Application research on piano teaching in colleges and universities based on remote wireless network communication

Before carrying out the system design, we first conduct research on the piano teaching situation in colleges and universities to understand the most basic user needs of teaching. The summary of the survey results is shown in Figure 2. The functions required for piano teaching are networking, intelligent piano equipment control, and remote management. The networking function is the basic guarantee of human–computer interaction and a necessary condition for realising the online to offline mode of piano teaching, which can provide a basic approach for cloud management of teaching topics and updating of teaching materials. The purpose of the networking function is to allocate all intelligent terminals reasonably through a unified network approach, which is convenient for piano teaching to manage and check the students’ devices. The intelligent piano device control function is to run and upload the teaching-end device through a pre-set programme, and conducts information induction and cloud management of teaching materials, such as videos and music scores, to increase the availability of the teaching and learning ends and to improve the user experience. The remote management function can help teaching to achieve equipment control, practice inspection, and improve teaching quality.

Fig. 2

After summarising the system functions, the development goals of the intelligent piano teaching system can be determined: to provide piano video lesson recording, online live teaching functions, visual and auditory effects collection and recording, remote control equipment, online interactive platform, homework inspection and submit the Q&A feature. The system design will be based on the STM32 single-chip microcomputer platform, network the video acquisition equipment and audio acquisition equipment, and cooperate with the software and hardware design of the motor module to achieve comprehensive control of the online piano teaching classroom. Through the coupling of mobile network, wireless Wi-Fi technology and ZigBee communication technology, remote wireless network college piano classroom teaching can be realised, so that teachers can realise non-contact remote teaching; and students can learn in the front-end web page or mobile APP and upload the exercises and other related exercises. Teachers and students communicate with each other. In addition, it is also equipped with visual system maintenance conditions. Through the computer Internet platform, the communication status and connection status of piano teaching equipment can be displayed, so that developers and users can maintain the equipment status and better grasp the system operation status. Most of the mechanical structures adopted in the system are additional structures, and corresponding matching mechanisms are used for application.

Judging from the current communication technology, due to the different communication environments and needs of each part of the online piano teaching in colleges and universities, the communication technology adopted will also be different, and the relevant software and hardware technical requirements and choices are also different [18, 19, 20]. To realise the complete online piano teaching communication, the system communication can be divided into two modules: (1) module 1 is the sensor signal transmission with STM32 microcontroller as the core and (2) module 2 is the information between the teacher terminal and the user terminal and the local piano teaching system Interactive communication. In the needs of sensor signal transmission, the current technical direction can be roughly divided into wired and wireless communications. Although wired communication runs stably, the complex wiring operation not only affects the appearance but also is not easy to maintain and care. Therefore, we abandoned the wired communication method and switched to the construction of the underlying structure of the wireless communication method. At present, the most used wireless communication technologies on the market include Wi-Fi technology, ZigBee technology, Bluetooth technology and NFC technology [21, 22, 23, 24]. Among them, Wi-Fi technology has the advantages of wide signal coverage and fast signal transmission, but it may have an impact on piano teaching due to network fluctuations, especially in live teaching. ZigBee technology has the advantages of low cost and low power consumption, but the transmission rate is slow. Bluetooth technology has high security, but it has poor fault tolerance in embedded development and poor compatibility with various technical architectures. NFC technology also has high security, but its transmission distance is too short. Therefore, combined with the needs of online piano teaching and the characteristics of signal transmission at the bottom perception layer, we have selected two communication methods to ensure the safety and effectiveness of information interaction. Among them, ZigBee technology is used as the main communication method of the signal sensing layer, which can meet the data volume of the bottom signal sensing layer. Wi-Fi technology is used as an auxiliary communication method to meet the communication detection function requirements due to the addition of a computer interface in the design.

In the second choice of information exchange communication, to realise the information cross-linking between the online teaching system and the user terminal, and considering that the huge online teaching market in colleges and universities needs to be promoted in a low-cost way, the controller signal and cloud server need to be promoted first. The connection is realised by using the Wi-Fi routing network with the technical route of the computer, and with the development of the mobile APP client and the web port to meet the operation and learning needs of different users.

