With the continuous development of 6G technology and a series of high-speed new services such as big data, cloud computing, and autonomous driving, data can only be transmitted for a higher transmission rate and a lower transmission delay. Meanwhile, it is necessary to pursue high transmission quality to ensure sound communication performance, and high-precision speed measurement, ranging, imaging, and wide-area perception are also required. Under different application scenarios and changes in various performance indicators, further breakthroughs should be achieved in communication, with better results yielded than traditional communication systems. The terahertz frequency band integrates the advantages of microwave communication and optical communication and features a high transmission rate, large capacity, strong directionality, high security, and good penetration. On the one hand, in radar applications, since its wavelength is very short at about 30 μm-3 mm, much smaller than the wavelength of microwave and millimeter waves, it can be employed to detect smaller targets and realize more accurate positioning. On the other hand, it has a wide range of frequencies and a very broad bandwidth to transmit nanosecond and picosecond pulses at thousands of frequencies. As a new sampling theory, compressed sensing needs to be quickly applied to various fields, such as speech coding, image processing, and radar detection. By taking advantage of the sparsity of the signal, the signal sampling frequency will be much smaller than the minimum signal sampling rate required in Nyquistian’s theorem, and then a discrete sample of the signal will be obtained by random sampling. Finally, the signal is reconstructed by a nonlinear algorithm. Among them, the field of compressed sensing and millimeter wave channel estimation has also yielded good results. To achieve higher spectral efficiency and restore high-quality signals, we adopt the compressed sensing method to process the integrated signals in the integrated communication and sensing system to achieve high-speed and high-quality signal transmission and sensing.

Digital signal processing is performed via MATLAB to generate communication signals and radar perception signals, which are converted into analog signals using digital-to-analog conversion devices, with the converted analog signals utilized to drive I/Q modulators. The I/Q modulator consists of two parallel Mach-Zehnder modulators. Then, the modulated signal is coupled with the external cavity lasers by the optically neutralized coupler, and the coupled signal is beaten via the photodiode. In the communication part, the signal is mixed with the local oscillator to downgrade the signal to the mid-frequency domain. The generated IF signal is converted into a digital signal via an analog-to-digital converter, and then the communication signal is recovered after digital signal processing. In the radar perception part, the integrated signal modulated by frequency division multiplexing is divided into upper and lower beams by the optical beam splitter, the upper beam is transmitted from the transmitter to space, and the receiver receives the echo signal. The echo signal and the downbeam signal are mixed and deskewed to obtain a single-frequency signal, which is adopted to drive the first MZM. Meanwhile, the pseudo-random sequence is employed to drive the second MZM modulator, and then the modulated signal is beaten and filtered to realize the Nyquist sampling process of the deoblique signal. Finally, the signal is reconstructed by the orthogonal matching pursuit algorithm to obtain the original signal.

In the experiment, the communication transmission of 16QAM signal under 8 Gbaud or the communication rate of 32 Gbit/s is realized (Fig. 11). When the measurement distance is no more than 1.6 m, the error distance is kept within 2 cm (Fig. 16). In the 1 m wireless simulation experiment, when the power ratio of the communication signal to the sensing signal is 7, the power of the communication signal is much greater than that of the sensing signal. Thus, the performance of the communication link is the best, the bit error rate is 0.0093, and the distance measurement error is large because the power of the sensing signal is too low with an error of 4.12 cm. Under the power ratio of 0.125, the power of the perception signal is much greater than that of the communication signal, and the measurement distance error is 0.413 cm. However, due to the low power of the communication signal currently, its performance is poor with a bit error rate of 0.0633. The power ratio at the intersection of the two curves is 0.75, which is the best coexistence of communication and perception in the simulation experiment (Fig. 18).

We propose an integrated optical terahertz synaesthesia system based on compressive sensing and theoretically analyze the principle of integrated signal generation and reception based on compressed sensing. Based on the joint simulation of MATLAB and VPI, the communication and perception performance of the integrated signal, the performance boundary of the integrated synaesthesia, and the influence of the compressed sensing algorithm on the perception accuracy are analyzed. The results show that the 16QAM-LFM signal can achieve a communication rate of 32 Gbit/s, and the bit error rate is lower than the soft decision threshold of forward error correction. In the compressive sensing simulation experiment, when the data compression ratio is 4 and the measurement distance is less than 1.6 m, the distance measurement error is less than 2 cm. Meanwhile, due to the data volume compression, the performance requirements of the analog-to-digital converter are lowered.