Low-hardware-complexity optical channel estimation method and device based on cascaded MZM architecture

Abstract

A low-hardware-complexity optical channel estimation method based on a cascade MZM framework comprises a continuous wave laser, a polarizer, a multi-channel arbitrary waveform generator, a Mach-Zehnder modulator, a voltage source, an erbium-doped optical fiber amplifier, an optical attenuator, a photoelectric detector, an oscilloscope and the like, and high-speed and high-parallelism real-time channel estimation is achieved under low hardware complexity. According to the method, pilot signals and subcarrier signals are respectively loaded by using the cascaded MZMs, mapping of radio frequency signals is directly carried out in the optical field, a large amount of matrix calculation in channel estimation is processed in the optical field, microsecond time delay consumed in complex signal processing steps such as digital-to-analog conversion and serial-to-parallel conversion is avoided, and the accuracy of channel estimation is improved. An ultra-low nanosecond delay and multi-channel parallel channel estimation effect is realized; the Fourier matrix is simplified by using the matrix condition number theory, and the complexity of the number of hardware devices and the light splitting loss is reduced on the premise of ensuring the channel estimation accuracy, so that the signal noise is improved to a certain extent, and the communication quality is improved.

Description

Technical Field

[0001] This invention belongs to the field of optical communication technology, specifically a low-hardware-complexity optical channel estimation method and device based on a cascaded Mach-Zernd modulator (MZM) architecture, which can achieve high-speed, high-parallelism real-time channel estimation with low hardware complexity. Technical Background

[0002] In the era of information explosion, the number of communication devices is growing rapidly, various service functions are constantly being updated, and various intelligent applications are emerging continuously. The ever-increasing data volume and increasingly complex application scenarios place high demands on data transmission rates and parallelism. However, traditional channel estimation methods based on electronic processors are struggling to overcome the limitations of the "electronic bottleneck," and improvements in latency, parallelism, and computing power are gradually lagging behind the growth in communication performance requirements. Therefore, in order to better support the ultra-low latency requirements of sixth-generation communication (6G), channel estimation is organically integrated with optical computing technology to achieve faster communication speeds and better communication quality.

[0003] Channel estimation aims to obtain channel state information (CSI) in wireless communication to achieve signal detection and recovery, thereby ensuring communication quality. Currently, traditional channel estimation methods mainly utilize electronic processors for computation, such as using pilot symbols for channel estimation (S. Ali, D. Zheng, and B. Jiao. "A new pilot-shared method for saving bandwidth cost of OFDM," Sci Rep. 14, 4528, 2024). Figure 1 As shown in (a), this process involves a large number of digital matrix calculations, leading to increased communication latency. Furthermore, channel estimation involves complex signal processing steps such as analog-to-digital conversion (ADC / DAC) and serial-to-parallel conversion (SPC / PSC), requiring hundreds of microseconds of time in the electronic processor for data access, processing, storage, and transmission. In the field of optics, optical computing technology has advantages in computing power, latency, and parallelism, enabling ultra-low nanosecond latency and multi-path parallel architectures. For the bottlenecks in speed and parallelism of traditional electronic channel estimation techniques, the advantages of low latency and high parallelism of optical computing can be leveraged, such as… Figure 1(b) shows that a large number of matrix calculations in channel estimation are processed in the optical domain, achieving low-latency optical parallel channel estimation, thus meeting the current requirements of the communication field for continuously improving transmission rates and system latency. The team led by Weiwen Zou at Shanghai Jiao Tong University proposed a channel estimation scheme based on optical matrix calculation, utilizing a cascaded architecture of Mach-Zenith modulators (MZM) to achieve optical parallel channel estimation (Zhao X, et al. "Photonic parallel channel estimation of MIMO-OFDM wireless communication systems", Optics express. 31, 1394-1408, 2023). However, this scheme has excessively high hardware complexity. As the scale of MIMO-OFDM increases, the number of devices increases exponentially. The increasing number of hardware devices (including Mach-Zenith modulators and photodetectors) leads to high system costs, making it difficult to have practical application value. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention proposes a low-hardware-complexity optical channel estimation method based on a cascaded MZM architecture. By employing a cascaded MZM architecture to load pilot and subcarrier signals separately, the matrix multiplication calculation for channel estimation is transferred to the optical domain. Furthermore, matrix condition number theory is used to simplify the Fourier matrix, further reducing the complexity of the channel estimation algorithm. While ensuring the accuracy of channel estimation, this low-complexity channel estimation algorithm can reduce hardware complexity and, to some extent, reduce splitting loss and improve the signal-to-noise ratio, thus making it applicable to larger-scale MIMO-OFDM systems.

