Large dynamic range signal detection method for underwater wireless optical OOK communication system

Abstract

The present invention belongs to the field of underwater wireless optical communication and signal detection. Disclosed is a large dynamic range signal detection method for an underwater wireless optical OOK communication system. The method comprises: performing upsampling on a received signal, extracting received signal sampling point vectors corresponding to symbol 0 and symbol 1 in a pilot, and respectively estimating mean values and covariance matrices of the received signal sampling point vectors corresponding to symbol 0 and symbol 1; establishing an optimization problem on the basis of the estimated mean values and covariance matrices of the received signal sampling point vectors, and using the Rayleigh theorem to solve the optimization problem to obtain an optimal dimensionality reduction vector; and computing the inner product of the dimensionality reduction vector and the sampling point vectors corresponding to the pilot signal, statistically determining probability distribution functions of the received signal corresponding to symbols 0 and 1 after the dimensionality reduction, and obtaining the intersection point of the two probability distribution functions as an optimal decision threshold for performing symbol detection on the dimensionality-reduced signal. The present invention can improve the detection performance with low complexity, thus improving the reliability of underwater wireless optical OOK communication systems.

Description

A method for detecting large dynamic range signals in an underwater wireless optical OOK communication system Technical Field

[0001] This invention belongs to the field of underwater wireless optical communication and signal detection, and particularly relates to a method for detecting large dynamic signals in an underwater wireless optical OOK communication system. Background Technology

[0002] With the development of communication technology, the massive amounts of data brought about by the intelligent era have placed demands on communication systems for lower latency and higher speeds. As a supplement to radio frequency (RF) communication, wireless optical communication offers advantages such as wide bandwidth, low power consumption, high speed, resistance to electromagnetic interference, and good confidentiality, attracting increasing research interest from both academia and industry. With the development of underwater information technology, underwater wireless optical communication is playing an increasingly important role. Compared to the high latency of underwater acoustic communication and the high attenuation of RF signals, underwater wireless optical communication has significant advantages in high-speed communication at the hundred-meter level due to the blue-green light absorption window in seawater. Current research on underwater optical communication mainly focuses on long-distance high-speed underwater wireless optical communication, link alignment issues at both ends of the receiver and transmitter, advanced modulation and demodulation and channel coding techniques, underwater channel models and experiments, underwater sensor networks and energy harvesting, and integrated air-space-ground-sea information networks. All of these rely heavily on the accuracy of signal detection at the receiving end.

[0003] In underwater wireless optical communication systems, receiver signal detection is an essential component. Accurate signal detection helps the receiver better recover transmitted data. Underwater transmission conditions are extremely complex, and multiple factors can interfere with communication efficiency. First, due to the absorption and scattering effects of water, the light signal passing through the water body is significantly attenuated, especially over long distances, where the signal strength becomes very weak. Second, the continuous flow of seawater leads to uneven temperature and humidity distribution, causing non-uniformity in the refractive index of light in the water, which in turn generates turbulence, affecting the stable propagation of the light signal. Therefore, the complex underwater environment not only presents challenges to the communication system but also poses a series of challenges to signal detection.

[0004] In underwater communication scenarios, photomultiplier tubes (PMTs) are commonly used as receiving devices. In low-light (high attenuation) scenarios, the number of photons detected within one symbol period is generally modeled as a Poisson distribution; in high-light (low attenuation) scenarios, the mean of signal sampling points within one symbol period is generally modeled as a Gaussian distribution. Therefore, existing underwater wireless optical OOK communication systems generally use the Poisson Maximum Likelihood Detection (PMLD) method based on a Poisson distribution and the Mean Power Detection (MPD) method based on a Gaussian distribution for signal detection. Both of these methods have limitations and do not meet the detection requirements of underwater high-dynamic scenarios. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for detecting high dynamic range signals in an underwater wireless optical OOK communication system. The detection scheme is simple in form and has low complexity, making it suitable for underwater high dynamic range light intensity scenarios. This method can improve the signal detection performance and reliability of the underwater wireless optical OOK communication system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for detecting large dynamic range signals in an underwater wireless optical OOK communication system includes the following steps:

[0008] Step 1: Upsample the received signal by M times to obtain the sampling point vector. Estimate the mean and covariance matrix of the received sampling point vectors corresponding to symbol 0 and symbol 1 based on the sampling point vectors of the received signal corresponding to the pilot.

