CN117896219A - LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Signal processing) optimization - Google Patents
LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Signal processing) optimization Download PDFInfo
- Publication number
- CN117896219A CN117896219A CN202410303673.8A CN202410303673A CN117896219A CN 117896219 A CN117896219 A CN 117896219A CN 202410303673 A CN202410303673 A CN 202410303673A CN 117896219 A CN117896219 A CN 117896219A
- Authority
- CN
- China
- Prior art keywords
- optimization
- algorithm
- lms
- population
- mean square
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 85
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000012545 processing Methods 0.000 title abstract description 6
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 66
- 241000287127 Passeridae Species 0.000 claims abstract description 44
- 230000003044 adaptive effect Effects 0.000 claims abstract description 28
- 238000010845 search algorithm Methods 0.000 claims abstract description 15
- 238000002945 steepest descent method Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 31
- 230000008569 process Effects 0.000 claims description 11
- 238000011156 evaluation Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 230000002431 foraging effect Effects 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 14
- 238000012549 training Methods 0.000 description 9
- 238000001914 filtration Methods 0.000 description 7
- 230000004044 response Effects 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 238000004836 empirical method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000005311 autocorrelation function Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 244000062645 predators Species 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000003997 social interaction Effects 0.000 description 1
Landscapes
- Filters That Use Time-Delay Elements (AREA)
Abstract
The invention discloses an LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure processing) optimization, and relates to the technical field of signal processing. The method comprises the following steps: determining an LMS equalization optimization algorithm; the LMS equalization optimization algorithm is constructed according to a minimum mean square error algorithm, a sparrow search algorithm and corresponding rule constraint; the minimum mean square error algorithm is constructed by taking the mean square error in the steepest descent method as a cost function and replacing the mean square error with the instantaneous error; optimizing initial parameters of the adaptive equalizer by using the LMS equalization optimization algorithm to obtain the adaptive equalizer after parameter optimization; the adaptive equalizer after parameter optimization is used for carrying out signal equalization on the QAM signal; the initial parameters include an initial step size and an order. The invention can make up for the neglect of the traditional improved mode to the initial parameter setting, fills the blank of the idealization of the initial parameter setting in the traditional mode, and improves the parameter optimizing effect of the equalizer.
Description
Technical Field
The present invention relates to the field of signal processing technologies, and in particular, to an LMS equalization optimization method, apparatus, and medium based on SSA optimization.
Background
In a digital communication system, the signal may be affected by multipath propagation, interference and noise during transmission, and the adaptive equalizer can effectively compensate for these distortions, thereby improving the reliability of data transmission.
Conventionally, adaptive equalizers use an adaptive algorithm, such as the LMS algorithm, to adjust their filter coefficients. The LMS algorithm does not need to know the statistical properties of the input signal and can learn and adjust weights over time, thus being the most basic and widely used algorithm in the field of adaptive filtering. Such algorithms operate based on the error between the received signal and the desired signal, and as the error changes, the algorithm dynamically adjusts the weights to be optimal and ultimately strives for a relatively minimum error. Classical LMS is used as a fixed step algorithm, and its most important parameters are the initial step μ and the order M.
In most designs, the selection of the optimal parameters for the initial step size μ, after deriving the theoretical upper limit, is empirically adjusted according to the simulation effect, and lacks the discussion of the related art. The selection of the order M is based on empirical methods, and also refers to the impulse response length, complexity requirement, over-fitting condition, noise and interference requirement, hardware resource allocation, etc. of the system.
For LMS adaptive equalizer based on fixed step length, after the step length deduces the theoretical selection range, the conventional method generally selects a step length far smaller than the upper limit to ensure stability and robustness, and the specific selection depends on an empirical method. But typically this range spans a large order of magnitude, the optional range is large, and the step size is critical to the outcome of the overall iteration, and conventional practice does not guarantee parameter optimization using empirical methods.
For a variable step length LMS self-adaptive equalizer, related technology directly modifies an algorithm structure, and variable step length coefficients are introduced, so that the algorithm can dynamically adjust the step length in an iterative process. However, this coefficient introduces division computations, adding significant difficulty and complexity to the associated hardware implementation. Similarly, the order is designed with dynamic variable order, but related researches are less, and the technical problems of hardware resource burden and complexity still exist.
Disclosure of Invention
The invention aims to provide an LMS equalization optimization method, equipment and medium based on SSA optimization, which can make up for the neglect of the traditional improvement mode on the initial parameter setting, has small hardware realization difficulty and can obtain the optimal initial parameter under the given environment in a short time.
