity for ultrafast signal detection, its application is limited by the random fluctuation time of the light source and the temporal resolution of the detectors. To realize ultrafast signal detection in various scenarios, researchers have proposed a series of technical solutions. Ryczkowski et al. used an extra-long dispersion fiber in the reference beam path to magnify the time fluctuation of the reference signal by five times, resulting in a reconstructed signal with five times pusle width (Fig. 7). A low response speed of the infrared detector limits the measurement of high-speed signals. Wu et al. transferred light to the spectral region using an ultrafast detector by employing the wavelength conversion concept, thus achieving temporal ghost imaging at a wavelength of 2 μm (Fig. 9). Using computational temporal ghost imaging, Xu et al. could detect fast signals beyond the bandwidth of the detectors by actively modulating the target signal with specific patterns (Fig. 11).

Further, we proposed a dual encryption scheme based on the micro-light-emitting diode (micro-LED) operating ultra bandwidth and a computational temporal ghost imaging algorithm, which was verified in a visible light communication system (Fig. 13). However, repeated measurements reduced the detection efficiency and limited the detection of nonrepeatable or aperiodic signals. To solve these problems, spatial multiplexing technology was introduced, based on which Jiang et al. proposed an information security scheme (Fig. 16). Moreover, Jiang et al. demonstrated a fast imaging scheme by increasing the camera frame rate via temporal ghost imaging.

Conclusions and Prospect In conclusion, temporal ghost imaging opens novel perspectives for dynamic imaging of ultrafast signals in various scenarios; however, it still needs further research and improvement. The development of hardware devices and improvement of the calculation algorithms will promote the development and application of temporal ghost imaging in optical communication, information security and transmission, equipment performance improvement, and other fields.