CN103685309A - Asynchronous request queue model oriented to map visualization tile service - Google Patents

Abstract

The invention provides an asynchronous request queue model for map visualization tile service access, and belongs to the technical fields of computer network technology and queue control. This model mainly applies the principle and method of queuing theory to solve the access problem of network map visualization tile service, and realize hierarchical, real-time and smooth map tile access by clients. First, set up a fixed-capacity request queue. When a new request arrives, determine whether the request is placed in the queue or discarded according to the queue status and priority; Get the process and return the result as a callback. This model involves technologies such as priority control under high-frequency visualization requests, concurrent request volume control, network asynchronous requests, and request queue management. It is a high-frequency, high-time-effective network asynchronous request queue model with priority control. It is applicable Access to map visualization tile network services.

Description

Asynchronous request queue model for map visualization tile service access

technical field

The present invention relates to computer network technology and queue control technology, in particular to priority control under high-frequency visualization requests, concurrent request volume control, network asynchronous request, request queue management and other technologies. The invention is a high-frequency and high-time-effective network asynchronous request queue model with priority control, which is suitable for map visualization tile network service access.

Background technique

Map visualization tile client requests have the requirements of high request frequency, high timeliness, and strict priority conditions. As the user continues to switch views, tile requests frequently occur in the background; the tiles in the center of the current view are always requested with the highest priority; at the same time, the tiles on the periphery and adjacent levels of the view are requested with different priority weights. It includes asynchronous concurrent access mechanism based on HTTP network protocol, priority request queue model and other related technologies.

Hypertext transfer protocol (Hypertext transfer protocol, HTTP) is a distributed, collaborative, communication protocol between hypermedia system applications, and is the basis for exchanging information on the World Wide Web. The commonly used HTTP request types are GET and POST, both of which are synchronous access methods, and the request is in a blocking state; the asynchronous implementation method is more complicated and difficult to program, but the request overhead is small, and it is suitable for application environments with frequent access. The main references include http://www.w3.org/Protocols/; http://progrium.com/blog/2012/12/06/async-http-responses-with-response-redirection/ , etc.

Queuing theory is an important branch of operations research and applied probability. In real life, the application examples of queuing theory can be seen everywhere and are playing an important role. It mainly consists of three parts: input process, queuing discipline and service system. The input process describes the rules by which customers arrive in the service system; the queuing rules describe whether the service is allowed to queue and whether the customers are allowed to queue; the service system mainly describes the number and organization of service desks, and the distribution of time required by customers. In the queuing system, customers and service agencies are most concerned about the length of the queue, the waiting time of customers in the system, their stay time and distribution, the busy period of the service desk, etc. These indicators are also the most important characteristics of the queuing system. The main references include QingchunJiang, Sharma Chakravarthy. Queueing analysis of relational operators for continuous data streams, Technical Report CSE, 2003(02).; Yingxian Liang. Application of mathematical logic in queue theory, Journal of Shenyang Institute, 2Engerona, etc.

The asynchronous request queue model for map visualization tile service access combines the asynchronous network request technology and the queuing method of priority control, which specifically involves technical issues such as map tile priority control, concurrent request volume control, and request queue management. , so far, it can be seen that there are no more systematic solutions in the literature/patents.

Contents of the invention

The purpose of the present invention is to establish an asynchronous network access request queue model that supports high-frequency access, has priority control and network access control, and has high timeliness for the access problem of networked map visualization tile services. Realize hierarchical, real-time, and smooth access to map tiles by the client.

The basic idea of the present invention is: set a fixed-capacity request queue, and each item in the queue can accommodate the information of a tile request; Enqueue or discard; on the other hand, create threads corresponding to the number of concurrent requests, take out the highest priority request from the queue to execute the tile acquisition process, and return the result in the form of a callback. This procedure instantiates an asynchronous request queue model with an input procedure, queuing discipline, and service system.

The technical solution of the present invention provides an asynchronous request queue model with priority and flow control, specifically including the following implementation steps:

1) The user defines an initial value of the number of connections, indicating the number of requests sent to the server at the same time;

2) Define a tile request queue with a fixed length (number of connections × 2), each item in the queue stores tile request information, including tile number, priority, timestamp, request status and return data pointer ;

3) Define a callback function for returning tiles asynchronously, and the callback information includes the tile number, timestamp and tile data corresponding to the tile request queue;

4) Generate and update the list of tiles to be requested in real time according to the viewpoint of the user browsing the map, and define the priority according to the view. The basic priority rules from high to low are: center of viewpoint, edge of viewpoint, adjacent lower layer, adjacent upper layer;

5) Start the queuing request thread to cyclically read each item in the tile pending request list, call the queue request function to request tiles, and stop the thread until the queue is empty;

6) When a tile request occurs, first determine whether the tile request queue is full, and if not, add the request to the queue; otherwise, if the priority of the current request is higher than the lowest priority item in the queue, replace it; when the above When the two conditions are not met, the current request is discarded;

7) Open multiple network request threads, the number of threads is consistent with the number of connections defined by the user, each thread performs the following operations: take out the request with the highest priority from the tile request queue, and parse it to form a URL service access address, Request tiles from the visualization service, and after the tiles are returned, remove the corresponding request from the tile request queue and tile pending request list, and return the result through the callback function.