Based on the needs of online piano teaching, the system functions need to have the functions of uploading and downloading teaching materials and remote control. Among them, the remote-control function is that the processor with the STM32 as the core controls the piano equipment remotely through logical operations such as analysis, calculation, arrangement, etc. after receiving the signal from each terminal sensor, combined with the instructions required by the teacher. Both teachers and students can debug the equipment in the piano classroom through the mobile phone APP client and teach after the performance. The teaching screen is recorded and uploaded by the multi-angle cameras arranged in the classroom. The uploading method can be connected to Wi-Fi, ≥ 10 m, it can also be the full coverage area of the signal after connecting to the 4G communication network. In general, the priority of user-based remote control is higher than the automatic control of the STM32 microcontroller. When the user does not perform remote control, the automatic control will be carried out spontaneously. The overall system structure diagram is shown in Figure 3.

Fig. 3

The implementation process of the specific communication process is as follows:

Communication node 1: Mobile phone signal accesses the network, the development web page interacts with users, uses JSON functions in the background to obtain the status of user operations, leases a cloud server, deploys a cloud TOMCAT server, and then encapsulates the web page. The communication IP points to the router through the TOMCAT server.

The development process of the software programme can be determined from the hardware structure of all levels of remote wireless network communication. Among them, the first thing that needs to be done is the interaction between the information obtained by the underlying sensors and the work orders of the mechanical structure. After that, the information is translated through the logic layer and then transmitted to the network communication module [25]. Because the work coordination of the Zigbee communication protocol and each terminal module in the system are the most basic information exchange process in the whole system and the largest development workload, the IAR Embedded Workbench development environment is selected for the software development of the STM32 microcontroller.

IAR Embedded Workbench is a top-level integrated development environment in the industry, referred to as ‘IAR’. It is an integrated software programme development environment developed by the world-renowned Swedish company IAR Systems for microprocessors. It enjoys high authority in the industry. The software development integrated environment supports ARM, AVR, MSP430 and other chip platforms, and also supports C/C++ language and other functions such as development, debugging, and online simulation.

In the funny piano teaching system based on remote wireless network communication technology, the control programme loaded into the microcontroller and other terminal control modules has three tasks to be completed [26].

Control the recording start or stop of the audio and video equipment required for the recording of piano lessons, and control the recorded video to be uploaded to the cloud server through the Wi-Fi routing network or 4G communication network.

To realise the above three control tasks, modular programming is carried out through C language, which is divided into ZigBee communication, W-Fi communication, sensor signal acquisition, generating PWM waveform to control the motor, remote/automatic working state switching, and other programme modules. The composition and control flow are shown in Figure 4.

Fig. 4

The model of the Wi-Fi module used in this design is ESP8266, which can be connected to the router equipment in the school classroom to realise remote control of motors, audio and video equipment, etc. The specific control steps are as follows:

The ZigBee wireless communication network sends the signal to the STM32 master for data processing.

The ESP8266Wi-Fi module supports three working modes: STA, AP, and STA+AP. The STM32 master sends the corresponding AT command to the ESP8266 module through serial communication to convert it to the corresponding working mode. By comparing the functions and characteristics of the three working modes, it is determined to adopt the STA working mode. The specific steps and sending AT commands are:

Convert to STA working mode: AT+CWMODE=1;

ZigBee is a short-range wireless communication technology. It is essentially a two-way wireless networking technology with a relatively low transmission rate. As a communication protocol, it can wirelessly network different digital devices so that they can communicate with each other and provide standardised instructions for the communication process [27]. To network through the ZigBee communication protocol, you can use the ZigBee protocol stack. The so-called protocol stack is the specific implementation form of the protocol, and it is the interface for developers to standardise the use of the protocol. Developers can programme through the API interface provided by the ZigBee protocol stack to efficiently implement ZigBee networking, which can greatly reduce the workload while ensuring standardisation [28].

The workflow of the ZigBee protocol stack can be roughly divided into several steps: system startup, driver initialisation, operating system abstraction layer initialisation and startup, and entering task polling. System initialisation is realised by the function osal_init_system, which realises the initialisation of hardware abstraction layer, network layer, tasks, etc. Among them, isalInitTasks is the initialisation function of the operating system task. After the system is initialised, the function osal_start_znp is called to poll the processing tasks. This function is a loop structure that continuously processes the task with the highest priority among all the tasks currently in the ready state.

In the long-distance wireless network communication system, the ZigBee wireless communication network usually needs to be connected to many devices, sensors, actuators and other nodes, because each node will generate energy loss during the routing process, and there will be new nodes in the process of use. It is added to the ZigBee communication network, so it is necessary to optimise the ZigBee routing structure through algorithms, reduce the path length, reduce network bottleneck nodes, and improve the reliability and stability of the entire network in the communication process. At present, there are two kinds of optimisation algorithms commonly used in ZigBee routing protocol: one is the Cluster-tree algorithm and the other is the On-demand Distance Vector (AODVjr) algorithm. The Cluster-tree algorithm is the most traditional and direct ZigBee routing protocol optimisation algorithm. It starts from the main coordinator, performs layering and traversal according to the tree structure, and forms a cluster tree network topology.