[0005] The technical solution of the present invention is as follows:

[0006] A low-hardware-complexity optical channel estimation method based on a cascaded MZM architecture.

[0007] Step 1: Simplify the Fourier matrix based on matrix condition number theory to build a low-complexity mathematical model for channel estimation. While ensuring a small error in the time-domain channel matrix, construct a time-domain parameter calculation model for optical channel estimation.

[0008] Step 2: Based on the principle of calculating time-domain parameters for optical channel estimation, construct a cascaded MZM channel estimation experimental architecture. Then, based on the principle of low-complexity channel estimation, select appropriate subcarriers and calculate the weight matrix.

[0009] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0010] 1) By utilizing a cascaded Mach-Zenith modulator (MZM) architecture, the matrix multiplication calculations in channel estimation, originally performed in the electronic domain, are transferred to the optical domain. This not only leverages the low latency and high parallelism advantages of optical computing but also reduces the burden on electronic processors, thereby improving overall communication efficiency.

[0011] 2) To reduce computational complexity while ensuring channel estimation accuracy, matrix condition number theory is introduced to simplify the Fourier matrix. By reasonably approximating or optimizing the Fourier matrix, the required computational complexity can be significantly reduced while maintaining sufficient channel estimation accuracy.

[0012] 3) This invention reduces the complexity of the number of hardware devices. It not only lowers the system's manufacturing cost and complexity but also improves its scalability and flexibility, enabling it to better adapt to MIMO-OFDM systems of different scales and requirements.

[0013] 4) Reduced splitting loss and improved signal-to-noise ratio: In the process of matrix multiplication calculation in the optical domain, by optimizing the optical path design and device parameters, the splitting loss can be reduced to a certain extent and the signal-to-noise ratio can be improved.

[0014] 5) Real-time optical channel estimation is achieved, which means that the channel state can be estimated and updated in a very short time, thereby meeting the stringent requirements of future communication systems for ultra-low latency. Attached Figure Description

[0015] Figure 1 This diagram illustrates a comparison between traditional electrical channel estimation methods and optical parallel channel estimation methods.

[0016] Figure 2 This is a schematic diagram of the specific architecture of the low hardware complexity optical channel estimation method based on the cascaded MZM architecture in an embodiment of the present invention.

[0017] Figure 3 This is a diagram of the operating device for the low-hardware-complexity optical channel estimation method based on the cascaded MZM architecture, according to an embodiment of the present invention.

[0018] Figure 4 The figures shown are experimental results of the low hardware complexity optical channel estimation method based on the cascaded MZM architecture according to an embodiment of the present invention, including signal-to-noise ratio-bit error rate curves and constellation diagrams. Among them, (a) is the bit error rate result using QPSK modulation at a carrier frequency of 1 GHz, (b) is the bit error rate result using 16QAM modulation at a carrier frequency of 1 GHz, (c) is the bit error rate result using QPSK modulation at a carrier frequency of 4 GHz, and (d) is the bit error rate result using 16QAM modulation at a carrier frequency of 4 GHz. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments, but this should not be construed as limiting the scope of protection of the present invention.

[0020] A low-hardware-complexity real-time optical channel estimation method based on wireless communication includes a continuous-wave laser, polarizer, multi-channel arbitrary waveform generator, Mach-Zehnder modulator, voltage source, erbium-doped fiber amplifier, optical attenuator, photodetector, oscilloscope, etc., enabling high-speed, high-parallelism real-time channel estimation with low hardware complexity. This method processes a large number of matrix calculations in the optical domain, mainly consisting of a cascaded Mach-Zehnder intensity modulator, photodetector, and optical attenuator.

