Channel model replacement parameter mapping algorithm and mapper based on particle swarm optimization

Abstract

The invention discloses a channel model replacement parameter mapping algorithm based on a particle swarm algorithm and a mapper, and relates to the technical field of channel model replacement. Comprising the following steps: inputting an original channel model and parameters thereof as well as a replacement model form and degree of freedom; executing a moment matching algorithm to obtain a particle swarm algorithm initial search position vector; initializing a position vector, a speed vector and a record vector for the M particles; the optimal record vector of the whole system is initialized; executing a speed and position iteration updating process; executing an observation recording process; judging whether the particle swarm algorithm converges or not; if so, returning an optimal record vector; otherwise, continuing to execute the speed and position iteration updating process.

Description

Technical Field

[0001] The present invention relates to the technical field of channel model replacement, and in particular to a channel model replacement parameter mapping algorithm and a mapper based on a particle swarm algorithm. Background Art

[0002] Due to the existence of various random mechanisms in the signal propagation medium, the performance analysis and optimization of communication systems and the adaptive processing of communication signals are highly dependent on accurate and easy-to-handle channel models. Accurate channel models have a high degree of practical relevance, can accurately quantify and predict the characteristics of communication signals in a specific propagation medium, and can facilitate further analysis of the reception and processing of communication signals. In addition to accuracy, the mathematical tractability of the channel model is equally important, because a series of communication system performance metrics based on the channel model (such as bit error rate, channel capacity, outage probability, diversity gain, etc.) need to be mathematically analyzed and simulated to provide information and guidance with practical analysis value.

[0003] However, in the actual development process, it is extremely challenging to achieve high accuracy and appropriate mathematical tractability of the communication channel model at the same time. Often, if you want high accuracy, the channel model inevitably needs to be complicated and special functions need to be introduced. For this reason, channel model replacement technology plays a role and is considered to be the best technical route to solve this analytical dilemma. With the support of channel model replacement technology, when analyzing and simulating communication systems based on highly complex or mathematically difficult channel models, this model can be replaced by a simpler distributed replacement model, as long as an appropriate parameter mapping relationship is established to keep the accuracy loss within a controllable range. The core of channel model replacement technology lies in the appropriate parameter mapping. Parameter mapping refers to the process of determining the unknown distributed parameter values ​​of the replacement model based on the given known parameters of the original channel model (also called the reference channel model or the target channel model), which is highly related to the replacement accuracy of the channel model replacement technology. An ideal parameter mapper can enable the replacement model to significantly reduce the computational complexity while retaining the main statistical characteristics of the original model. In the past few decades, due to the simplicity and ease of operation of the parameter mapping method based on the moment matching criterion, this method has been widely used in the design of most channel model replacement parameter mappers. However, in-depth research has found that the parameter mapper based on the moment matching criterion has the following major defects:

[0004] 1. Limitations of heuristic algorithms: The moment matching criterion is essentially a heuristic algorithm that relies on the low-order moment information of the model for parameter mapping. Although this method can provide an acceptable replacement model in many cases, its results are often difficult to guarantee to be globally optimal. Especially in complex channel environments, the parameter mapper based on the moment matching criterion may produce large deviations, resulting in low accuracy of the replacement model.

[0005] 2. Unstable replacement accuracy: Since the moment matching criterion only considers low-order moment information, its results are less sensitive to changes in channel characteristics, especially the capture of the tail end characteristics of the channel distribution model is significantly insufficient. In some specific cases, the replacement model generated by the parameter mapper based on the moment matching criterion may be far from the optimal solution for channel model replacement, resulting in the accuracy and reliability of system performance analysis being affected. This instability makes it difficult for the parameter mapper based on the moment matching criterion to meet analysis requirements in some application scenarios with high precision requirements.

[0006] 3. Falling into local optimal solutions and ignoring global optimal solutions: The matrix matching criterion is prone to falling into local optimal solutions during parameter mapping and cannot effectively explore the global parameter space. This means that under complex channel conditions, the replacement effect of the parameter mapper based on the matrix matching criterion may be compromised and the advantages of the channel model replacement technology cannot be fully utilized.