The above steps are characterized by:

步骤2)和步骤3)定义了瓦片请求队列和回调函数的内容，确定了异步访问的形式；同时定义了瓦片请求队列的长度，确定了瓦片流量控制的形式；

Step 2) and step 3) define the content of the tile request queue and the callback function, and determine the form of asynchronous access; at the same time, define the length of the tile request queue, and determine the form of tile flow control;

Step 4) and step 5) are external input processes, generating tile requests with priority according to the user's viewpoint, and the background thread sends tile access requests to the queue model cyclically at fast time intervals;

步骤4)和步骤5)中叙述的瓦片待请求列表和瓦片请求队列形成了一个双队列模式，前者维护请求发起方的队列缓存，由用户程序生成，内容符合步骤2)中的定义；后者维护处理请求的队列缓存，是优先级和流量控制的结构基础；

The tile pending request list and tile request queue described in step 4) and step 5) form a double queue mode, the former maintains the queue cache of the request initiator, which is generated by the user program, and the content conforms to the definition in step 2); The latter maintains a queue cache for processing requests, which is the structural basis for priority and flow control;

Step 6) The core mechanism of priority in the queue model is realized, and in the process of continuously sending requests from external threads, it is ensured that the request entering the queue is the request with the highest priority;

步骤7)通过多线程实现了瓦片请求的流量控制、优先级控制和异步特性，具体来讲，线程数与用户定义的连接数一致，控制了流量；按照优先级顺序处理请求，实现了优先级控制；通过回调函数返回结果，实现了异步化。

Step 7) Through multi-threading, the flow control, priority control and asynchronous characteristics of the tile request are realized. Specifically, the number of threads is consistent with the number of connections defined by the user, and the flow is controlled; the requests are processed according to the order of priority, which realizes Priority control; the result is returned through the callback function, which realizes asynchrony.

Compared with the prior art, the present invention has the following characteristics: (1) it supports high-frequency real-time tile access, and external user programs can send requests continuously and in batches to obtain real-time response effects; (2) the priority control and The combination of flow control ensures fast, timely and accurate response to tile requests while effectively reducing the load of the visualized tile service.

Description of drawings

Fig. 1 general flow of the present invention

Figure 2 Schematic diagram of priority generation based on two-dimensional view

Figure 3 Schematic diagram of priority generation based on 3D view

Figure 4 Multi-threaded access to the tile request queue

Figure 5 two-dimensional display renderings

Detailed ways

The present invention requires the user to input the initial number of connections, which is determined by indicators such as network conditions, and the default value is 6. The following implementation steps will be described using this value as an example.

Fig. 1 illustrates the overall process of the present invention, which is composed of three parts as a whole: (1) input process: the external program generates a tile request with priority according to the view display state, and continuously sends the request to the queue model module; ( 2) Queuing rules: Based on the principle of priority and flow control, the queue model will concurrently respond to the top 6 requests with the highest priority; (3) Service system: The queue model accesses the tile service and returns the tile result, and sends the corresponding request from It is removed from the queue and handed over to an external program for processing through a callback function.

The structure of a tile request defined by the present invention includes the following parts:

瓦片编号：由用户自定义，用于在异步调用机制中唯一标识一个瓦片，本发明采用[X，Y，D]表示，将全球范围划分为平面网格(瓦片)，X、Y、D分别表示横向编号、纵向编号和深度编号，这是一种常见的瓦片划分方法，本发明不限于此种划分方法，在此不赘述；

Tile number: defined by the user, used to uniquely identify a tile in the asynchronous call mechanism, the present invention uses [X, Y, D] to represent, and divides the global scope into flat grids (tiles), X, Y , D represent horizontal numbering, vertical numbering and depth numbering respectively, this is a kind of common tile division method, the present invention is not limited to this kind of division method, do not go into details here;

优先级变量：用一个整形变量代表瓦片请求的优先级，在队列模型内部处理时，按照变量值越大优先级越高的原则排序，优先处理变量值大的请求，具体数值的确定由用户自定义，后续将以二维和三维视图为例进一步说明；

Priority variable: An integer variable is used to represent the priority of the tile request. When processing inside the queue model, it is sorted according to the principle that the larger the variable value is, the higher the priority is. Priority is given to processing requests with large variable values. The specific value is determined by the user Customization, which will be further explained by taking 2D and 3D views as examples;

时间戳：表示瓦片请求进入队列模型时的时间，精确到毫秒，用于判断瓦片请求是否超时，若超时则认为目标服务器没有该瓦片数据，将请求从队列中移除；

Timestamp: Indicates the time when the tile request enters the queue model, accurate to milliseconds, and is used to determine whether the tile request times out. If it times out, it is considered that the target server does not have the tile data, and the request is removed from the queue;

Request status: tile requests include three states: pending request, requesting, and request completed, respectively representing the request entering the queue model, accessing tile service, and tile service returning results. What needs to be explained here is that when the request status is in progress When a request is made, the tile request will be locked and will not be replaced by other requests until abnormal conditions such as result return or service timeout occur.