[0021] Step 1:

[0022] 1.1 Constructing a low-complexity channel estimation matrix based on matrix condition number theory

[0023] In a MIMO-OFDM system, the transmitting antenna M T and receiving antenna M R The channel model between them can be represented as

[0024]

[0025] Where N is the number of subcarriers in the OFDM symbol, and the transmitted signal X i (k) and received signal Y j (k) is the digital symbol loaded on the k-th subcarrier, where X i =(X i [0],...,X i [N-1]) T and Y j =(Y j [0],...,Y j [N-1]) T Z j (k) is Gaussian white noise on the subcarrier, where Z j =(Z j [0],...,Z j [N-1]) T H i,j (k) represents the channel state information on the k-th subcarrier from transmit antenna i to receive antenna j. For the multichannel model, the time-domain channel matrix h i,j With the frequency domain channel matrix H i,j The relationship between them can be represented as

[0026]

[0027] Where L is the number of actual channel paths, F n,l(n = 0, 1, 2, ..., N-1) are the elements of the Fourier matrix, with values ​​e. -j2πnl / N .

[0028] Since the channel parameters are correlated in both the time and frequency domains, the above linear equations have a unique solution. Because the number of equations N is much larger than the number of unknowns L, the problem can be simplified to solving L linearly independent equations. Since the Fourier transform matrix is ​​a Vandermonde matrix, any L chosen equations are linearly independent; that is, any L chosen equations can satisfy the remaining NL equations, thus simplifying the Fourier matrix from N×L dimensions to L×L dimensions.

[0029] In practical communication systems, the frequency domain channel matrix H i,j The presence of noise will cause some disturbance to the results. Randomly selecting L equations will lead to an increase in the time-domain channel matrix h. i,j There is a significant error. Taking this error into account, the frequency domain channel matrix H... i,j With the time-domain channel matrix h i,j The relationship between them is represented as

[0030] H L,1 +σH L,1 =F L，L (h L,1 +σh L,1 )

[0031] Where σH L,1 and σh L,1 These represent the errors in the frequency-domain channel matrix and the time-domain channel matrix, respectively. To reduce time-domain errors, according to matrix condition number theory, the condition number in a Fourier matrix can be defined as...

[0032]

[0033] Where ||F L,L || represents the matrix L 2 Norm. Obtained from the triangle inequality.

[0034]

[0035] The matrix condition number reflects h i,j For H i,j The sensitivity of h is such that the smaller the matrix condition number, the lower the sensitivity. i,j Subject to H i,j The smaller the impact of fluctuations, the better, even in the frequency domain channel H i,j Due to noise interference, the time-domain channel h i,j It won't deviate too far from the true value. Assuming N is a composite number, according to the prime factorization theorem, it can be factored into the form m × L. To ensure K(F... L,LTo minimize the error, we take L rows and all L columns of the original Fourier matrix at intervals m to form a new matrix. Under this condition, the matrix condition number is 1, thus obtaining the channel state information with the minimum error. Based on the simplified Fourier matrix, h i,j With H i,j The relationship between them can be represented as

[0036]

[0037] 1.2 The simplified channel estimation matrix is ​​used for calculating the time-domain parameters of optical channel estimation.

[0038] By constructing the synthesis matrix To express weight information, the MIMO-OFDM-based communication process can be transformed into...

[0039] Y j =Wh j +Z j (j = 1, 2, ..., M) R )

[0040] The time-domain channel estimation result obtained using the least squares algorithm can be expressed as follows:

[0041]

[0042] In an OFDM-MIMO system, the radio frequency signal in the m-th OFDM signal can be represented as:

[0043]

[0044] Where s m (k) represents the QAM / PSK signal modulated on the k-th carrier in the m-th OFDM signal. At the transmitter, f k The value is f c +kΔf(k=0,1,...,N-1), where f c Let represent the carrier frequency, Δf represent the subcarrier spacing, and the m-th OFDM symbol refer to the time interval from (m-1)T to mT, where T = 1 / Δf represents the period of one OFDM symbol. Based on the orthogonality of OFDM signals, the baseband symbols transmitted by the transmit antenna can be demodulated as follows:

[0045]

[0046] When the transmitting antenna inserts a pilot signal in the first OFDM symbol, the symbol carried by the k-th carrier of the receiving antenna j can be represented as:

[0047]

[0048] The formula for calculating time-domain parameters in optical parallel channel estimation can be expressed as follows:

[0049]

[0050] Where w l,i It is the weight matrix W -1 The sum of the elements in the l-th row.