[0009] Step 2: Based on the estimated mean and covariance matrix of the received sampling point vector, establish an optimization problem, solve the optimization problem using Rayleigh's theorem, and find the optimal dimension reduction vector;

[0010] Step 3: Take the inner product of the received signal sampling point vector obtained in Step 1 and the optimal dimension reduction vector to obtain the dimension-reduced received signal. Statistically calculate the probability distribution functions of the received signals corresponding to the pilot symbols 0 and 1 in the dimension-reduced received signal. Find the intersection of the two probability distribution functions as the optimal decision threshold, which is used to perform symbol detection on the dimension-reduced received signal.

[0011] The beneficial effects of this invention are as follows:

[0012] (1) This invention solves the problem of poor transmission performance of underwater wireless optical OOK communication system from the perspective of signal detection. The detection method is simple and has low complexity. It is applicable to large dynamic scenarios with various light intensities and is more suitable for underwater wireless optical OOK communication system.

[0013] (2) The present invention provides a signal detection method with better performance than traditional methods, which can improve detection performance with lower complexity and improve the reliability of underwater wireless optical OOK communication system. Attached Figure Description

[0014] Figure 1 is a block diagram of the communication system of the large dynamic signal detection method of the underwater wireless optical OOK communication system of the present invention.

[0015] Figure 2 is a schematic diagram comparing the bit error rate performance under 5x upsampling provided in the embodiment of the present invention. Detailed Implementation

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] This invention discloses a large dynamic range signal detection method for an underwater wireless optical OOK communication system, mainly comprising the following steps: upsampling the received signal, extracting the received sampling point vectors corresponding to symbols 0 and 1 in the pilot signal, and estimating the mean and covariance matrices of the received sampling point vectors corresponding to symbols 0 and 1 respectively; establishing an optimization problem based on the estimated mean and covariance matrices of the received sampling point vectors, solving the optimization problem using Rayleigh's theorem, and finding the optimal dimension-reduced vector; taking the inner product of the dimension-reduced vector and the sampling point vectors corresponding to the pilot signal, statistically obtaining the probability distribution functions of the received signals corresponding to symbols 0 and 1 after dimension reduction, and finding the intersection of the two probability distribution functions as the optimal decision threshold for symbol detection of the dimension-reduced signal. Specifically,

[0018] Step 1: Upsample the received signal by M times to obtain the sampling point vector. Estimate the mean and covariance matrix of the received sampling point vectors corresponding to symbol 0 and symbol 1 based on the sampling point vectors of the received signal corresponding to the pilot. This includes the following steps:

[0019] Step 1.1: Assuming the transmitter sends a signal for N symbol times, the analog signal at the receiver during the i-th symbol time is upsampled by a factor of M to obtain the sampling point vector during the i-th symbol time. ,element This represents the value of the k-th sampling point within the i-th symbol time. , The superscript T denotes the transpose of the matrix;

[0020] Step 1.2: Extract the pilot sampling point vector from the sampling point vector obtained in Step 1.1. Let be the set of sampling point vectors corresponding to the symbol 0 in the pilot signal, where the sampling point vectors are represented as . ; Let be the set of sampling point vectors corresponding to symbol 1 in the pilot signal, where the sampling point vectors are represented as Estimate the mean of the sampling point vectors corresponding to symbol 0 and symbol 1 respectively. and :

[0021] ,

[0022] ,

[0023] in, Represents the number of elements in the set;

[0024] Step 1.3: Estimate the covariance matrix of the sampling point vectors corresponding to symbol 0 and symbol 1 respectively. and :

[0025] ,

[0026] .