[0007] 4. Poor adaptability: The parameter mapper based on the moment matching criterion has the disadvantage of poor adaptability when dealing with different types of channel model replacement problems. Its fixed mapping method is difficult to cope with diverse channel characteristics, resulting in the replacement model often not performing well in complex and changing channel environments, and a complex parameter adjustment process is required to obtain an acceptable replacement model.

[0008] Due to the existence of the above defects, the existing parameter mappers based on the moment matching criterion have a large room for improvement in practical applications. To address these problems, a new parameter mapping method is urgently needed to overcome the limitations of the moment matching criterion. Summary of the invention

[0009] In view of the deficiencies in the above-mentioned prior art, the present invention provides a channel model replacement parameter mapping algorithm and a replacement mapper based on a particle swarm algorithm to overcome the limitations of the moment matching criterion, and can effectively search for the global optimal solution in a complex parameter space, thereby significantly improving the accuracy and stability of the replacement model.

[0010] The specific technical solutions are as follows:

[0011] The channel model replacement parameter mapping algorithm based on particle swarm algorithm includes:

[0012] Input the original channel model and its parameters, replace the model form and degrees of freedom;

[0013] Execute the moment matching algorithm to obtain the initial search position vector of the particle swarm algorithm;

[0014] Initialize the position vector, velocity vector and record vector for M particles; and initialize the optimal record vector for the entire system;

[0015] Execute the iterative update process of velocity vector and position vector;

[0016] Perform observation and recording process;

[0017] Determine whether the particle swarm algorithm converges; if converged, return the best record vector; otherwise, continue to execute the iterative update process.

[0018] The above scheme further initializes the particle swarm algorithm:

[0019] Assume that there are M particles, each of which is associated with a position vector, a velocity vector, and a record vector, which are denoted as and Where 1<=m<=M; the best record vector among all M particles is recorded as Initialize the initial position vector and velocity vector of all particles:

[0020] b m =r m =r * =b MM ,

[0021]

[0022] Among them, b MM is the K-order moment matching equation Solution; rand([-1,1]) is a function for generating uniform random numbers in the range [-1,1].

[0023] The above scheme further performs a speed and position iterative update process, including:

[0024]

[0025] Among them, individual acceleration c1 and group acceleration c2 are two macro parameters of the particle swarm algorithm, and the two together determine the convergence efficiency of the particle swarm algorithm.

[0026] The above scheme further performs an observation and recording process, including

[0027] The iterative update recording process of all M particles and the system optimality recording process are carried out as follows:

[0028]

[0029] Among them, r m represents the record vector, r * represents the best record vector among all M particles.

[0030] The above scheme further:

[0031] The criterion for judging whether the particle swarm optimization algorithm has converged is: after executing the above particle swarm optimization algorithm λ>=100 times, check the time domain variance statistics and take the average over the number of iterations:

[0032]

[0033] If it is greater than or equal to the threshold ε, it means that the particle swarm algorithm has not converged, and the iteration should continue according to the above steps; if it is less than the threshold, it means that it has converged, stop the iteration, and return As a result of replacing the mapping parameters with the best channel model.

[0034] A channel model replacement mapper based on particle swarm algorithm applies the channel model replacement parameter mapping algorithm based on particle swarm algorithm.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention proposes to achieve efficient global optimization in complex parameter space through particle swarm algorithm. Its significant technical effects include: improving the accuracy of parameter mapping, enhancing the stability of replacement model, and significantly improving the efficiency of communication system performance analysis and optimization. In addition, the present invention is easy to implement, has strong adaptability, can flexibly respond to different channel conditions and communication environments, and provides a strong technical support for communication system design and optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is the basic architecture of the general parameter mapper for channel model replacement;

[0038] Figure 2 It is a flow chart of the channel model replacement parameter mapping algorithm based on particle swarm algorithm;

[0039] Figure 3 is the relative entropy utility surface composed of the characteristic parameters θ and ν of the gamma distribution;

[0040] Figure 4 It is a corresponding technical effect diagram obtained by performing channel parameter mapping on the particle swarm algorithm of the present invention, the traditional moment matching algorithm and the relative entropy minimization algorithm;

[0041] Figure 5 It is a graph of algorithm convergence effect produced by the particle swarm algorithm of the present invention under different macro parameter conditions;

[0042] Figure 6 It is a relationship diagram between the number of iterations and the utility function of the particle swarm algorithm of the present invention under different macro parameter conditions. DETAILED DESCRIPTION

[0043] The following is a further detailed description of the embodiments of the invention in conjunction with the accompanying drawings to make the objectives, technical solutions and technical effects of the invention more clearly presented.