[0051] Step 2: Based on the cascaded MZM architecture, the multiplication calculation of the channel estimation matrix is ​​transferred to the optical domain.

[0052] Based on the above formula for calculating time-domain parameters in optical parallel channel estimation, Figure 2 This paper presents the architecture of a simplified, low-hardware-complexity real-time optical channel estimation method. A cascaded MZM architecture is used to load pilot and subcarrier signals onto different MZMs. The multiplication calculation of the channel estimation matrix is ​​transferred to the optical domain through nonlinear effects and modulation processes in the optical domain. In the optical parallel channel estimation sub-architecture, the first-stage MZM directly receives the RF signal, avoiding the time delay caused by various signal conversions. The second-stage MZM divides the channel into L branches (much smaller than the total number of subcarriers N in an OFDM signal) to load L different subcarrier demodulated signals. Subsequently, an optical attenuator loads the channel estimation weight information, and finally, a photodetector converts the optical signal into an electrical signal, thereby transferring the multiplication calculation of the channel estimation matrix to the optical domain.

[0053] Step 3: Select appropriate subcarriers during the optical domain computation and calculate the corresponding weight matrix.

[0054] like Figure 3 As shown, to verify the feasibility of this architecture, an experimental verification was established based on a 2×2 OFDM-MIMO wireless communication system, which included four subcarrier signals. Based on the channel estimation matrix under low complexity, the first and third subcarrier signals were selected and loaded into the second-level MZM. The weight matrix loaded into the third-level MZM was calculated based on the synthesis matrix W as (0.5 0.5; 0.5 - 0.5). Compared to the original scheme, this invention avoids dividing the channel path into four parts on the second-level MZM, thus avoiding hardware redundancy. In terms of specific signal settings, the carrier frequencies of the input second-level MZM RF signals were set to 1 GHz and 4 GHz, with a frequency interval of 100 MHz. At a sampling rate of 60 GHz, signal generation, processing, and acquisition were implemented using the corresponding experimental setup. Since the experimental setup AWG only supports real numbers, when loading complex-form signals and weight information into the MZM, the real and imaginary parts should be separated, i.e., the experimental setup should be reused twice.

[0055] Step 4: Obtain the time-domain channel state information for each path based on subcarrier selection and weight matrix.

[0056] Based on a cascaded MZM architecture, the optical parallel channel estimation architecture will receive the radio frequency signal y j (t) is modulated onto an optical carrier, and then optical computing technology is used to combine the selected first subcarrier signal / third subcarrier signal and the calculated weight information w. l,i By multiplying the two signals and then adding them together and integrating them in the time domain over the 0-T time period, the time-domain channel state information for each path can be obtained.

[0057]

[0058] Table 1

[0059] Table 1 shows the root mean square error between experimental and simulation results in different frequency bands of the time-domain channel matrix of the low-hardware-complexity optical channel estimation method based on the cascaded MZM architecture in this embodiment of the invention: the root mean square error between experimental and simulation results in different frequency bands of the time-domain channel matrix is ​​all within 10. -4 about.

[0060]

[0061] Table 2

[0062] Table 2 is a simplified comparison of the hardware equipment and splitting loss complexity of the low hardware complexity optical channel estimation method based on the cascaded MZM architecture in the embodiments of the present invention: the complexity of hardware equipment and splitting loss can be reduced from O(N) to O(1).

