Abstract
Enterprise is an indispensable factor of production in the country’s economic and social development, and the operation of the enterprise is indispensable. The development of strategic operation of the enterprise is facing many problems, and the survival of the enterprise is facing serious threats. The purpose of this article is to study the method of establishing a reliable enterprise strategic management system based on GPRS wireless communication and neural network. Let the company continue to grow in a sustainable and healthy way. This article puts forward the importance of corporate strategic management under GPRS wireless communication. Strategic business management can improve the foresight and initiative of enterprises, overcome short-term behaviour, and provide a clear direction for the development of the enterprise. In the experimental data of this article, it can be seen from 2016 that the demand for talent management by enterprises is lower than that in 2019. By 2019, enterprises will have the demand is as high as 85.3%, so enterprise management development needs should be taken seriously. The error between the actual output of the network and the expected output is controlled within 5%, which shows that the established neural network has a good evaluation effect and can be used to evaluate the talents of business operators. This also shows that the established BP neural network can fully absorb the judgment experience of experts and the actual employment of enterprises. The results show that the evaluation results obtained according to the evaluation network model have certain guiding significance for the selection and assessment of employers and the self-evaluation of management talents.
1. Introduction
While adapting to mobile Internet life, the requirements for the transmission quality and transmission speed of the wireless communication system are also increasing. From the wired network to the wireless network, the difference is not only the physical connection method but also changes the way people use the network and behavior habits. The era of a phone line or a network cable has gradually passed, and wireless Internet access anytime, anywhere has become an inevitable trend. The essence of operation is to optimize the allocation of corporate resources and ultimately achieve the functional activities of the organization’s goals. The ultimate goal of operation is to maximize the benefits of the enterprise. The development of an enterprise depends on effective operation, and the realization of operation depends on excellent operation to realize the added value of the enterprise. Therefore, it is more important to study the construction method of the enterprise management talent evaluation system according to the actual situation of the enterprise. This article’s research on the evaluation of corporate management talents will optimize the current talent evaluation methods and evaluation processes and balance the impact of expert experience and corporate employment tendencies on the results of talent evaluation. Achieve consistent talent evaluation results based on actual business conditions.
By strengthening the management of the development strategy, it will help companies establish a more scientific and appropriate development strategy plan to ensure the development of the company. Especially with regard to risk countermeasures capabilities, companies still have big shortcomings. Therefore, the group’s risk management system should be based on a strategic development plan to track risks in the implementation of the strategy in order to address them. In order to improve the response to various business risks and improve the company’s management level and risk countermeasure capabilities, companies need to establish a more complete development strategy management mechanism.
Bennism proposed that 5 g wireless network is very important and has attracted great attention from academia and industry. Fundamentally speaking, the network design method is the most effective. On the contrary, there is still a lack of a reliable process to make decisions. In order to realize this vision, after providing the definition of delay and reliability, we carefully studied the various contributing factors of URLLC and its inherent trade-offs. Then, focus on the various technologies and methods related to URLLC requirements, and their application through selected use cases [1]. Zhang Z pointed out that the main challenge encountered at present is that some interference and signal differences are encountered when implementing FD wireless devices. In this paper, their advantages and disadvantages are listed and which device is better compared [2]. In order to adapt to the field, Ganini introduced a new method of learning. In this method, the architecture of the trained neural network is prominent. This method is influenced by the adaptive theory. As the training progresses, this method promotes the emergence of the following functions: (i) the