Abstract: The combination of embedded devices and the Internet represents the true future of embedded technology. This paper presents an embedded network intelligent control system that combines an embedded operating system with the Internet. First, the architecture of the embedded network intelligent control system based on a Web server is given. Then, the software system is designed, and the message format for information exchange between the embedded network intelligent control terminal and the Web server is presented. A parameter calculation method and load scheduling algorithm for the scheduling server are proposed. Finally, experimental results and conclusions are given. Keywords: Embedded operating system; Web server; Scheduling algorithm 1 Introduction With the development of Internet technology, the widespread application of information appliances, and the miniaturization and specialization of EOS (Embedded Operating System), EOS is beginning to evolve from single, weak functions to highly specialized, strong functions. Embedded systems, marked by Internet-based architecture, are in a stage of rapid development. Currently, most embedded systems are still isolated from the Internet, but with the development of the Internet and the increasingly close integration of Internet technology with information appliances, industrial control technology, etc., the combination of embedded devices and the Internet will represent the true future of embedded technology. The embedded network intelligent terminal control system studied in this paper organically combines the embedded operating system and the Internet by directly connecting the microprocessor system to Ethernet. The following will describe the system architecture and software system of the embedded network intelligent control system. 2 System Architecture of Embedded Network Intelligent Control System The application environment and network topology of the embedded network intelligent control system are shown in Figure 1. The topology shown in Figure 1 consists of an embedded network intelligent terminal, a scheduling server, a Web server group [2], and an Internet host. The embedded network intelligent terminal uses the uCLinux operating system [1]. The terminal is directly connected to the Ethernet and transmits data to the Web server of the Internet wide area network. The scheduling server plays the role of load scheduling in this system, so that no server in the Web server is overloaded. The Web server group collects data from the embedded network intelligent terminal system and responds to the access of the wide area network host. The host in the wide area network can randomly access the Web server to obtain data from the embedded network intelligent terminal, and the host with specific permissions can directly access the embedded network intelligent terminal. Figure 1 System application environment and network topology The embedded network intelligent terminal consists of a microprocessor, external interface, network information processing, level conversion, extended memory, etc. The hardware structure of the embedded network intelligent terminal is shown in Figure 2. Figure 2 Hardware structure of embedded network intelligent terminal 3 Software system 3.1 Software architecture of embedded network intelligent terminal According to the environment shown in Figure 1, this paper selects uClinux operating system as the software system development platform. Figure 3 shows the overall architecture of the software system. The software system is divided into two parts: the controlled object software part and the embedded network intelligent terminal software part. The controlled object software part involves how the sensor collects the signal of the controlled object, how to drive the controlled object and peripherals, and how to transmit the signal to the embedded network intelligent control terminal. The embedded network intelligent terminal software part involves kernel modification, operating system porting, file processing, design of graphical user interface, task processing, how to add application to uClinux, and information exchange between the embedded network intelligent control terminal and the Web server. Figure 3 Overall architecture of the software system 3.2 Information exchange between embedded network intelligent control terminal and Web server The Web server is an important means of exporting information from remote devices. It can allow modification of device parameters through the form of the Common Gateway Interface (CGI) to perform remote monitoring [2]. The Web server has become the standard for remote management, especially for devices without traditional user interfaces (such as the embedded intelligent terminal designed in this paper). The Web browser can communicate with the remote device and display the corresponding data. In the topology environment shown in Figure 1, the Web server, client and embedded network intelligent terminal use Hypertext Protocol [3] (HTTP) and network programming to realize information exchange and remote management. HTTP is an ASCII-based protocol. HTTP uses the standard synchronous request/response mode and client/server structure on top of the TCP/IP protocol [4]. When the client makes a request to the HTTP server, it needs to send an HTTP request message, the format of which is shown in Figure 4. Figure 4 HTTP request message format Figure 4 shows that the HTTP request message consists of a request line, optional headers and optional message body fields. The request line establishes the stage for subsequent activities. The optional headers indicate the protocol requested by the client and the information the server needs to know about the client before responding. The HTTP request message contains very little information, but through this information, the server can be informed of its "capabilities and preferences". Similarly, the HTTP response adopts a similar structure, which will not be elaborated here. 3.3 Network data submission When the embedded network intelligent terminal application submits network data (the application in the software layer of Figure 3), the network communication between the embedded network intelligent terminal and the wide area network Web server adopts the client/server mode, that is, network programming is performed in the uCLinux operating system environment. Implementing network programming (information transmission) includes the following five basic processes: Creating a socket: calling socket(); Binding a socket: calling bind(); Creating a connection: calling connect(); Listening for connections: calling listen(); Establishing a connection: calling accept(). 3.4 Scheduling Algorithm of the Scheduling Server A key factor affecting the reliability of the embedded network intelligent control system shown in Figure 1 is the network performance of the scheduling server. Reasonable parameter calculation, mapping, and load positioning are crucial for improving network performance. 3.4.1 Parameter Calculation and Mapping CPU utilization directly reflects server performance. The following describes the process of obtaining CPU utilization in the Linux operating system. In the Linux operating system, this is achieved by reading the /proc/stat file, which records the variables sys, user, nice, and idle. When calculating CPU utilization, these variables only need to be read twice, and then the formula is used: Total_1=user_1+nice_1+sys_1+idel_1; Total_2=user_2+nice_2+sys_1+idel_2; IntCpuRate=（int（（（float（（user_2+sys_2+nice_2）-（user_1+sys_1+nice_1））/（float（total_2-total_1））＊100）; where total_1 is the sum of parameters read in the first reading, and total_2 is the sum of parameters read in the second reading. For other parameters, such as the size of the buffer, they can be obtained by calling the ioct1() system function of the Linux operating system. After obtaining the corresponding parameters, the corresponding parameters are mapped to the load location algorithm. The main algorithm used in the parameter conversion process is the dynamic threshold algorithm[5]. The mathematical model of the algorithm is as follows: In the above formula, B is the capacity of the system interface buffer space, Qi(t) is the queue length of the i-th Web server, Q(t) is the current total queue length of the system, T(t) is the threshold parameter for no longer connecting to the session, and α is the adjustment factor. As can be seen from the above formula, the dynamic threshold algorithm dynamically adjusts the control threshold according to the system state. The size of the threshold is proportional to the idle resources in the current system. When the buffer space of a Web server exceeds the threshold, the Web server will be blocked and no more load will be allocated to the Web server. 3.4.2 Load positioning strategy Based on the parameters obtained above, these parameters are passed to the scheduling server. The scheduling server selects the appropriate server by using the weighted round-robin scheduling algorithm [5], that is, the load is positioned on the specific Web server. The algorithm for the localization strategy is described as follows: Assume there is a set of web servers represented as S = {S0, S1, …, Sn-1}, W(Si) represents the weight of web server Si, variable i represents the web server selected in the last iteration, variable cw represents the weight currently being scheduled, max(S) represents the maximum weight of all web servers in set S, and gcd(S) represents the greatest common divisor of the weights of all web servers in set S. Variables i and cw are initially initialized to zero. The algorithm is described as follows: while (true) { if (i == 0) { cw = cw - gcd(S); if (cw <= 0) { cw = max(S); if (cw == 0) return NULL; } } else i = (i + 1) mod n; if (W(Si) >= cw) return Si; } 4. Experimental Environment and Results The experimental topology adopts the simplest system shown in Figure 1. The embedded network intelligent terminal uses the uClinux operating system, and corresponding network communication programs are added to the uClinux operating system. The scheduling server uses Linux 9.0 as the operating system and implements the scheduling algorithm in Section 3.3. Web server groups and Internet hosts are common forms. Figure 5 shows the data packets obtained using the Ethereal network protocol analyzer, which demonstrates the effectiveness of the system designed and the algorithm used in this paper. Figure 5. Acquired Data Packet 5. Conclusion The experiments show that the system structure, load calculation method, and scheduling algorithm proposed in this paper realize the remote control of the embedded network intelligent control terminal. By employing a weighted round-robin scheduling algorithm and a dynamic threshold algorithm, the system load scheduling problem is solved, resulting in good network performance and high practical application value. The implementation of this system provides a good network platform for further research on embedded network remote control system services. References [1] Li Jing, Zhang Weixin. An implementation method of real-time tasks under uClinux[J]. Microcomputer Information, 2005(1): 129-130 [2] Wang Dongbin, Fang Binxing, Yun Xiaochun. Design and implementation of network monitoring technology based on Web management[J]. Computer Engineering, 2002(10), 2003-2004 [3] J. Gettys J. Mogul Hypertext Transfer Protocol —— HTTP/1.1[S] RFC2068 [4] M. Tim Jones Embedded System TCP/IP Application Layer Protocol[M]. Beijing: Electronic Industry Press, 2003. 91-96. [5] Hahme EL, Choudhury A K. Dynamic queue length threshold for multiple loss priorities. IEEE/ACM Trans Networking[J], 2002, 10(3): 368-380