In order to achieve the above object, the present invention provides the following solutions:
an LMS equilibrium optimization method based on SSA optimization comprises the following steps:
Determining an LMS equalization optimization algorithm; the LMS equalization optimization algorithm is constructed according to a minimum mean square error algorithm, a sparrow search algorithm and corresponding rule constraint; the minimum mean square error algorithm is constructed by taking the mean square error in the steepest descent method as a cost function and replacing the mean square error with the instantaneous error;
Optimizing initial parameters of the adaptive equalizer by using the LMS equalization optimization algorithm to obtain the adaptive equalizer after parameter optimization; the adaptive equalizer after parameter optimization is used for carrying out signal equalization on the QAM signal; the initial parameters include an initial step size and an order.
Optionally, the LMS equalization optimization algorithm is used to optimize initial parameters of the adaptive equalizer, so as to obtain the adaptive equalizer after parameter optimization, which specifically includes:
Setting initial conditions; the initial conditions comprise an initial population, and the number and range of target parameters of an optimized minimum mean square error algorithm; the initialization population comprises a plurality of sparrows, and each sparrow is a different combination of an initial step length and an order;
performing fitness evaluation according to the initial conditions, and determining an objective function; the objective function is a stable steady state error value under the set iteration times;
And carrying out parameter optimization on the objective function by utilizing a sparrow search algorithm, determining an optimal parameter, and obtaining the self-adaptive equalizer after parameter optimization according to the optimal parameter.
Optionally, the method for determining the initialization population comprises the following steps:
Setting a filter initialization coefficient;
Calculating an equalizer output signal based on the filter initialization coefficients;
calculating an equalizer error based on the equalizer output signal; the equalizer error is the error of the equalizer output signal and the desired signal;
Adjusting equalizer parameters according to the equalizer errors, and returning to the step of calculating equalizer output signals based on the filter initialization coefficients until the equalizer is ended to obtain an output sequence and an error sequence;
The output sequence and the error sequence are determined as an initialization population.
Optionally, the fitness evaluation is performed according to the initial condition, and the objective function is determined, which specifically includes:
Calculating a steady-state mean square error according to the initial condition;
and carrying out fitness evaluation on the position of each sparrow by using the steady-state mean square error, and determining an objective function.
Optionally, the rule constraint of the sparrow search algorithm includes at least one finder and follower mechanism.
Optionally, the rule constraint of the sparrow search algorithm specifically includes:
The first, the discovery person is generally located in a better position, has higher food amount reserve, can search for the area with more abundant food amount, and provides the position direction of foraging for other follower; in the analogy model, the position where the finder is located depends on the size of the fitness value of the individual, and the smaller the fitness value is, the better the fitness value is;
Secondly, when any individual in the population is found dangerous or frightened, the individual can give out a sound to inform other individuals of the danger, and the position needs to be transferred; when the alarm value is higher than the safety threshold value, the discoverer can lead the population to other safety areas to search food, and if threat still exists at the new position, the population position is continuously transferred;
Third, the identity of both the follower and the third is dynamically changed during foraging of the population; the discoverer has richer food sources, when a certain follower has better food sources, the follower becomes the discoverer, the ratio of the discoverer to the follower is generally unchanged in the population, so that the two are fixed in number in the model, and when a certain follower individual becomes the discoverer, the discoverer becomes the follower, and the identities of the two are dynamically changed;
Fourth, as the concentration of food sources becomes lower as the followers find food in the vicinity of the finder, the poorer their position in the overall population, i.e., the poorer the fitness value, the followers fly to other locations to find food in an attempt to acquire more food sources;
Fifth, during the process of searching for food, the follower always follows the finder with more food sources, and then looks for food from the vicinity of the finder; some followers monitor the discoverers all the time and have a certain probability of competing for food in order to increase their predation rate;
sixth, when individuals in the population find a hazard, individuals at the edges of the population may fly quickly to a safer area, while individuals in the middle of the population may randomly move.
The invention also provides electronic equipment which is characterized by comprising a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the LMS balance optimization method based on SSA optimization.