main learning tasks of the source domain can be distinguished, and (ii) the conversion between domains cannot be distinguished. According to experiments, by expanding with several standard layers and new gradient reversal layers, this kind of adaptive action can be realized by almost all feed forward models. [3]. Classification is the classification method that Jinnianlei has been paying attention to in data resources in recent years. This is one of the data mining methods. On this basis, the neural network method is proposed, but the research found that it is not suitable for data mining method. This method can be used. Accurately extract simple symbol companies from the neural network. In order to meet the requirements, the network must be trained. Then use the network to analyze the activation value [4]. In this paper, Jaderberg proposes a text recognition system, that is, text location, recognition in natural scene images, and text-based image retrieval. Then the convolution neural network is trained to recognize the words in the whole proposal area at the same time, which is different from the previous system based on character classifier. It shows the best advantages of this method, and strict experiments are carried out. [5]. Alnoukari proposed that the integration of business intelligence and corporate strategic management has a direct impact on modern and flexible organizations. This integration helps decision-makers implement their corporate strategy, easily adapt to changes in the environment, and gain a competitive advantage. A BSC-BI framework is proposed, which promotes the integration of business intelligence and balanced scorecard methods. Case studies in the field of telecommunications demonstrate the implementation of the BSC-BI framework [6]. The purpose of VitollaF is to fill the gaps in the existing literature. These documents not only involve the application of social-oriented philosophy to the theme of strategic corporate social responsibility (CSR) integration but also the systematic analysis of the strategic CSR management process, and the creation of a social management philosophy and foundation. The link between the CSR integration dynamic approach of the strategic management process. What is the most important strategic management process for integrating corporate social responsibility according to the concept of social management [7]. Liston-Heyes proposed a planned corporate environmental behavior model, which emphasizes the values and attitudes of managers to the environment, environmental intentions, and the background in which these intentions are formed and transformed into actual performance. The mentality shared by those who give the country the responsibility to protect the environment. A series of hypotheses are generated and tested in the context of a unique data set using structural equation models [8]. Through the research experiments of scholars, we know that there is no advanced and complete mechanism for corporate strategic management, and companies are facing management problems. Therefore, strengthening corporate strategic management and developing a good management system are the top priorities.
The innovations of this article are as follows: (1) the research on the evaluation of enterprise management talents theoretically enriches and expands the scope of use and evaluation content of talent evaluation. From the perspective of research methods, it is introduced in the process of enterprise management talent evaluation. (2) Introduce the theoretical knowledge and related practice of GPRS and develop a set of highly reliable, complete, and true system based on GPRS wireless communication for corporate strategic management.
2. Signal Detection Algorithm of 1 MIMO System Based on GPRS
2.1. GPRS
GPRS is the abbreviation of General Packet Radio Service, and it is a mobile data service available to GSM mobile phone users [9]. GPRS sends and receives data in the form of frames, and the charging method has also changed from the traditional on-time charging to flow-based charging. In the long run, this approach can greatly reduce the user’s cost of use [10]. As long as no data is being transferred, you do not have to pay even if you are always “online.” To use a “phone call” analogy, when using a GSM+WAP mobile phone to access the Internet, it is like a phone call that starts to be billed as soon as it is connected, whereas with GPRS+WAP it is much more reasonable, like a phone call that is not charged as soon as it is connected, but only when the conversation is calculated. In short, it truly embodies the principle of pay less for less use and pay more for more use. The system structure diagram of GPRS is shown in Figure 1.
2.1.1. 1 MIMO System
1 MIMO system (multiple-input multiple-output, MIMO) is a technology that uses multiple antennas for wireless transmission and reception of data. By using this technology, the data transmission speed can be doubled [11]. The system model is shown in Figure 2.