[0044] like Figure 2 As shown, the present invention discloses a channel model replacement parameter mapping algorithm based on particle swarm algorithm, comprising:

[0045] Input the original channel model and its parameters, replace the model form and degrees of freedom;

[0046] Execute the moment matching algorithm to obtain the initial search position vector of the particle swarm algorithm;

[0047] Initialize the position vector, velocity vector and record vector for M particles; and initialize the optimal record vector for the entire system;

[0048] Execute the iterative update process of velocity vector and position vector;

[0049] Perform observation and recording process;

[0050] Determine whether the particle swarm algorithm converges; if converged, return the best record vector; otherwise, continue to execute the iterative update process.

[0051] The above scheme is based on the particle swarm optimization algorithm to find the optimal solution. In the channel model replacement parameter mapping, the algorithm first determines the initial search position vector through the moment matching algorithm to provide a starting point for the particle swarm search; then, the algorithm initializes the position vector, velocity vector and record vector for each particle, and records the best record vector of the entire system; by iteratively updating the velocity and position of the particle, the algorithm continuously approaches the global optimal solution; in this process, the algorithm also records the observation results of each particle to evaluate the quality of its solution; when the particle swarm algorithm reaches the convergence condition, the algorithm returns the current best record vector as the final parameter mapping result. Based on the above scheme, the accuracy and efficiency of the channel model replacement parameter mapping are significantly improved. Its core advantage is that it can automatically adjust parameters to adapt to complex channel environments, thereby ensuring that the communication system can achieve optimal performance under different conditions. The application of the above scheme enables the communication system to simulate the actual channel conditions more accurately when performing performance analysis and optimization, thereby providing a solid technical foundation for system design and optimization.

[0052] In existing communications and signal processing, the propagation attenuation of signals on the communication channel caused by path loss, shadowing and multipath fading will reduce the signal reception quality of the receiver. This effect leads to a time-varying and random channel fading envelope, and can be characterized by a channel envelope model, which is usually a continuous random distribution with a non-negative real domain and consists of a set of distribution parameters. += [0, ∞] represents the random channel fading envelope. The information of the channel model can be fully represented by its probability density function (PDF) fG(g) or the equivalent cumulative distribution function:

[0053] (CDF)FG(g)=Prob{G<g}=∫-O^g{fG(t)}dt

[0054] This is called the original channel model, also known as the reference channel model or target channel model. For most practical situations, we can assume that the original channel model can be fully characterized by N known parameters, expressed as:

[0055] a=[a_1,a_2,…,a_N].

[0056] The form of the channel envelope model and the values ​​of related parameters need to be determined through distribution fitting tests. Usually, the original channel model will adopt a more complex form and more parameters (N is large) to ensure high modeling accuracy. In fact, although the original channel model can well match empirical observations, it may sometimes cause difficulties in analysis and processing, or be too complex in simulation. Therefore, the channel model replacement technology aims to find another random distribution model with a probability density function f^-G(g) or an equivalent cumulative distribution function F^-G(g). The model is characterized by k free parameters β = [b_1, b_2, ..., b_K] (usually K≤N), as an alternative to replace the original channel model. This model is called a channel replacement model.

[0057] The process of determining the unknown parameter set β = [b_1, b_2, ..., b_K] through the known parameter set a = [a_1, a_2, ..., a_N] is called parameter mapping. By determining the appropriate bb = [b_1, b_2, ..., b_K], the channel replacement model F^_G(g) should be similar to the original channel model F_G(g) in a statistical sense; parameter mapping is essentially an optimization process to minimize the statistical distribution distance under constraints.

[0058] As attached Figure 1 As shown, the general architecture of the channel model replacement parameter mapper is shown. The mapper is mainly used for adaptive calculation to minimize the statistical distance between the original channel model F_G(g) and its replacement model F^-G(g).