The invention also provides a computer readable storage medium storing a computer program which when executed by a processor implements an LMS equalization optimization method based on SSA optimization as described above.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
The invention discloses an LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Shell) optimization, wherein the method comprises the steps of determining an LMS balanced optimization algorithm; the LMS equalization optimization algorithm is constructed according to a minimum mean square error algorithm, a sparrow search algorithm and corresponding rule constraint; the minimum mean square error algorithm is constructed by taking the mean square error in the steepest descent method as a cost function and replacing the mean square error with the instantaneous error; optimizing initial parameters of the adaptive equalizer by using the LMS equalization optimization algorithm to obtain the adaptive equalizer after parameter optimization; the adaptive equalizer after parameter optimization is used for carrying out signal equalization on the QAM signal; the initial parameters include an initial step size and an order. The invention can make up for the neglect of the traditional improved mode to the initial parameter setting, fills the blank of the idealization of the initial parameter setting in the traditional mode, and improves the parameter optimizing effect of the equalizer.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an LMS equalization optimization method based on SSA optimization;
FIG. 2 is a schematic diagram showing the steps of the sparrow optimizing algorithm in the present embodiment;
FIG. 3 is a logic flow diagram of the equalization optimization method in the present embodiment;
Fig. 4 is a constellation diagram before and after filtering by the LMS algorithm in the present embodiment; wherein part (a) is a training sequence diagram, and part (b) is a transmission sequence diagram; part (c) is a schematic diagram of a received sequence; part (d) is an equalizer output schematic;
FIG. 5 is a diagram showing the constellation before and after filtering by the SSA-LMS algorithm for the first time in this embodiment; wherein part (a) is a training sequence diagram, and part (b) is a transmission sequence diagram; part (c) is a schematic diagram of a received sequence; part (d) is an equalizer output schematic;
FIG. 6 is a diagram showing the constellation before and after filtering by the SSA-LMS algorithm for the second time in this embodiment; wherein part (a) is a training sequence diagram, and part (b) is a transmission sequence diagram; part (c) is a schematic diagram of a received sequence; part (d) is the equalizer output schematic.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide an LMS equalization optimization method, equipment and medium based on SSA optimization, which can make up for the neglect of the traditional improvement mode on the initial parameter setting, has small hardware realization difficulty and can obtain the optimal initial parameter under the given environment in a short time.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
As shown in fig. 1, the present invention provides an LMS equalization optimization method based on SSA optimization, including:
Step 100: determining an LMS equalization optimization algorithm; the LMS equalization optimization algorithm is constructed according to a minimum mean square error algorithm, a sparrow search algorithm and corresponding rule constraint; the minimum mean square error algorithm is constructed by taking the mean square error in the steepest descent method as a cost function and replacing the mean square error with the instantaneous error; wherein the rules constraint of the sparrow search algorithm includes at least one finder and follower mechanism.
Step 200: optimizing initial parameters of the adaptive equalizer by using the LMS equalization optimization algorithm to obtain the adaptive equalizer after parameter optimization; the adaptive equalizer after parameter optimization is used for carrying out signal equalization on the QAM signal; the initial parameters include an initial step size and an order.
As a specific embodiment of step 200, it includes:
Step 210: setting initial conditions; the initial conditions comprise an initial population, and the number and range of target parameters of an optimized minimum mean square error algorithm; the initialization population comprises a plurality of sparrows, and each sparrow is a different combination of an initial step length and an order;
Step 220: performing fitness evaluation according to the initial conditions, and determining an objective function; the objective function is a stable steady state error value under the set iteration times; the method specifically comprises the following steps:
Calculating a steady-state mean square error according to the initial condition; and carrying out fitness evaluation on the position of each sparrow by using the steady-state mean square error, and determining an objective function.
Step 230: and carrying out parameter optimization on the objective function by utilizing a sparrow search algorithm, determining an optimal parameter, and obtaining the self-adaptive equalizer after parameter optimization according to the optimal parameter.
The method for determining the initialization population comprises the following steps:
Setting a filter initialization coefficient; calculating an equalizer output signal based on the filter initialization coefficients; calculating an equalizer error based on the equalizer output signal; the equalizer error is the error of the equalizer output signal and the desired signal; adjusting equalizer parameters according to the equalizer errors, and returning to the step of calculating equalizer output signals based on the filter initialization coefficients until the equalizer is ended to obtain an output sequence and an error sequence; the output sequence and the error sequence are determined as an initialization population.
Based on the above technical solution, the following embodiments are provided.
In order to solve the problems in the prior art, the present embodiment uses a sparrow search algorithm to optimize the initial parameters of the LMS, including the initial step size and the order, so as to obtain a better result in the adaptive equalization of the received signal, without increasing the difficulty of hardware implementation. The SSA calculates the optimal initial parameters under the given environment in a short time through the steps of initializing, foraging, migration, updating global optimal solution, judging termination conditions and the like.
In the embodiment, two important parameters of the LMS are innovatively optimized, so that neglect of the traditional improvement mode on initial parameter setting is made up, and the gap of idealization of the initial parameter setting in the traditional mode is filled. Different from the mode of modifying the LMS algorithm in the general mode, the method returns to true without complicating the design, improves parameter setting at the source, and finally can reach the level similar to the improved algorithm.
The LMS algorithm optimized based on the method can improve the effect of the LMS adaptive equalizer to a certain extent, and is close to the effect which can be achieved by improving the LMS algorithm. The method has a certain reference meaning for the optimization of the initial parameters of other improved LMS algorithms, and has a heuristic value for realizing the step length or the order change of the LMS along with iteration by using SSA.