[0063] Experiments show that the proposed real-time optical channel estimation method based on wireless communication, with low hardware complexity, directly maps radio frequency signals in the optical domain, processing a large number of matrix calculations in channel estimation in the optical domain, and eliminating complex signal processing steps such as digital-to-analog conversion and serial-to-parallel conversion. This invention leverages the advantages of optical computing technology in terms of computing power, latency, and parallelism to achieve an ultra-low nanosecond latency and multi-path parallel channel estimation architecture, aligning with the current trend of "optoelectronic convergence." Table 1 shows that the root mean square error between experimental and simulation results for different frequency bands in the time-domain channel matrix is ​​all within 10. -4 Around 100°C, which can meet the accuracy requirements of channel estimation. Figure 4 The experimental results for different carrier frequencies under different modulation schemes are shown in the diagram. The bit error rate after channel estimation is significantly lower than the forward error correction threshold, which can fully meet the reliability requirements of wireless communication systems. The constellation diagram more intuitively demonstrates that this invention can enhance the resolution of baseband symbols, enabling symbols to be restored to their original quadrants.

[0064] Meanwhile, this invention proposes a low-hardware-complexity real-time optical channel estimation method based on wireless communication, reducing the N branches in the second-stage MZM to L branches (the actual number of channel paths L is much smaller than the total number of subcarriers N in the OFDM signal), thereby achieving the lowest system cost by reducing hardware complexity. As shown in Table 2, due to the reduction in the number of splitters in the second-stage MZM, the complexity of hardware devices (including Mach-Zernd modulators, photodetectors, etc.) and splitting losses can be reduced from O(N) to O(1), thereby correspondingly improving the signal-to-noise ratio and communication quality. In practical MIMO applications, N is usually between 100 and 1000, while L is usually around 10. Therefore, this invention can reduce system cost by tens or even hundreds of times without changing the system architecture as the number of subcarriers N increases. With the improvement of optoelectronic integrated chip technology and other related algorithms, the low-hardware-complexity real-time optical channel estimation method based on wireless communication is expected to be used in large-scale MIMO-OFDM systems to achieve higher speed and parallelism.

Claims

1. A low-hardware-complexity optical channel estimation method based on a cascaded MZM architecture, characterized by: Step 1. Use matrix condition number theory to simplify the Fourier matrix to construct a low-complexity channel estimation matrix, and obtain the optical channel estimation time-domain parameters based on the simplified Fourier matrix; Step 2. Using a cascaded MZM architecture, the pilot signal and subcarrier signal are loaded onto different MZMs respectively. Through nonlinear effects and modulation processes in the optical domain, the multiplication calculation of the channel estimation matrix is ​​transferred to the optical domain. Step 3. In the optical domain calculation process, select appropriate subcarriers and calculate the corresponding weight matrix. The subcarriers and weight matrix are determined based on the channel estimation matrix obtained in Step 1 and the calculation principle of the optical channel estimation time domain parameters. Step 4. Based on the subcarrier selection and weight matrix calculation in Step 3, the channel estimate is obtained, thereby obtaining the time-domain channel state information on each path and realizing optical channel estimation technology with low complexity.

2. The method according to claim 1, characterized in that, It also includes analyzing and comparing channel estimation results under different frequency bands and modulation schemes to evaluate different communication conditions.

3. An apparatus for implementing the low-hardware-complexity optical channel estimation method based on a cascaded MZM architecture as described in claim 1 or 2, characterized in that, The device shown in Figure 3 includes, from front to back, a continuous wave laser (1), a polarizer (2), a multi-channel arbitrary waveform generator (3), a first-stage MZM (4), a second-stage MZM (5), a voltage source (6), a third-stage MZM (7), an erbium-doped fiber amplifier (8), an optical attenuator (9), a photodetector (10), and an oscilloscope (11). The continuous wave laser (1) generates an optical signal, which is then maximized by the polarizer (2); The multi-channel arbitrary waveform generator (3) generates radio frequency signals and subcarrier signals, which are then loaded onto the first-stage MZM (4) and the second-stage MZM (5) through two channels, respectively. The voltage source (6) generates weight information for the third-level MZM; The first-stage MZM (4) and the second-stage MZM (5) receive radio frequency signals and subcarrier signals from the multi-channel arbitrary waveform generator (3) respectively and modulate them; the third-stage MZM (7) receives weight information from the voltage source (6) and modulates it. The modulated optical signal is transmitted sequentially through the erbium-doped fiber amplifier (8) and the optical attenuator (9) to the photodetector (10) and converted into an electrical signal; The electrical signals acquired by the oscilloscope at different carrier frequencies are accumulated, and the final time-domain channel information is obtained by time-domain integration.

Images