[0059] The moment matching criterion is the most commonly used parameter mapping method and has been the dominant method for channel model parameter mapping since 2000. Mathematically speaking, the moment matching method is very simple and easy to operate. In order to obtain the values ​​of the K free parameters in the replacement channel model, we only need to solve the following K equations:

[0060]

[0061] That is, the original channel model and the replacement channel model are moment matched from the 1st order moment to the Kth order moment. Parameter mapping based on moment matching only benefits from its simplicity, and in most cases a closed form parameter mapping relationship can be derived; however, the practicality and optimality of the moment matching criterion are difficult to guarantee on a strict basis.

[0062] More rigorously, according to information theory, the statistical distance between two random distributions should be measured by relative entropy (also known as KL divergence):

[0063]

[0064] The smaller the relative entropy, the closer the two random distributions are from the perspective of information theory; in the application context of channel model replacement, relative entropy represents the information loss when the original channel model ff-G(g) is approximated by the replacement model ff^_G(g) with a given parameter mapping bb=[b_1,b_2,…,b_K]. In this way, the problem of finding the optimal channel replacement model can be transformed into a K-dimensional band-limited optimization problem of finding the parameter set bb=[b_1,b_2,…,b_K] that minimizes the relative entropy:

[0065]

[0066] Although minimizing relative entropy can ensure the information-theoretic optimality of channel model replacement, it is computationally challenging to use the criterion of minimizing relative entropy for analysis and optimization. Due to the existence of logarithmic functions in the relative entropy integral expression and the fact that most integrands contain special functions such as gamma function, beta function, Q function, etc., the relative entropy between most channel models cannot usually be expressed as a closed-form solution (the integral cannot be solved), and most optimization problems themselves are also non-concave in nature. Therefore, the criterion of minimizing relative entropy can be used as the best benchmark to quantify and compare the utility of parameter mapping in the channel model replacement problem, but it is usually difficult and computationally unrealistic to directly apply this criterion to infer the optimal mapping parameter set bb = [b_1, b_2, ..., b_K] through a = [a_1, a_2, ..., a_N], especially when N and K are large.

[0067] In order to complement the use of moment matching and the parameter mapping criteria of minimizing relative entropy, so that the parameter mapping process is accurate, fast, and easy to solve, the present invention proposes to use the particle swarm algorithm to perform parameter mapping for channel model replacement. In order to improve the computational efficiency, the present invention first proposes to use the moment matching result as the search starting point for the particle swarm algorithm parameter mapping, and at the same time, for the first time, proposes to use relative entropy to quantify and compare the adaptability of particles to observe the accuracy change of channel model replacement during the execution of the particle swarm algorithm; at the same time, in order to effectively test the convergence of the particle swarm algorithm and terminate the algorithm in time to save settlement resources and speed up the parameter mapping optimization process, the present invention also first proposes a particle swarm algorithm convergence detection criterion applicable to parameter mapping optimization problems.

[0068] Specifically, suppose there are M particles, each of which is associated with a position vector, a velocity vector, and a record vector, denoted as and Where 1≤m≤M. The best record vector among all M particles is recorded as It gives the position vector corresponding to the best utility from the start of the particle swarm algorithm to the observation time point, that is, the mapping parameter set. The present invention proposes for the first time that for all particles, the initial position and velocity are determined by the following initialization process:

[0069]

[0070] where b MM is the K-order moment matching equation Solution; rand([-1,1]) is a function for generating uniform random numbers in the range [-1,1].

[0071] The present invention also stipulates that the iterative update processes of all M particles are independent of each other and follow the following iterative update rules for speed and position:

[0072]

[0073] Among them, individual acceleration c_1 and group acceleration c_2 are two macro parameters of the particle swarm algorithm, and the two together determine the convergence efficiency of the particle swarm algorithm.

[0074] The present invention also provides that the fitness of a specific particle mm (1 <= m <= M) is calculated by the relative entropy The smaller the relative entropy value, the higher the adaptability of the particle. Based on this, the present invention proposes that the iterative update recording process of all M particles and the system optimality recording process are performed in the following manner:

[0075]

[0076] At the same time, the present invention also proposes a criterion for the convergence of the particle swarm algorithm for the parameter mapping problem, that is, the algorithm stopping criterion: after executing the above particle swarm optimization algorithm λ (λ>=100) times, check the time domain variance statistics, that is, take the average over the number of iterations:

[0077]

[0078] If it is greater than or equal to a certain threshold ε, it means that the particle swarm algorithm has not converged, and the iteration should continue according to the above process; if it is less than the threshold, it means that the particle swarm algorithm has converged, the iteration stops, and returns As a result of replacing the mapping parameters with the best channel model.