According to the embodiment, a sparrow search optimization method with excellent global optimizing capability is fully utilized to capture the selection of initial step length and order number when the steady state error of an objective function is stable and minimum under the limit iteration times in a minimum mean square error algorithm, given data are configured to be as close to a communication environment as possible, optimized data meeting termination conditions are used for being put into application to perform signal equalization work, training data and population parameters are adaptively adjusted when the environment is changed, and the optimized result is updated to complete optimization. Compared with the similar technology, the algorithm provided by the invention has smaller steady-state error and faster convergence speed when carrying out signal equalization.
The technical solution adopted by the embodiment for achieving the above purpose is to provide a minimum mean square error algorithm self-adaptive equalization method based on sparrow search algorithm optimization with the following structure, which comprises the following specific steps:
step 1) initializing a population, and determining the number and the range of optimized target parameters;
step 2) evaluating the fitness, and determining the objective function as a stable steady state error value under the given iteration times;
Step 3) completing the optimizing process through a sparrow optimizing algorithm, and obtaining an optimal parameter result for signal equalization;
the specific steps of initializing the population are as follows:
the main idea of the LMS algorithm is to take the mean square error in the steepest descent method as a cost function, and replace the mean square error with an instantaneous error, and the algorithm process can be described as follows:
1) Filter initialization coefficients: ;
2) Calculating an equalizer output signal: ;
3) Calculating an error of the equalizer output signal and the desired signal: ;
4) Adjusting equalizer parameters according to the error: .
5) Returning to the step 2), repeating until the end, and obtaining an output sequence and an error sequence.
The initial step size affects the convergence speed and steady state error, and the two tend to repel each other. For the selection of step size μ, its theoretical range is , where tr [ R ] is the trace of R,/> is the maximum eigenvalue of the autocorrelation matrix R of the input signal,/> . The autocorrelation function can be theoretically calculated or estimated only on the basis of complete knowledge of the system model and the signal generation process, so that it may be difficult to acquire the complete autocorrelation function in real time with an on-line or stepwise process. Typically, when a smooth or ill-conditioned assumption can be made, the computation trace may be relatively simple and can be incorporated as a parameter range reference for the initializing population.
While the order affects the filter length or the number of weights w (n) updated per iteration. The range of values is relatively broader, and if we have information about the system impulse response, the order M should cover at least the point in time when the system impulse response decays to a smaller value. For example, if the impulse response is known to decay to zero or near zero after T time units, then the order M should be greater than T.
The value range of mu can be deduced to a certain extent based on the condition of the input signal, and the size of the initial population in the optimization algorithm is determined by introducing the range, so that the initial population is used as the generation range of the random initial solution defined by the solution space.
The specific steps for evaluating the fitness are as follows:
Firstly, for the equalizer to simulate the impulse response of the channel and perform the result after the channel equalization, a unified performance evaluation criterion, namely the steady-state mean square error MSD, is required. MSD is the set average value that produces the mean square estimation error for the difference between the filter coefficients and the impulse response of the channel, and is defined as follows:
Where is the matrix trace,/> is the mathematical expectation of the random variables. MSD is typically expressed in its dB form:
in the field of adaptive filtering, the objective is typically to surround a minimized MSD, ensuring that the approximation coefficients converge exactly to their optimal counterparts. Therefore, the introduction of steady state MSD as a guideline in SSA-based optimization becomes an obvious choice, closely coupled with the principle of minimizing the differences in adaptive filter coefficient estimation. In SSA, each sparrow in the population symbolizes a different combination of LMS step size and filter order, projected into a multidimensional search space. The process of assessing the location (solution) of each sparrow by defining an MSD-based objective function usefully quantifies the effectiveness of the reduction in estimation error embodied in the LMS parameters. A smaller MSD represents better alignment with the optimal filter coefficients and therefore translates to better fitness and enhances the impact of sparrow guide populations in subsequent iterations.
The steps of obtaining the result through the sparrow optimizing algorithm are shown in fig. 2:
After the initial population setting and the objective function conditions are determined, the SSA algorithm simulates the division and social interaction between different individuals in the sparrow population when the sparrow population searches for food, and optimization is performed through relevant steps, as shown in the above graph. In order to make the SSA algorithm simpler, some behaviors are idealized, corresponding rules are formulated, and a mathematical model is established. The following rules exist:
1. The discoverer is generally located in a better position, has a higher food reserve and can search for areas with more abundant food, and provides foraging position directions for other followers. In analogy to the model, the position where the finder is located depends on the size of the fitness value of the individual, the smaller the fitness value, the better.