[0027] Step 2: Based on the estimated mean and covariance matrix of the received sampling point vector, establish an optimization problem, solve the optimization problem using Rayleigh's theorem, and find the optimal dimension-reduced vector, including the following steps:

[0028] Step 2.1, Assume the dimension-reduced vector of the received signal sampling point vector is... Then, after dimensionality reduction, the mean and variance corresponding to the symbol 0 are respectively and ;

[0029] After dimensionality reduction, the mean and variance corresponding to symbol 1 are respectively and :

[0030] ,

[0031] ;

[0032] Step 2.2: With the objective of minimizing the intra-class difference between symbols 0 and 1 after dimensionality reduction and maximizing the inter-class difference between symbols 0 and 1, solve the following optimization problem:

[0033] ,

[0034] in, Represents the optimal dimensionality reduction vector. Representative function The vector corresponding to the maximum value ,function Represented as:

[0035] ,

[0036] In the above formula, the denominator represents the intra-class differences between symbols 0 and 1 after dimensionality reduction, and the numerator represents the inter-class differences between symbols 0 and 1 after dimensionality reduction. Let L be the 2-norm. According to Rayleigh's theorem, the optimal solution to the optimization problem is:

[0037] ,

[0038] In the formula, the superscript -1 indicates the inverse operation;

[0039] Step 2.3: The PMT is a commonly used receiving device in underwater wireless optical OOK communication systems. Based on its working principle, the time interval is greater than its basic pulse width (PMT). The correlation between the sampling points is low, therefore the sampling interval is greater than 100°C. When, the two covariance matrices Both are diagonally dominated matrices. For the two covariance matrices... Diagonal approximation of a matrix:

[0040] ,

[0041] but

[0042] ,

[0043] In the formula, diag represents the diagonal approximation operation.

[0044] Step 3: Take the inner product of the received signal sampling point vector obtained in Step 1 and the optimal dimensionality reduction vector found in Step 2 to obtain the dimensionality-reduced received signal; Calculate the probability distribution functions of the received signals corresponding to symbol 0 and symbol 1 of the pilot signals, and find the intersection of the two probability distribution functions as the optimal decision threshold for symbol detection of the dimensionality-reduced received signal. This includes the following steps:

[0045] Step 3.1: Calculate the optimal dimensionality reduction vector obtained in Step 2.3. The inner product of the sampling point vectors corresponding to symbols 0 and 1 in the pilot signal is taken to obtain the dimension-reduced received signal corresponding to the pilot signal. The probability distribution functions of the received signals corresponding to symbols 0 and 1 in the dimension-reduced signal are statistically obtained respectively. The intersection point of the probability distribution functions of the received signals corresponding to symbols 0 and 1 in the dimension-reduced signal is quickly found using a bisection method, and this intersection point is used as the optimal decision threshold. ;

[0046] Step 3.2: When making a decision on the i-th symbol, the sampling point vector within the time interval of the i-th symbol is compared with the optimal dimensionality reduction vector obtained in step 2.3. Perform the inner product to obtain the dimension-reduced received signal, denoted as . .like If the i-th symbol is true, then the i-th symbol is judged as 1; otherwise, it is judged as 0. The symbol judgment is as follows:

[0047] .

[0048] Example

[0049] The proposed method was tested using the communication system framework shown in Figure 1. First, a pseudo-random binary sequence was pulse-shaped and then converted from digital to analog to drive a transmitting laser for signal transmission. The optical signal reached the receiving photomultiplier tube via an underwater wireless optical communication channel, where background light introduced channel noise. The signal output from the photomultiplier tube at the i-th symbol time was converted from analog to digital to output a sampling point vector. The recovered data was obtained after signal detection. In the experiment, a 4m long water tank was selected as the underwater channel. A green LD was used at the transmitting end, and a PMT was used at the receiving end. An electronically controlled attenuator and a filter were placed before the receiving end to attenuate the signal and filter out noise. In order to simulate the dynamic underwater environment, the voltage of the electronically controlled attenuator was adjusted to obtain different received signal optical power, so that the photomultiplier tube traversed three working regions: waveform region, transition region, and photon counting region.

[0050] Figure 2 shows a comparison of the bit error rate (BER) curves of the signal detection method of this invention (the proposed method) and the traditional signal detection methods (average power detection and Poisson maximum likelihood detection) as a function of the optical power at the receiving end when upsampling by 5 times. Specifically, due to the low upsampling factor, the Poisson maximum likelihood method in the traditional signal detection method cannot detect the signal; when the BER is at... When (below the forward error correction threshold) The average power detection methods of this invention and conventional methods require optical power of -48.5 dBm and -45.5 dBm, respectively, while the proposed detection method has an optical power gain of 3 dBm. Furthermore, the proposed method, after diagonal simplification, has lower algorithm complexity and does not introduce significant bit error rate degradation. This demonstrates that this invention can improve signal detection performance with lower complexity and enhance the reliability of underwater wireless optical OOK communication systems. Therefore, the signal detection method proposed in this invention is more suitable for underwater wireless optical OOK communication systems in high dynamic scenarios.