[0079] The following takes the famous lognormal-gamma channel replacement combination as an example, and compares it with the replacement effects produced by the moment matching criterion and the minimum relative entropy criterion, to demonstrate the working principle and beneficial effects of the channel model replacement parameter mapper based on the particle swarm algorithm. The relevant principles, methods and effects are also applicable to other channel replacement combinations. Figure 3 Represents the relative entropy utility surface composed of the characteristic parameters θ and ν of the gamma distribution: wherein the blue * point represents the initial position vector generated by the moment matching method; the blue o point represents the position vector that finally converges after executing the particle swarm algorithm proposed in the present invention; the red + point represents the optimal point generated based on the minimization of the relative entropy criterion; the blue dotted line represents the iterative trajectory of the optimal record vector.

[0080] Figure 3 The data in the figure show that the channel parameter mapper design based on the particle swarm algorithm proposed in the present invention can effectively converge to the optimal mapping parameter combination: the blue 0 point coincides with the red + point, and the effect is better than the parameter mapper based on the moment matching criterion: the blue 0 point reaches a lower utility surface than the blue * point. In addition, selecting the parameter combination generated by the moment matching criterion as the initial search position of the particle swarm algorithm can effectively improve the computational efficiency and speed up the algorithm convergence and parameter mapping process, that is, the starting point of the algorithm, the blue * point itself, is already at a lower point in the utility surface.

[0081] In the above scheme, the particle swarm algorithm is used to efficiently solve the channel parameter mapping problem, thereby realizing efficient channel model replacement, which has the following characteristics:

[0082] 1. The particle swarm algorithm used in the scheme uses the channel parameter mapping results returned by the moment matching algorithm as the common initial position state of all particles in the particle swarm algorithm, rather than randomly initializing the particle position state in the entire parameter space. This unified initialization setting utilizes the experience that the channel parameter mapping results returned by the moment matching algorithm are usually relatively close to the global optimal solution, which can save the computing power of the particle swarm algorithm and thus accelerate convergence.

[0083] 2. The particle swarm algorithm used in the scheme uses the relative entropy of the reference distribution to the alternative distribution as a measure of particle fitness; that is, the smaller the relative entropy of a particle, the stronger the fitness of the particle is. Compared with traditional linear fitness measures, such as mean square error, relative entropy is a nonlinear measure that can efficiently detect subtle differences between the reference distribution and the alternative distribution, thereby making a more accurate judgment on whether the particle swarm algorithm has converged.

[0084] 3. Due to the complex boundary conditions of the parameter space in the channel parameter mapping problem, particles often go beyond the boundary of the parameter space, resulting in invalid solutions. If the traditional particle swarm algorithm is used and the out-of-bounds problem is not effectively controlled, these recorded invalid solutions may further interfere with the subsequent execution of the particle swarm algorithm, causing misjudgment of algorithm convergence or giving invalid convergence results.

[0085] To solve this problem, the particle swarm algorithm used in the above scheme adds position state boundary restrictions. Once the position state of a particle is determined to be outside the boundary, it is forced to stay on the boundary with the shortest Euclidean distance, thereby ensuring that all particles will not exceed the state space boundary during the random search process.

[0086] 4. Finally, the above scheme solves the difficulty of judging the convergence of the particle swarm algorithm due to the use of a unified moment matching algorithm for initialization according to the characteristics of the channel parameter mapping problem. Specifically, since the random initialization in the entire parameter space is abandoned and a unified moment matching result is used as the position state initialization for all particles, it can be considered that in the initialization phase when the particle swarm algorithm begins to execute, all particle states are perfectly converged, but they have not converged to the global optimal solution (i.e., the point with the minimum relative entropy in the entire parameter space); therefore, if the convergence judgment is made based on the convergence judgment criterion of the traditional particle swarm algorithm based on the random initialization method, the particle swarm algorithm used in this scheme will be immediately judged as converged in the initialization phase, thereby terminating the iterative search process of the particle swarm and returning the solution of the moment matching algorithm (i.e., the solution in the initialization state).

[0087] In view of this, combined with the complexity of the channel parameter mapping problem and previous research experience, the above scheme improves the convergence judgment condition of the traditional particle swarm algorithm: first, a judgment protection period window is set on the number of iteration rounds, that is, the convergence judgment is not performed within the λ rounds after initialization, so as to ensure that the particles are fully randomly diffused and in the λ rounds after initialization, and move and iterate according to the basic rules of the particle swarm algorithm. In this way, it can be ensured that the particle swarm will not be judged as converged because of sharing an initialized position state; secondly, it is proposed to use the variance of the relative entropy of the particle swarm as the judgment quantity of the algorithm convergence, that is, when the number of iteration rounds of the particle swarm algorithm exceeds λ rounds and the variance of the relative entropy is less than ε, the particle swarm algorithm is considered to have converged. Compared with the linear convergence judgment standard, the variance of the relative entropy proposed in this scheme is a strong convergence standard, which can ensure a higher judgment accuracy.

[0088] The corresponding technical effects of the particle swarm algorithm proposed in this scheme are compared with the traditional moment matching algorithm (low complexity, poor utility) and the relative entropy minimization algorithm (extremely high complexity, optimal utility) for channel parameter mapping. Figure 4 shown.

[0089] The algorithm convergence effect of the particle swarm algorithm proposed in this scheme under different macro parameter conditions is as follows Figure 5 As shown, different colors represent the number of iterations required for the algorithm to converge, which proves the effectiveness and efficiency of the particle swarm algorithm proposed in the above scheme for the channel parameter mapping problem.

[0090] The relationship between the number of iterations and the utility function (relative entropy) of the particle swarm algorithm proposed in this scheme under different macro parameter conditions is as follows: Figure 6 As shown, the rationality and effectiveness of the convergence protection window proposed in the above scheme are proved.

[0091] The above description is only a preferred feasible embodiment of the present invention and is not intended to limit the scope of the patent application of the present invention. All equivalent changes, equivalent substitutions or modified changes performed within the technical spirit and principles suggested by the present invention should be included in the scope of patent protection covered by the present invention.

Claims

1. A channel model replacement parameter mapping algorithm based on particle swarm algorithm, characterized in that: include: Input the original channel model and its parameters, replace the model form and degrees of freedom; Execute the moment matching algorithm to obtain the initial search position vector of the particle swarm algorithm; Initialize the position vector, velocity vector and record vector of M particles; and initialize the optimal record vector of the entire system; Execute the iterative update process of velocity vector and position vector; Perform observation and recording process; Determine whether the particle swarm algorithm converges; if converged, return the best record vector; otherwise, continue to execute the iterative update process.

2. The channel model replacement parameter mapping algorithm based on particle swarm algorithm according to claim 1 is characterized in that: Initialize the particle swarm algorithm: Assume that there are M particles, each of which is associated with a position vector, a velocity vector, and a record vector, which are denoted as and Where 1<=m<=M; the best record vector among all M particles is recorded as Initialize the initial position vector and velocity vector of all particles: b m =r m =r * =b MM , Among them, b MM is the K-order moment matching equation Solution; rand([-1,1]) is a function for generating uniform random numbers in the range [-1,1].

3. The channel model replacement parameter mapping algorithm based on particle swarm algorithm according to claim 2 is characterized in that: Execute the iterative update process of speed and position, including: Among them, c1 is the individual acceleration and c2 is the group acceleration.

4. The channel model replacement parameter mapping algorithm based on particle swarm algorithm according to claim 2 is characterized in that: Perform observation recording process, including: The iterative update recording process of all M particles and the system optimality recording process are carried out as follows: Among them, r m represents the record vector, r * represents the best record vector among all M particles.

5. The channel model replacement parameter mapping algorithm based on particle swarm algorithm according to claim 2 is characterized in that: The criterion for judging whether the particle swarm optimization algorithm has converged is: after executing the above particle swarm optimization algorithm λ>=100 times, check the time domain variance statistics and take the average over the number of iterations: If it is greater than or equal to the threshold ε, it means that the particle swarm algorithm has not converged, and the iteration should continue according to the above operation; if it is less than the threshold, it means that it has converged, stop iteration, and return As a result of replacing the mapping parameters with the best channel model.

6. A channel model replacement mapper based on particle swarm algorithm, characterized in that: Apply the channel model replacement parameter mapping algorithm based on particle swarm algorithm described in any one of claims 1-5.

Images