2. When any individual in the population is found dangerous or frightened, the individual will sound a beep to inform other individuals that there is a danger, and the location needs to be transferred. When the alarm value is higher than the safety threshold, the discoverer brings the population to other safety areas for searching food, and if threat still exists at the new position, the population position is continuously transferred.
3. The identity of both the follower and the population is dynamically changed during foraging of the population. The discoverer has richer food sources, when the discoverer has better food sources, the discoverer becomes the discoverer, the ratio of the discoverer to the discoverer is generally unchanged in the population, so that the two are fixed in number in the model, and when the discoverer individual becomes the discoverer, the discoverer becomes the discoverer, and the identities of the two are dynamically changed.
4. As the concentration of food sources in the vicinity of the finder becomes lower, the poorer their location in the overall population, i.e. the poorer the fitness value, the followers fly to other locations to find food in an attempt to obtain more food sources.
5. During the search for food, the follower always follows the finder with more food sources, and then looks for food from the vicinity of the finder. Some followers will constantly monitor discoverers and have some probability of competing for food in order to increase their predation rate.
6. When individuals in the population find a hazard, individuals at the edges of the population may fly quickly to a safer area, while individuals in the middle of the population may randomly move.
In SSA, the optimal individuals within the population will have priority to acquire food during the search. As a seeker, it can obtain a larger search range for foraging than a follower. During each iteration, the position of the seeker is updated as follows:
Wherein is the individual position of sparrow, i is the current iteration number, itermax is the maximum iteration number; alpha is a random number in [0,1 ]; the/> 、/> is an early warning value and a safety value respectively; q is a random number obeying normal distribution; l is a1 xd matrix, where each element is 1.
When , this means that there are no natural enemies around, the seeker can conduct a global search. If/> this means that some sparrows have found predators, all sparrows have to take relevant action. During the foraging process, some followers monitor the seeker at all times. Once the seeker finds a better food, they immediately leave the current location to compete for the food. The food is immediately available if they win the competition, otherwise the rule (4) needs to be continued. The follower's location update is as follows:
Wherein X p is the position of the optimal seeker, and X worse is the current global worst position; n is population size. A is a1×d matrix, each element has a random amplitude of 1 or-1, where A + is defined as follows:
When i > n/2 shows that the state of the ith follower with lower fitness value is poor, the ith follower needs to fly to other places to find food. In the algorithm, the primitive text assumes that 10% -20% of the individuals within the population (20% set in this embodiment) are aware of the danger, and that the initial positions of these individuals are randomly generated in the population:
Wherein X best is the current global optimal position, beta is a step control parameter, and the value of the step control parameter is a normal distributed random number with the obeying mean value of 0 and the variance of 1; k is a random number within [ -1,1 ]; f is an fitness value, and f g、fw is the current optimal fitness value and the worst fitness value respectively; epsilon is a constant that avoids denominator of 0. f i＞fg indicates that this sparrow is at the edge of the population, f i=fg, which indicates that sparrows in the middle of the population are aware of the danger and need to be brought close to each other to avoid predation.
And repeating the iterative process until the fitness condition or the objective function based on the step 2) reaches a threshold value, and outputting an optimization result.
According to the steps shown in fig. 3, a specific experimental protocol is provided:
1. Initializing a population phase.
1) Parameters and ranges are defined. As can be seen from the description of the LMS algorithm procedure in step 1) above, the target parameters include step size and order, so the number is 2.
The value range of the target parameter directly influences the size of the optimized solution space, and in principle, a larger search range is limited as much as possible, so that the initial sparrow population can fully explore the parameter space, the local minimum is prevented from being trapped, and the global optimized solution is obtained. Therefore, for the initial step size, the parameter range is set to (0, 1) when the relevant statistical characteristic information of the input signal is lacking, and the value of the upper limit 1 depends on a large amount of experience rules. The equation is referred to as the initial population range when learning or being able to estimate the trace of the autocorrelation matrix R of the input signal.
2) By encoding the step size μ and the filter order M as a solution vector of SSA, each sparrow represents one possible combination of parameters. The variable number dim is set to 2 in the get_ Functions _detail file, and the ranges of the lower and upper bounds lb and ub are set as the case may be according to the above derivation. Subsequently, the various population locations are randomly initialized as described in step 3).
2. Assessment of fitness
1) And determining the fitness function. The fitness function represents the MSD result generated when the combination of parameters represented by the ith sparrow at the jth dimension acts on the LMS at the t-th iteration, and is used as an objective function in the optimization process, which describes the quality of a solution, and the lower the requirement is, the better the lower the requirement is. MSD has large data dispersion degree on linear scale, and is converted into smaller order of magnitude and unchanged monotonicity for comparison. In the constellation, the size of MSD is expressed as the concentration degree of symbol points, and the lower the MSD is, the more concentrated the symbol points are.