[0051] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A large dynamic signal detection method for an underwater wireless optical OOK communication system, characterized in that, The method comprises the following steps: Step 1, M times up-sampling is performed on a received signal to obtain a sample point vector, and mean values and covariance matrices of received sample points corresponding to symbols 0 and 1 are estimated according to a sample point vector of the received signal corresponding to a pilot; Step 2, an optimization problem is established according to the estimated mean values and covariance matrices of the received sample points, and the optimization problem is solved by using Rayleigh theorem to find an optimal dimension reduction vector; Step 3, an inner product is performed on the received signal sample point vector obtained in step 1 and the optimal dimension reduction vector to obtain a reduced received signal, probability distribution functions of the reduced received signal corresponding to symbols 0 and 1 of the pilot are counted, and an intersection of the two probability distribution functions is found as an optimal decision threshold for performing symbol detection on the reduced received signal.

2. The method of claim 1, wherein the method is a large dynamic signal detection method for an underwater wireless optical OOK communication system. The step 1 comprises: Step 1.1, assuming that the sending end sends a signal of N symbol times, the received end simulates the signal in the i-th symbol time to obtain a sampling point vector in the i-th symbol time wherein the elements denotes the value of the kth sample point in the ith symbol time, ， A superscript T represents a transpose of a matrix; Step 1.2, from the sampling point vector A sample point vector corresponding to a pilot signal is extracted, Recall Let S0 denote the set of vectors of sampling points corresponding to the symbols 0 in the pilot signal. A set of vectors of sample points corresponding to symbol 1 in the pilot signal, respectively estimating the mean of the vectors of sample points corresponding to symbol 0 and symbol 1 With ： ， ， wherein, A number of elements in a set is represented by N; Step 1.3, covariance matrices of the sample point vectors corresponding to symbols 0 and 1 are respectively estimated With ： ， 。 3. The method of claim 1, wherein the method is a large dynamic signal detection method for an underwater wireless optical OOK communication system. The step 2 comprises: Step 2.1, let the reduced dimension vector of the received signal sample point vector be , the mean and variance corresponding to symbol 0 after dimensionality reduction are and ; the mean and variance of the symbol 1 after dimensionality reduction are and ： ， ； Step 2.2, the following optimization problem is solved with an objective of minimizing an intra-class difference of symbols 0 and 1 after dimension reduction and maximizing an inter-class difference of symbols 0 and 1 after dimension reduction: ， wherein denotes the optimal dimensionality reduction vector, representative function a reduced dimension vector of the vector of received signal sample points corresponding to the maximum , intermediate function A 2-norm is represented by ||.||2; ， In the above formula, the denominator represents the within-class difference of the symbols 0 and 1 after dimension reduction, and the numerator represents the between-class difference of the symbols 0 and 1 after dimension reduction, According to Rayleigh theorem, an optimal solution of the optimization problem is: ， In the formula, a superscript -1 represents an inverse operation; Step 2.3, diagonal approximation is performed on the estimated covariance matrices of the sample point vectors corresponding to symbols 0 and 1 With Then ， In the formula, diag represents a diagonal approximation operation. ， The step 3 comprises:

4. The method of claim 1, wherein the method is a large dynamic signal detection method for an underwater wireless optical OOK communication system. Then the i-th symbol is judged as 1, otherwise as 0, and a symbol decision expression is as follows: Step 3.1, receiving the probability distribution function of the received signal corresponding to symbol 0 and symbol 1 in the pilot signal of the received signal after statistical dimension reduction, and using bisection method to quickly find the intersection point of the probability distribution functions of the received signal corresponding to symbol 0 and symbol 1 after dimension reduction as the optimal decision threshold ； Step 3.2, when deciding on the i-th symbol, the vector of sample points in the i-th symbol time is compared to the optimal reduced dimension vector Taking the inner product, we obtain the reduced dimension received signal, denoted as , if ​ 。

Images