Figure 2 shows a V-BLAST MIMO system model with transmitting antennas and () receiving antennas [12] as shown in Formula (1):
The corresponding received signal vector is expressed as Formula (2):
The multipath effect caused by the multiantenna transmission signal makes the symbol information transmitted in the MIMO system superimpose on each antenna at the receiving end, as Formula (3):
2.1.2. MIMO Channel Capacity Analysis
According to different input and output antenna arrangements, the multiantenna system can be divided into three types: single-input multiple-output SIMO system, multiple-input single-output MISO system, and multiple-input multiple-output MIMO system [13]. SVD is performed on the channel matrix through singular value decomposition. Operation can get formula (4):
Among them, and are both Emirates matrices and satisfy ; is a diagonal matrix, and the elements on the diagonal are the singular values of matrix , that is, the arithmetic square root of the eigenvalues of , which brings the above formula into the MIMO system vector model get Formula (5):
Formula (5) is obtained by multiplying to the left of Formula (6):
Since is a diagonal matrix, if is used to represent the singular value of matrix , then there is Formula (7):
It can be seen from the above Formula (7) that the equivalent receiving element (when ) is independent of the transmitted symbol, which is equivalent to thinking that the channel gain is 0 to obtain Formula (8):
is the noise power; represents the received power of the i-th antenna at the receiving end. Since the channel state information at the transmitting end is unknown, the method of equal distribution of all antennas is often used when power allocation is performed. Assuming that the total transmit power is , then the total MIMO channel capacity is Formula (9):
Formula (10) can be obtained by derivation:
The experimental analysis surface usually assumes that the bandwidth is large enough. Because the channel state has certain randomness, the expected capacity is usually taken as Formula (11) in the experiment:
Formula (11) carries out the channel capacity test of the MIMO system with the number of antennas [1], [2], [4], [8], [14], where the number of antennas is [1] The SISO system can be regarded as a special case of MIMO system.
2.2. Signal Detection Algorithm of MIMO System
MIMO signal detection is based on the known receiving end signal vector , and the channel state information (ISI) matrix obtained through channel estimation, and the environmental noise subject to the Gaussian distribution with a mean value of 0 and a variance of . A process of estimating the value of [15]. Signal detection technology is the key to the smooth application of MIMO wireless communication technology. The research of detection algorithms with good performance and relatively lower complexity is one of the hot topics in the field of MIMO wireless communication [16].
2.2.1. Maximum Likelihood Detection Algorithm
The maximum likelihood detection algorithm is abbreviated as the ML algorithm [17]. and since the system model is formula (12):
The maximum likelihood algorithm needs to search and calculate all possible combinations of the signal vector b (these combinations are called candidate solutions) one by one, Formula (13): is a known parameter, the conditional probability distribution mentioned earlier is equivalent to the following Gaussian distribution, which is Formula (14):
Maximizing the conditional probability is equivalent to , so the maximum likelihood detection algorithm can be expressed by Formula (15):
Its physical meaning is to represent the energy of the signal, represents the signal space, which represents the value space of the i-th element of the signal vector [18]. In order to obtain the optimal solution that satisfies the conditions, so the ML algorithm is a kind of all possible symbols for the sender. Combining the ergodic search algorithm, the best detection performance can be achieved by finding the signal point where the cost function can be minimized as the best emission vector [19]. Because of its traversal search characteristics, ML algorithm has become the best detection algorithm in MIMO system. It can not only make the detection error rate reach the lowest but also can obtain all the hierarchical gains of the system.
2.3. Interference Cancellation Detection Algorithm
Serial interference cancellation detection algorithm was first proposed by scholars. Its basic principle is improved on the basis of wireless communication [20]. It detects each symbol in turn, eliminates the influence of detection symbols, and improves the purpose of detection. The basic flow and steps are as follows: the SiC detection algorithm based on wireless communication is the detection algorithm.