Success of the optimization may be determined when results smaller after the optimization than the last iteration, or symbol points are observed more concentrated in the constellation.
2) And (5) environment configuration. The environment configuration requirements are as close as possible to the target channel condition, in consideration of the actual scene of aviation communication, under the environment of onboard radar and navigation under the general condition, airport ground communication and the like, the 16-QAM modulation simulates the condition under the harsher condition, the complex signal with the length part of n 1 and the real part both in is added as a training sequence before the original signal, the random integer array in [0,15] is generated and mapped onto 16 possible symbols of the 16-QAM by the rear part n 2, or the random integer array in [0,63] is generated and mapped onto 64 possible symbols of the 64-QAM.
The LMS algorithm itself can complete the filtering operation without the statistical properties of the input signal, but in actual situations, the content of the training sequence is often added for better quick yield, initialization, robustness, performance evaluation, etc., and for better showing the filtering variation in a limited number of iterations, a simple training sequence based on 4-QAM modulation is added.
3. Iterative optimization stage
After the pre-work is completed, SSA is used for optimizing, and the solution vector is continuously and iteratively updated according to the returned objective function condition, and the pseudo code is as follows:
SSA pseudocode:
Maximum iteration G
PD number of producers
SD, perceived dangerous sparrow number
R 2 alarm value
Begin
Initializing parameters and initializing a population;
While()
Sorting fitness function values of all individuals and finding out the current optimal individual and the worst individual;
For i=1：PD
The discoverer quickly explores the whole solution space and performs extensive searching;
End for
For i=PD+1：N
the follower searches for a better position around the finder and performs local search;
End for
For j=1：SD
early warning individuals of danger, migration to other locations
End for
Obtaining a current new location
Comparing the quality of the new position through the fitness function value, and updating if the new position is higher
The number of iterations is increased by 1, i.e. t=t+1
End while
Outputting the position of the optimal individual
End
4. Results application and environment change discrimination
And independently running the optimized result in the SSA-LMS to verify the validity of the optimized result. In practical use, if the channel condition is changed, the relevant information of the environment configuration is returned to the process of evaluating the fitness, and if the input signal characteristic is changed, the initialization condition is returned to the process of initializing the population.
5. Simulation experiment
To verify the validity of SSA-LMS, classical SSA was the subject of comparison. The environment setting uses the configuration preset above. The result of iterating 10000 times after training 300 times based on training sequence of 4-QAM modulation under the signal to noise ratio of 30dB is shown in figure 4. Within a limited number of iterations, classical algorithms work but convergence of the results is not ideal. Continuing the iteration under the condition of too small a step length can sacrifice the running speed to obtain better convergence accuracy, and continuing the iteration under the condition of too large a step length can not have significant influence on the result, so that the effective results in the limit of the iteration times. After optimizing, new parameters are introduced, and the operation results are shown in fig. 5 and 6, the results of multiple operations can be known to show a certain consistency, and the reliability and the robustness of the method are proved, so that the simulation experiment is effective.
The embodiment has the following beneficial effects:
the optimization algorithm is chosen to be more advanced. Compared with other intelligent particle swarm optimization algorithms, the sparrow optimizing algorithm used in the invention can achieve better optimizing speed in limited optimizing time on the primary parameter optimizing result of the LMS, has stronger global searching capability and has higher advancement.
The optimization efficiency is higher. Each sparrow of SSA can be considered an independent search agent, which makes the algorithm easily parallelizable, thereby increasing computational efficiency.
Less hardware resources are occupied. The method avoids the hardware design difficulty caused by LMS complex design by using a software pre-optimization mode, and does not need to consider the processing of division in the variable step size coefficient.
The invention also provides electronic equipment which is characterized by comprising a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the method.
The invention also provides a computer readable storage medium storing a computer program which, when executed by a processor, implements a method as described above.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the core concept of the invention; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (8)
1. An LMS equilibrium optimization method based on SSA optimization is characterized by comprising the following steps:
Determining an LMS equalization optimization algorithm; the LMS equalization optimization algorithm is constructed according to a minimum mean square error algorithm, a sparrow search algorithm and corresponding rule constraint; the minimum mean square error algorithm is constructed by taking the mean square error in the steepest descent method as a cost function and replacing the mean square error with the instantaneous error;
Optimizing initial parameters of the adaptive equalizer by using the LMS equalization optimization algorithm to obtain the adaptive equalizer after parameter optimization; the adaptive equalizer after parameter optimization is used for carrying out signal equalization on the QAM signal; the initial parameters include an initial step size and an order.