The address assignment of the Cluster-tree algorithm follows the following method:

The main coordinator establishes a new communication network, its own address is 0, and the depth d₀ = 0 in the network;

The routing strategy of Cluster-tree is a routing node calculates the next hop of the packet according to the destination network address of the received packet. Suppose the router network address is A and the depth is d, then the router will first pass the formula (1) A<D<A+Cskip(d−1) A < D < A + {C_{skip}}\left({d - 1} \right) Determine whether the destination node is its own descendant node, if it is true, it is, and if it is not true, it is not. If the destination node is a descendant node of A, the following formula must be judged (2) D>A+Rm*Cskip(d) D > A + Rm*{C_{skip}}\left(d \right) If it is true, that is, the destination node is the terminal child node of A, then the address N of the next hop node is D, and it can be sent directly to D; otherwise, if the destination node is other nodes of A, it is required that the next hop node is A which routing child node of, send to this child node, the calculation method is as follows (3) A+1+[D−(A+1)Cskip(d)]×Cskip(d) A + 1 + \left[ {{{D - \left({A + 1} \right)} \over {{C_{skip}}\left(d \right)}}} \right] \times {C_{skip}}\left(d \right) If Eq. (1) does not hold, it means that D is not the descendant node of A, and the next hop node is the parent node of A.

In the C4.5 algorithm in the Cluster-tree algorithm, for a given data set D, if there are m unequal value class label attributes, it is divided into m different classes Ci (i = 1, 2, …, m), denote the set of tuples corresponding to class Ci in the data partition D as Ci,d, where D represents the number of tuples in D, and Ci,d represents the number of tuples:

For the desired information needed: (4) Info(D)=−∑i=1mpilog2(pi) Info\left(D \right) = - \sum\limits_{i = 1}^m {p_i}{\log _2}\left({{p_i}} \right) Among them, p_i is the probability that the tuple belongs to class Ci, and Info(D) represents the entropy of D. Assuming that attribute A has v unequal values {a₁,a₂,…,a_v}, D is divided into v different subsets {D₁,D₂,…,D_v}, whose Dj represents that on A has a tuple of value a_j that is contained in D. Therefore, the information entropy obtained by calculating attribute A is: (5) InfoA(D)=∑j=1v[Dj][D]×Info(Di) Info_{A}\left(D \right) = \sum\limits_{j = 1}^v {{\left[ {{D_j}} \right]} \over {\left[ D \right]}} \times Info\left({{D_i}} \right) For the corresponding information gain values: (6) Gain(A)=Info(D)−InfoA(D) Gain\left(A \right) = Info\left(D \right) - Inf{o_A}\left(D \right) Next, the training sample set is divided with the value of attribute A as the reference standard, and its initial information amount SplitInfo_A(D) is: (7) SplitInfoA(D)=−∑j=1v[Dj][D]×log2([Dj][D]) SplitInfo_{A}\left(D \right) = - \sum\limits_{j = 1}^v {{\left[ {{D_j}} \right]} \over {\left[ D \right]}} \times {\log _2}\left({{{\left[ {{D_j}} \right]} \over {\left[ D \right]}}} \right) The information gain rate is the ratio of the information gain to the initial amount of information (8) GainRatio(A)=Gain(A)SplitInfoA(D) GainRatio\left(A \right) = {{Gain\left(A \right)} \over {SplitInf{o_A}\left(D \right)}} That is to say, the information gain rate is the information gain obtained by the unit initial information amount and is the relative information amount uncertainty measure.

Choose a fitness function: (9) f(x)=w1×x1+1w2×x2+w3×x3 f\left(x \right) = {w_{1 \times}}{x_1} + {1 \over {{w_{2 \times}}{x_2} + {w_{3 \times}}{x_3}}} Among them, x₁ represents the classification accuracy of the rule set; x₂ represents the number of all rules in the rule set; x₃ represents the number of “attribute name = attribute value” in the rule set, that is, the complexity of the rule set; w₁, w₂, w₃ represent the impact factor of x₁, x₂, x₃.

Denote the fitness of each selection of the algorithm as f_i, then the probability of operation i being selected and copied is defined as: (10) Pi=fi∑j=1nfi, i=1,2,…,n {P_i} = {{{f_i}} \over {\sum\limits_{j = 1}^n {f_i}}},\quad i = 1,2, \ldots,n