Zero forcing detection of a single symbol: use the detection algorithm to detect the first element of the transmission vector, as shown in Formula (16):
Interference cancellation: after the detection in the previous step, the first symbol has been recovered, namely , and the the detected is subtracted from the received signal to obtain a new signal model to be detected, the form of which is shown in Formula (17):
is the first column of the channel matrix , and is the equivalent received signal vector obtained after eliminating the signal component . Through the continuous iterative calculation of the above two steps, all the transmitted symbols can be finally detected and recovered NS. From the previous algorithm analysis, it can be seen that if the first detected signal has an error, and it will interfere with the detection of the following symbols. Therefore, the SIC algorithm has an error propagation phenomenon. In order to reduce the impact of this phenomenon, increase the detection accuracy. Because even the linear detection algorithm can obtain more accurate detection results under the condition of large signal-to-noise ratio, the priority is to detect the symbols with large received signal-to-noise ratio, and to detect the symbols of each layer in order of the signal-to-noise ratio can effectively reduce the error propagation. The impact of this will improve the detection accuracy of the system [21]. The detection signal-to-noise of the i-th layer is shown in Formula (18):
Compared with the SIC detection algorithm, OSIC detection algorithm adds a sorting step, so the OSIC algorithm can be summarized as: sorting-zeroing-interference cancellation three parts. Taking the zero-forcing sequencing serial interference cancellation algorithm as an example, the specific algorithm flow chart is shown in Figure 3:
As shown in Figure 3, compared with the SIC detection algorithm OSIC detection algorithm adds a sorting step, so the OSIC algorithm can be summarized as sorting-zeroing-interference cancellation three parts. Take the zero-forcing sequencing serial interference cancellation algorithm as an example. In the algorithm, the row with the smallest second norm in the filter matrix is obtained by calculation. This process is equivalent to obtaining the layer with the largest signal-to-noise ratio as the current layer to be detected. The largest layer is the current detection layer.
2.4. SD Sphere Detection Algorithm
The algorithm is based on the maximum likelihood criterion like the ML algorithm. The basic idea is to restrict or narrow the search range by determining a multidimensional sphere with a vector as the center and a radius to reduce the number of candidate solutions that need to be searched. Thereby, the computational complexity of the maximum likelihood algorithm is reduced, the operation speed is improved, and the detection accuracy is ensured [22]. The realization of this algorithm needs to solve two main problems: how to determine the appropriate search radius , and how to determine the search grid points are within the range of the sphere. As shown in Figure 4:
Usually it is very difficult to select a very ideal radius. The radius cannot be selected too small, otherwise there may be no optimal solution or even no grid points in the range of the sphere, and the radius selection is too large, and it cannot reduce the calculation complexity. In order to determine whether the point is in the circle or not, you only need to determine whether its coordinate range meets the conditions one by one. Since is an upper triangular square matrix, searching upwards from the Nth layer by layer can finally determine whether the point is within the search range.
When , as shown in Formula (19):
From this, the upper and lower limits of can be determined as Formula (20):
2.5. Neural Network Method
This paper applies the constructed enterprise management talent evaluation model to the evaluation of enterprise management talents, uses the weights determined by the entropy method to calculate the evaluation results of the research objects, and uses the trained network as an evaluation tool. Use the same sample other than the training sample set [23] for evaluation. The BP network uses a continuous differential possible function with specific threshold characteristics as the activation function of the neuron. Of course, other similar nonlinear functions can also be used. If we combine neural network and enterprise talent management model, we can get the result which shown in Figure 5:
3. Experiment and Analysis
3.1. Entropy Method Experiment and Analysis
Part of the learning sample set of this study is selected from the entropy method used in the previous research to determine the index weight data, and the other part of the learning sample comes from the evaluation data of the investigation of talents who cannot be managed by the same department in different enterprises [24]. Although the data sets are from different sources, the meaning embodied in the data gives a good indication of the differences between the topics, and therefore no comparative data are involved in this paper. Since the value definition of different management talents by the human resources department of the enterprise also has certain learning significance, the combination of the two parts of learning samples can finally train a neural network that not only conforms to the evaluation results of the expert system but also combines the actual situation of the enterprise. The input indicators of the learning sample and the weights of each indicator are shown in Table 1:
From Table 1, it can be seen that the feasibility of using salary level as an output variable in this paper is first demonstrated theoretically. In order to make the final analysis result more scientific, the salary of management talents of the same level in different companies may vary greatly. This is mainly reflected in the fact that different industries have different practices, different management philosophies and remuneration strategies, different geographical locations of different companies, different levels of value or contribution to the company, and the length of time the incumbent has been in the position. Normalization it is also more difficult, so by consulting the information, this article uses the logarithmic function to reduce the salary level data difference. After the data difference is reduced, then learn from the previous normalization method to directly divide the output value of the original data by the maximum value of the data set value and get the output data after normalization. According to the commonality of management talents pointed out in the previous study, management talents actually create value for the enterprise, and the amount of value created has the same measurement standard in the same enterprise, and the amount of value created corresponds to the salary level of the position. Pay structure design is the process of establishing a linear or nonlinear link between the relative value of each job position in an organisation and the corresponding out-of-pocket pay. It allows the level of pay for each job position to correspond to its relative value. There is an inevitable connection between them. In this sense, it is reasonable to use the salary level corresponding to the position as the output of the comprehensive evaluation of management talents. The trend chart of a company’s demand for talent management in 2016 and 2019 is shown in Figure 6:
It can be seen from Figure 6 that in 2016, it can be seen that the demand for talent management by enterprises is lower than that in 2019. By 2019, the demand for talent management by enterprises is as high as 85.3%, so the demand for enterprise management development should be taken seriously. The manager’s management activities are to maximize the benefits and the purpose of the enterprise’s management activities. Behavioral science theory believes that the essence of management is to improve work efficiency. On the surface, the two seem to be different, but in general, the direct purpose of management is to improve work efficiency, and the improvement of work efficiency ultimately serves to increase profits, so the essence of management In fact, it is to realize the maximization of profits, which is the process of turning limited input into maximizing output.
Finally, through comprehensive evaluation and data statistics, the learning sample set required for neural network modeling is obtained as shown in Table 2:
It can be seen from Table 2, the final evaluation object, for the score of the indicator, part of it comes from the comprehensive evaluation process of experts, combined with the weight obtained by the value entropy method, as the input of the learning sample, the comprehensive evaluation score of each sample can be obtained. And normalized processing, so that the output of the learning sample is obtained, and the other part of the learning sample selects the salary level corresponding to the survey sample as the output [25].
3.2. Experiment and Analysis of BP Neural Network
Entropy is a measure of uncertainty. The greater the amount of information, the smaller the uncertainty and the smaller the entropy; the smaller the amount of information, the greater the uncertainty and the larger the entropy. According to the properties of entropy, we can calculate the entropy value to judge the randomness and disorder of an event, and we can also use the entropy value to judge the degree of dispersion of a certain indicator, the greater the degree of dispersion of the indicator, the greater the influence of the indicator on the comprehensive evaluation. It is hoped that the network model can absorb the evaluation experience of experts, and at the same time, it will test the samples are judged with high precision. NeruoSolutions, a very useful software at present, has a visual operation interface, which can directly perform network creation and network testing, and can also directly manipulate excel files to complete the learning process. As shown in Figure 7.
Figure 7 selects the standardized worksheet interface, clicks the train option under the menu TrainNetwork to set the number of iterations to 6000, trains the network, and then clicks the test option in TestNetwork to test the data. Through network learning and testing, you can select testing in the main interface NeruoSolutions to get the desired and the actual output, this is the output and expected output of the network after training.
It can be seen from Figure 8 that the network input value after network training is very close to the expected output result, indicating that the trained BP network has achieved a good learning effect. After the BP neural network training is completed, the test sample set is started. The data is tested on the network, the comprehensive evaluation results of the enterprise management talents are obtained, and the test results in the following Table 3 are obtained. Finally compare the plan with the actual, and the network output curve in Figure 9 is obtained.
It can be seen from Table 3 that the neural network must also integrate the actual employment of the enterprise, and the enterprise has a maximum expectation of 96% of talent management capabilities. The importance of enterprise talent management is obvious. Therefore, the learning sample set must not only include comprehensive evaluation data as output but also reflect the actual evaluation tendency of enterprise talents. According to the actual value contribution of management talents to the enterprise and the salary feedback of the enterprise to management talents, it must satisfy a certain positive relationship. Therefore, in order to obtain output data that is more representative of different companies’ employing tendencies and is easy to obtain, this paper selects the salary level corresponding to enterprise management talents to represent the actual results of the company’s evaluation of talents and regards the salary level as the actual output of the network.
As shown in Figure 9, the neural network should be combined with the talent management of the enterprise. The highest expectation of the enterprise for talent management ability is 96%. It shows that enterprise talent management is very important. Therefore, the learning sample set should not only include the comprehensive score but also reflect the actual score of enterprise management talents. According to the actual value contribution of management talents to the enterprise, it is necessary to make more scientific judgments on the data about their value contribution tested. At the same time, based on the empirical evidence, it further confirms the use of the BP neural network to evaluate the business ability of the enterprise, scientificity, and feasibility. The experimental results show that applying the entropy method and BP neural network to the evaluation process of enterprise management talents can obtain better evaluation results, providing a new way of thinking for the evaluation of enterprise management talents, and the evaluation process is faster and more accurate. According to the evaluation network model, the evaluation results have certain guiding significance for the selection and assessment of employers and the self-evaluation of management talents.
4. Discussion
This article analyzes how to develop a set of highly reliable, complete, and true system for corporate strategic management based on GPRS wireless communication. Strengthen development strategy management, help companies establish more scientific and appropriate development strategy planning, so as to ensure the company’s development. The theoretical knowledge of GPRS wireless communication and neural network is explained. Due to the development of digital communication technology and the pursuit of high-quality wireless communication by users, after the end of the century, the wireless communication system has developed into a second-generation wireless communication system represented by digital communication technology. These systems use more advanced systems. Digital technology hopes to use the various advantages of wireless networks to design a reliable enterprise strategic management mechanism that can be accepted by the public. This paper makes reasonable use of the signal detection algorithm of the 1 MIMO system. The ML algorithm has become the detection algorithm with the best performance in the MIMO system because of its traversal search characteristics. The basic idea of establishing a neural network in the experimental analysis of this paper is to briefly analyze the number of network layers, the number of neurons in each layer, the learning rate of neurons, and the activation function, and then use NeruoSolutions software. Train and test the BP network model and analyze the experimental results. Through the analysis of the experimental results, we can see that the network model has a good evaluation effect. The next step is to look at the technical or other improvements that can be made to the network to make it more widely used.
5. Conclusions
This article mainly starts from the theoretical knowledge of GPRS wireless communication and neural network and discusses how to establish a set of highly reliable business management strategy development model based on GPRS wireless communication and neural network. Based on the signal detection algorithm and neural network model of the 1 MIMO system, it can be seen that the BP neural network is used in the process of evaluating enterprise management talents, and it can obtain better evaluation results and providing a new idea for the evaluation of enterprise management talents and the evaluation process. Faster and more accurate, the evaluation results obtained according to the evaluation network model have certain guiding significance for the selection and assessment of employers and the self-evaluation of management talents. For example, the network model can be used in large companies where the actual situation of a person cannot be determined based on information about that person, but where a conclusion is urgently needed, so that a relatively reasonable assessment can be made. Based on the wide range of related scientific fields involved in the GPRS Wireless Communication Application Research Institute, the concept of enterprise management has always been disputed, and the author’s knowledge has not yet reached the perfect state. The author’s academic theory and business capabilities are relatively weak, and there are still many deficiencies. At the same time, the author is constantly discovering and solving problems, striving to be the best.
Data Availability
No data were used to support this study.
Conflicts of Interest
The author states that this article has no conflict of interest.
Acknowledgments
This work was supported by Social Science Planning Project of Shandong Province (Research on the realization path of inclusive finance in Shandong Province based on ecosystem perspective; Project No:(19CJRJ05).