2. The SSA-based optimized LMS equalization optimization method of claim 1, wherein the optimizing the initial parameters of the adaptive equalizer by using the LMS equalization optimization algorithm to obtain the parameter-optimized adaptive equalizer specifically comprises:
Setting initial conditions; the initial conditions comprise an initial population, and the number and range of target parameters of an optimized minimum mean square error algorithm; the initialization population comprises a plurality of sparrows, and each sparrow is a different combination of an initial step length and an order;
performing fitness evaluation according to the initial conditions, and determining an objective function; the objective function is a stable steady state error value under the set iteration times;
And carrying out parameter optimization on the objective function by utilizing a sparrow search algorithm, determining an optimal parameter, and obtaining the self-adaptive equalizer after parameter optimization according to the optimal parameter.
3. The SSA-based LMS equilibrium optimization method of claim 2, wherein the initializing population determination method is as follows:
Setting a filter initialization coefficient;
Calculating an equalizer output signal based on the filter initialization coefficients;
calculating an equalizer error based on the equalizer output signal; the equalizer error is the error of the equalizer output signal and the desired signal;
Adjusting equalizer parameters according to the equalizer errors, and returning to the step of calculating equalizer output signals based on the filter initialization coefficients until the equalizer is ended to obtain an output sequence and an error sequence;
The output sequence and the error sequence are determined as an initialization population.
4. The SSA-based LMS equalization optimization method of claim 2, wherein the fitness evaluation is performed according to the initial condition, and the determining the objective function specifically comprises:
Calculating a steady-state mean square error according to the initial condition;
and carrying out fitness evaluation on the position of each sparrow by using the steady-state mean square error, and determining an objective function.
5. The SSA-based LMS equilibrium optimization method of claim 1, wherein the rules constraint of the sparrow search algorithm includes at least one finder and follower mechanism.
6. The SSA-based LMS equilibrium optimization method of claim 5, wherein the rule constraint of the sparrow search algorithm specifically comprises:
The first, the discovery person is generally located in a better position, has higher food amount reserve, can search for the area with more abundant food amount, and provides the position direction of foraging for other follower; in the analogy model, the position where the finder is located depends on the size of the fitness value of the individual, and the smaller the fitness value is, the better the fitness value is;
Secondly, when any individual in the population is found dangerous or frightened, the individual can give out a sound to inform other individuals of the danger, and the position needs to be transferred; when the alarm value is higher than the safety threshold value, the discoverer can lead the population to other safety areas to search food, and if threat still exists at the new position, the population position is continuously transferred;
Third, the identity of both the follower and the third is dynamically changed during foraging of the population; the discoverer has richer food sources, when a certain follower has better food sources, the follower becomes the discoverer, the ratio of the discoverer to the follower is generally unchanged in the population, so that the two are fixed in number in the model, and when a certain follower individual becomes the discoverer, the discoverer becomes the follower, and the identities of the two are dynamically changed;
Fourth, as the concentration of food sources becomes lower as the followers find food in the vicinity of the finder, the poorer their position in the overall population, i.e., the poorer the fitness value, the followers fly to other locations to find food in an attempt to acquire more food sources;
Fifth, during the process of searching for food, the follower always follows the finder with more food sources, and then looks for food from the vicinity of the finder; some followers monitor the discoverers all the time and have a certain probability of competing for food in order to increase their predation rate;
sixth, when individuals in the population find a hazard, individuals at the edges of the population may fly quickly to a safer area, while individuals in the middle of the population may randomly move.
7. An electronic device comprising a memory for storing a computer program and a processor that runs the computer program to cause the electronic device to perform the SSA-based LMS equalization optimization method according to any of claims 1-6.