The external user program generates tile requests according to the above structure definition according to different view characteristics. Figure 2 illustrates a tile request priority generation rule for a two-dimensional view. The bold solid line box represents the window boundary of the two-dimensional view, and the dotted line box represents the boundary of the view cache. The squares are tiles. The number in represents the priority, and the priority descends spirally from the center of the viewpoint to the surroundings. FIG. 3 illustrates a tile request priority generation rule for a three-dimensional view. According to the order of tiles involved in the viewing angle from near to far, tile requests with priority from high to low are correspondingly generated.

Figure 4 illustrates the multi-threaded working mechanism of the queue model, in which the execution path of the external queuing request thread is indicated by the blue arrow line, and the execution path of the queue model tile request thread is indicated by the red arrow line; above the "request interface" is the external user Program, the following "request interface" is the queue model program.

The following is an example of a processing solution for an external program. Whenever the view changes and a new tile requirement is generated, the external user program generates a tile request to form a tile pending request list, and then starts the queuing request thread to continuously and cyclically send tile requests to the queue model, and the thread works The pseudocode of the process (the blue line above the "request interface" in Figure 4) is as follows:

At the same time, the external program implements the callback function, which is called by the queue model program. The function prototype is as follows:

CallbackTile(tile number, tile data)

The callback function first copies the tile data and submits it to the upper-level program for processing, and then removes the corresponding tile request from the tile pending request list. In order to avoid multi-thread access conflicts, the list needs to be locked when removing.

The main part of the queue model consists of two parts: the request interface (the blue line below the "request interface" in Figure 4) and the tile request thread (the red line below the "request interface" in Figure 4). The former is expressed in pseudocode as follows:

The request interface realizes the queuing rule when the request arrives in the queue model, and the tile request thread realizes the working process of the service system. As described in the above pseudo code, the thread starts when the request arrives, and the maximum number of threads is 6. The execution process Expressed in pseudocode as follows:

The above steps implement a queue model with a maximum concurrent request of 6 tiles, and use queue rules to avoid repeated requests and enable high-priority tiles to get real-time responses. The tile request queue is set to twice the number of connections, that is, 12. On the one hand, it ensures the real-time nature of tile requests and acts as a time priority rule. On the other hand, it avoids excessively long queues and consumes resources.

The display effect of the two-dimensional view realized by the present invention is shown in FIG. 5 . The size of the view is 1091 pixels in width and 636 pixels in height, and the size of each tile is 256 pixels in width and 256 pixels in height. The number of tiles required in the picture is 20, and the priority order of the request is spirally decreasing from the center of the view to the surrounding. The actual display order is shown in the red Arabic numerals in the picture. When the average response time of a single tile is 0.5 seconds, it takes less than 3 seconds to display all the views. The invention is suitable for high-frequency and fine-grained network access under priority control, and can realize hierarchical, real-time and smooth access to networked map visualization tile services.

Claims (3)

1. An asynchronous request queue model for map visualization tile service access, characterized in that it includes the following implementation steps: Step 1, the user defines an initial value of the number of connections, indicating the number of requests sent to the server at the same time; Step 2. Define a tile request queue whose length is twice the number of connections. Each item in the queue stores tile request information, including tile number, priority, timestamp, request status, and returned data pointer. ; Step 3, define a callback function for asynchronously returning tiles, the callback information includes the tile number, timestamp and tile data corresponding to the tile request queue; Step 4: Generate and update the list of tiles to be requested in real time according to the viewpoint of the user browsing the map, define the priority according to the view, and the basic priority rules from high to low are: center of viewpoint, edge of viewpoint, adjacent lower layer, adjacent upper layer ; Step 5, start the queuing request thread to cyclically read each item in the tile pending request list, call the queue request function to request tiles, and stop the thread until the queue is empty; Step 6, when a tile request occurs, first judge whether the tile request queue is full, if not, add the request to the queue; otherwise, if the priority of the current request is higher than the lowest priority item in the queue, replace it; when When the above two conditions are not met, the current request is discarded; Step 7: Start the network request threads whose number is consistent with the number of connections defined by the user. Each thread performs the following operations: take out the request with the highest priority from the tile request queue, parse it to form a URL service access address, and send it to the visualization The service requests a tile, and after the tile data is returned, the corresponding request is removed from the tile request queue and tile pending request list, and the result is returned through the callback function. the 2. The asynchronous request queue model according to claim 1, characterized in that a synchronous HTTP network access protocol is used to realize the asynchronous tile access form, support high-frequency real-time tile access, and external user programs can be continuously , Send requests in batches to obtain real-time response effects. the 3. The asynchronous request queue model according to claim 1, characterized in that priority control and flow control are combined to ensure fast, timely and accurate response to tile requests while effectively reducing the load of the visualized tile service. the

Images