8. A computer readable storage medium, characterized in that it stores a computer program, which when executed by a processor implements an SSA-based LMS equalization optimization method according to any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410303673.8A CN117896219A (en) | 2024-03-18 | 2024-03-18 | LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Signal processing) optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410303673.8A CN117896219A (en) | 2024-03-18 | 2024-03-18 | LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Signal processing) optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117896219A true CN117896219A (en) | 2024-04-16 |
Family
ID=90641712
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410303673.8A Pending CN117896219A (en) | 2024-03-18 | 2024-03-18 | LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Signal processing) optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117896219A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102137052A (en) * | 2011-03-11 | 2011-07-27 | 哈尔滨工程大学 | Variable step length least mean square channel equilibrium method based on gradient vector |
| CN102883243A (en) * | 2012-10-15 | 2013-01-16 | 苏州上声电子有限公司 | Method and device for balancing frequency response of sound reproduction system through online iteration |
| CN112511473A (en) * | 2021-02-01 | 2021-03-16 | 睿迪纳(南京)电子科技有限公司 | Automatic step length LMS time domain equalization filter and implementation method thereof |
| CN114792066A (en) * | 2022-03-15 | 2022-07-26 | 湖北工业大学 | Wireless charging system compensation network optimization method for improving sparrow search algorithm |
| US20230394316A1 (en) * | 2022-06-07 | 2023-12-07 | Wuhan University | Direct current (dc)/dc converter fault diagnosis method and system based on improved sparrow search algorithm |
| CN117527044A (en) * | 2023-11-08 | 2024-02-06 | 大连大学 | Dynamic deployment method of software defined satellite network controller based on sparrow algorithm |
-
2024
- 2024-03-18 CN CN202410303673.8A patent/CN117896219A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102137052A (en) * | 2011-03-11 | 2011-07-27 | 哈尔滨工程大学 | Variable step length least mean square channel equilibrium method based on gradient vector |
| CN102883243A (en) * | 2012-10-15 | 2013-01-16 | 苏州上声电子有限公司 | Method and device for balancing frequency response of sound reproduction system through online iteration |
| CN112511473A (en) * | 2021-02-01 | 2021-03-16 | 睿迪纳(南京)电子科技有限公司 | Automatic step length LMS time domain equalization filter and implementation method thereof |
| CN114792066A (en) * | 2022-03-15 | 2022-07-26 | 湖北工业大学 | Wireless charging system compensation network optimization method for improving sparrow search algorithm |
| US20230394316A1 (en) * | 2022-06-07 | 2023-12-07 | Wuhan University | Direct current (dc)/dc converter fault diagnosis method and system based on improved sparrow search algorithm |
| CN117527044A (en) * | 2023-11-08 | 2024-02-06 | 大连大学 | Dynamic deployment method of software defined satellite network controller based on sparrow algorithm |
Non-Patent Citations (1)
| Title |
|---|
| CONG ZHOU，CHUN ZHANG等: ""The LMS filtering algorithm optimized using an improved fractional-order cuckoo search algorithm"", 《2023 38TH YOUTH ACADEMIC ANNUAL CONFERENCE OF CHINESE ASSOCIATION OF AUTOMATION (YAC)》, 29 January 2024 (2024-01-29), pages 1345 - 1350 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110460880B (en) | Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network | |
| CN110890930B (en) | Channel prediction method, related equipment and storage medium | |
| Wang et al. | Modulation classification of underwater communication with deep learning network | |
| KR20220031624A (en) | Devices and Methods for Machine Learning Assisted Precoding | |
| CN107947761B (en) | Variable threshold value proportion updating self-adaptive filtering method based on fourth order of least mean square | |
| CN112085050A (en) | Antagonistic attack and defense method and system based on PID controller | |
| CN110601777A (en) | Method for estimating satellite-ground downlink co-channel interference under low-orbit mobile satellite constellation | |
| CN114863226A (en) | Network physical system intrusion detection method | |
| US20230311003A1 (en) | Decision model training method and apparatus, device, storage medium, and program product | |
| CN114707575A (en) | SDN multi-controller deployment method based on AP clustering | |
| CN110190909A (en) | A kind of signal equalizing method and device for optic communication | |
| CN109150237B (en) | Robust multi-user detector design method | |
| CN116579418A (en) | Privacy data protection method for model segmentation optimization under federal edge learning environment | |
| CN117896219A (en) | LMS (least mean Square) balanced optimization method, equipment and medium based on SSA (secure Signal processing) optimization | |
| Zhang et al. | Deepcc: Bridging the gap between congestion control and applications via multiobjective optimization | |
| CN114499601A (en) | Large-scale MIMO signal detection method based on deep learning | |
| CN107070825B (en) | Wavelet weighted multi-mode blind equalization method based on simulated annealing wolf pack optimization | |
| CN107018103B (en) | Wavelet constant modulus blind equalization method based on adaptive step size monkey swarm optimization | |
| CN116633830B (en) | Seed mutation operation scheduling method based on firefly algorithm | |
| CN110474663B (en) | Iterative intelligent signal detection method based on neural network | |
| Xu et al. | Modified constant modulus algorithm based on bat algorithm | |
| CN113036762B (en) | SSA algorithm-based multi-channel power signal mode rapid identification method | |
| CN110505293A (en) | Cooperation caching method based on improved drosophila optimization algorithm in a kind of mist wireless access network | |
| CN110190832A (en) | Become regularization parameter multitask sef-adapting filter network | |
| Li et al. | AIIR-MIX: Multi-Agent Reinforcement Learning Meets Attention Individual Intrinsic Reward Mixing Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |