KR102073684B1 - Method for transmitting and receiving data for data distribution service - Google Patents

Abstract

Disclosed is a method for transceiving data for a data distribution service which can improve transceiving performance. The method for transceiving data for a data distribution service preferably comprises: a step in which a data analysis unit analyzes properties of data to be transceived and classifies data in accordance with analyzed data properties; a step in which a high-frequency data transceiving unit applies a batching profile to transceive user data with a small data size and a high transceiving frequency among the classified data; a step in which a large-capacity data transceiving unit applies an application level fragment profile to transceive user data with a large data size and a high priority among the classified data; a step in which a background large-capacity data transceiving unit applies a background large-capacity data profile to transceive user data with a large data size and a low priority among the classified data; and a step in which a general data transceiving unit applies a general profile to transceive user data with a large data size and a low transceiving frequency among the classified data.

Description

Data transmission and reception method for data distribution service {METHOD FOR TRANSMITTING AND RECEIVING DATA FOR DATA DISTRIBUTION SERVICE}

The present invention relates to a data transmission / reception method for a data distribution service, and more particularly, data for a data distribution service for automatically classifying user data according to characteristics and transmitting and receiving the automatically classified user data according to each characteristic. It relates to a transmission and reception method.

Many heterogeneous devices in a variety of industries use distributed networks to not only constantly distribute and share data in real time, but also to facilitate the dynamic involvement and withdrawal of participants over time. It requires data-centric communication technology.

There are many technologies for data-centric communication. Among them, publish / subscribe technologies are widely used in various application fields, and a representative protocol using them is a data distribution service (DDS).

DDS is a middleware established by the OMG (Object Management Group) organization. It maximizes the flexibility of node configuration by allowing free participation or withdrawal from the network data domain in a distributed environment, and sharing data in real time based on publishing / subscribing technology. And the ability to distribute.

These DDSs are used in many fields such as defense, aviation, aerospace, transportation, robotics, securities, medicine, financial transactions, and automation systems in various fields.

When transmitting and receiving user data using DDS, topic size and DDS Quality of Service (QoS) Policy tuning are representative factors that affect performance.

A topic is a structure of user data that you want to receive or send with a unique name in the DDS domain. You can send / receive one user data at a time by defining a topic with the same size as the user data size in consideration of system requirements or network conditions, and send and receive a single user data by dividing it several times by defining a small sized topic. have.

The DDS QoS policy is a set of parameters for data flow on the network and is a set of parameters for real-time operation control. There are about 20 types of standard DDS QoS policies, and one DDS QoS policy has several to tens of attribute values.

DDS QoS policy tuning is for system developers to set attribute values of DDS QoS policy according to system requirements such as operation control, communication quality, and transmission timing.

When setting DDS QoS policy attribute values needed to satisfy system requirements during DDS QoS policy tuning process, it is necessary to set them so that they do not conflict with the characteristics of the same or different DDS QoS policy attributes. Is needed.

If DDS QoS policy attribute values of transmitting / receiving nodes or data conflict with each other during DDS QoS policy tuning, a QoS incompatible error may occur, and data transmission and reception may not be possible, rather than a performance level of a system. In addition, finding the cause of QoS non-error among many DDS QoS policy attribute values is not an easy problem because it requires a lot of time and know-how.

If a QoS nonconformance error occurs, review all the tuned DDS QoS policy attributes or purchase a separate debugging tool to reduce the scope of the cause of the QoS nonconformance error before debugging.

As described above, since the conventional DDS QoS policy tuning has difficulty, DDS middleware manufacturers have to charge an additional consulting fee for DDS QoS policy tuning according to system requirements separately from the DDS middleware cost.

Consulting costs are different by each DDS middleware manufacturer and system size, but most of them are as expensive as DDS middleware purchasing cost, so many businesses do not request consulting and change the basic DDS QoS policy of sample provided little by little.

Korea Patent Publication No. 10-1236729 (Notice date 2013.02.28.)

SUMMARY OF THE INVENTION The present invention has been made to solve this problem, and provides a data transmission / reception method for automatically classifying user data according to characteristics and transmitting / receiving automatically classified user data according to DDS QoS policies for each characteristic. There is a purpose.

Data transmission and reception method for a data distribution service according to an embodiment of the present invention for achieving the above object, the data analysis unit analyzes the characteristics of the data to be transmitted and received, and classifies the data according to the characteristics of the analyzed data step; Transmitting, by a high frequency data transmission / reception unit, user data having a small data size and high transmission / reception frequency among the classified data by applying a batching profile; Transmitting / receiving, by a large data transmission / reception unit, user data having a high data size and high priority among the classified data by applying an application level fragment profile; A background mass data transmitting / receiving unit, transmitting and receiving user data having a large data size but low priority among the classified data by applying a background mass data profile; And transmitting and receiving, by the general data transmitting and receiving unit, user data having a small data size and low transmission / reception frequency among the classified data by applying the general profile.

In an embodiment of the present invention, the transmitting and receiving of the user data having a large data size and high priority may be performed by dividing the data to be transmitted into a plurality of data at an application level.

In another embodiment of the present invention, the step of transmitting and receiving user data with a small data size and a high frequency of transmission and reception includes: a coalescing packet structure in which a pair of a plurality of DDS RTPS headers and user data information are gathered in one IP packet. By using, it is preferable that the step of transmitting and receiving user data having a small data size and a high frequency of transmission and reception.

In the data transmission / reception method for the data distribution service of the present invention, DDS middleware can be easily applied to a system by automatically classifying user data according to characteristics and transmitting / receiving the automatically classified user data according to DDS QoS policies for each characteristic. Will be.

By applying the DDS QoS policy most suitable for each user data characteristic, the transmission and reception can be improved.

DDS QoS policies are applied according to the characteristics of user data, so even if the DDS QoS policy is tuned, only the profile of the corresponding DDS QoS policy can be applied to the system without changing the application.

1 is a view schematically showing the configuration of a data transmission and reception apparatus for a data distribution service according to an embodiment of the present invention.
2 is a diagram illustrating a packet structure used for high frequency data transmission and reception according to the present invention.
3 is a diagram illustrating a batch DDS packet structure applied to the present invention by way of example.
FIG. 4 is a diagram exemplarily illustrating a difference between a case where a large amount of data is divided into small sized topics and transmitted and received using one large topic.
5 is a diagram schematically illustrating an operation of a synchronous DataWriter, an asynchronous DataWriter, and a flow controller QoS policy.
6 is a flowchart illustrating a data transmission / reception method for a data distribution service according to an embodiment of the present invention.
7 is a view schematically showing a test operation process for high frequency data transmission and reception according to the present invention.
8 and 9 illustrate exemplary test results for high frequency data transmission and reception according to the present invention.
10 is a diagram illustrating a test result for large data transmission and reception according to the present invention by way of example.

DETAILED DESCRIPTION OF THE INVENTION The following detailed description of the invention refers to the accompanying drawings that show, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It is to be understood that the various embodiments of the invention are different but need not be mutually exclusive. For example, certain shapes, structures, and characteristics described herein may be embodied in other embodiments without departing from the spirit and scope of the invention in connection with one embodiment. In addition, it is to be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the invention. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present invention is defined only by the appended claims, along with the full scope of equivalents to which such claims are entitled. Like reference numerals in the drawings refer to the same or similar functions throughout the several aspects.

Hereinafter, an apparatus and method for transmitting and receiving data for a data distribution service according to an exemplary embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a view schematically showing the configuration of a data transmission and reception apparatus for a data distribution service according to an embodiment of the present invention.

As shown in FIG. 1, a data transmission / reception apparatus for a data distribution service according to an embodiment of the present invention may include a data analyzer 110, a high frequency data transceiver 120, a large data transceiver 130, and a background. Mass data transmission and reception unit 140, the general data transmission and reception unit 150 may be made.

The data analyzer 110 classifies the type of user data to be transmitted by analyzing characteristics (eg, size, transmission / reception period, transmission / reception frequency, etc.) of user data to be transmitted and received using the DDS middleware.

The user data types classified according to the characteristics of user data are small and high frequency data with high frequency of transmission and reception, large data with large data size, and large data size but low priority. It can be classified into low-priority large data (background large-capacity data) that should be done, and general data having a small data size and low frequency of transmission and reception.

The high frequency data (e.g., user location information, target information, etc.) having a small data size and high transmission / reception rate transmits and receives user data having a size of 1 KB or less at a frequency of several thousand times per second.

Since such high frequency data transmits and receives a large number of data in a short time, the following DDS QoS policy attributes should be basically considered.

RESOURCE_LIMIT - 고빈도 데이터를 송수신하므로 버퍼 크기는 한 번에 전송하는 데이터 개수 이상 설정

RESOURCE_LIMIT-sends and receives high frequency data, so the buffer size is set to at least the number of data sent at a time

HEARTBEAT - 송신측의 버퍼를 빨리 비워야 하므로 주기를 빠르게 설정

HEARTBEAT-Set the cycle faster because the sender's buffer should be emptied quickly

MAX_HEARTBEAT_RETRIES - 그룹에서 탈퇴한 수신단을 빠르게 비활성화(inactive)로 처리하기 위하여 데이터 전송 주기와 HEARTBEAT 주기 및 RESOURCE_LIMIT 크기를 고려하여 설정

MAX_HEARTBEAT_RETRIES-Set considering the data transmission cycle, HEARTBEAT cycle, and RESOURCE_LIMIT size in order to quickly deactivate a receiver that has left the group.

RELIABILITY - 위치 정보와 같이 주기적으로 송신 되고, 최신 데이터 값이 유효한 경우, BEST_EFFORT로 설정 (또는 RELIABILITY가 RELIABLE인 경우, HISTORY 속성을 KEEP_LAST로 설정)

RELIABILITY-periodically sent with location information, set to BEST_EFFORT if the latest data value is valid (or set HISTORY attribute to KEEP_LAST if RELIABILITY is RELIABLE)

When transmitting and receiving high frequency user data using DDS, a packet structure that can be used can be classified as shown in FIG.

The structure (a) of FIG. 2 is a basic packet structure of DDS for data transmission, in which one DDS RTPS header and one user data are included in one IP packet.

In this case, for example, when 1,000 user data is transmitted, 1,000 IP headers, a DDS RTPS header, and a submessage of RTPS, which is user data, are transmitted and received. When a problem occurs, the unit for retransmission also becomes an entire packet including an IP header, a DDS RTPS header, and a submessage of RTPS.

The structure of FIG. 2B is a method in which a user bundles a plurality of user data at an application level and defines and uses one new user data.

In this case, since the DDS RTPS and IP header information to be transmitted / received and analyzed are less than the structure of FIG. can do.

However, it is not new user data but first user data that is defined, but it is new user data that is wrapped. Therefore, it is difficult to use debugging tools provided by DDS development company, and it is impossible to use keys provided by DDS. have. In addition, additional logic is needed to add new user data and divide the collected data into more complex implementations.

Finally, the structure (c) of FIG. 2 is a coalescing structure in which a pair of multiple DDS RTPS headers and user data information is collected in one IP packet.

If only the related DDS QoS policy value is set on the transmitting side, transmission and reception are possible in an integrated structure. This method also reduces the number of IP headers, eliminates the need for users to add additional wrapping-related processing, and allows key and debugging tools to be used in many ways. The unified structure is mainly used when a large number of data retransmissions are needed at once for the NACK response.

Meanwhile, in order to transmit a large amount of data over a certain size, tuning of related DDS QoS policies is required. In particular, large data (e.g., audio, video, database, etc.) that transmits and receives user data once several megabytes per second is mostly data that requires reliable transmission such as audio, video, and database. The same DDS QoS policy attributes should be considered.

RESOURCE_LIMIT - 토픽 크기가 큰 경우, 버퍼 크기를 크게 설정하면 응용 프로그램이 사용하는 메모리 또한 비례하여 증가. 노드의 가용 메모리를 고려하여 설정 (메모리가 작은 임베디드 시스템은 특히 주의해서 설정)

RESOURCE_LIMIT-If the topic size is large, setting a large buffer size also increases the memory used by the application. Setup taking into account the available memory of the node (special care should be taken for embedded systems with low memory)

RELIABILITY - 사용자 데이터를 손실 없이 송수신해야 하므로 KEEP_ALL로 설정 (RESOURCE_LIMIT로 설정한 버퍼에 가득 찬 경우, 전송 측에서는 버퍼가 비워질 때까지 설정된 시간만큼 대기하게 되므로 일시적 또는 영구적으로 통신이 단절된 것처럼 동작)

RELIABILITY-User data should be transmitted and received without loss. Set to KEEP_ALL (If the buffer set by RESOURCE_LIMIT is full, the sender waits for the set time until the buffer is empty, so it acts as if communication was temporarily or permanently disconnected.)

MAX_MESSAGE_SIZE, RECV_SOCKET_BUFFER, MAX_FRAGMENTS_PER_SAMPLE - 토픽 및 사용자 데이터의 크기, 전송 프로토콜의 종류를 고려하여 설정

MAX_MESSAGE_SIZE, RECV_SOCKET_BUFFER, MAX_FRAGMENTS_PER_SAMPLE-set considering the size of the topic and user data, and the type of transport protocol

As described above, when the data analysis unit 110 classifies the type of user data to be transmitted by analyzing characteristics (eg, size, transmission / reception period, transmission / reception frequency, etc.) of user data to be transmitted / received, the classified user According to the data type, high frequency data is applied to the high frequency data transmission / reception unit 120, general data is applied to the general data transmission / reception unit 150, and large data is applied to the large-capacity data transmission / reception unit 130, and priority. Low volume data is applied to the low-capacity large data transmission / reception unit 130.

The high frequency data transceiver 120 may transmit and receive high frequency data classified by the data analyzer 110 using an integration method (see FIG. 2C).

Specifically, in the case of high frequency data that transmits and receives small sized user data at a frequency of several thousand times per second, transmission and reception are performed using a coalescing method to reduce the number of transmission and reception of a common part such as a message header. In this case, the high frequency data transceiver 120 preferably sets the HEARTBEAT cycle quickly after sufficiently setting the buffer size.

The high frequency data transceiver 120 may transmit / receive high frequency data classified by the data analyzer 110 according to a batch QoS policy.

The batch QoS policy is a method of further improving the integrated structure (structure (C) of FIG. 2) described in the high frequency data transmission / reception method, and is a DDS QoS policy capable of batching with various settings. The batching DDS packet structure may be as shown in FIG. 3.

Using the batching method, as shown in FIG. 3, one IP packet is transmitted in a structure in which one DDS RTPS header and several user data are collected. By default, DDS transfers user data immediately after the user calls the write () API. However, if a batch QoS policy is set, a repository called a batch is created inside the DDS stack, the user data is collected in a batch, and then the user data collected when the transmission conditions are satisfied is transmitted at once as a sub message. There are three transfer conditions: total data size accumulated in the batch, the number of user data, and a flush cycle. You can adjust the size of the batch, the number of data collected in one batch, and the flush cycle values to suit user data size, network performance, and the state of computer resources such as CPU and memory. Can be used

The high frequency data transmission / reception unit 120 performs transmission and reception on high frequency data using a batching profile for transmitting and receiving user data having a small data size and a high frequency of transmission and reception.

The batching profile may be generated based on a batch QoS policy used for high frequency data transmission and reception.

The large-capacity data transmission / reception unit 130 may divide the large-scale data classified by the data analysis unit 110 in consideration of network performance and divide user data at an application level, and then transmit and receive the data according to a batch QoS policy, which is a high frequency data transmission / reception method. .

For example, in order to transmit 5MB of user data, 5MB of data may be sent at the application level at one time, but at the application level, the data may be divided into small data in advance, divided into several times, and then sent and received again. You can use the 5MB size transfer method.

In the case of large data transmission and reception, performance can be adjusted in accordance with the topic size to be transmitted and received, so it is very important to determine the appropriate topic size according to the node and network performance and status.

4 exemplarily illustrates a difference between transmitting and receiving a large amount of data into smaller topics and transmitting and receiving using one large topic.

In case of transmitting / receiving using a small-sized topic as shown in FIG. 4 (a), at the application level, one large data is divided into several small user data and transmitted, and the received small user data is collected into one large data. We need additional logic. In addition, compared to the method of transmitting and receiving using the large topic of FIG. 4 (b), additional costs such as time delay, CPU, and memory resource usage are incurred in dividing into smaller user data. However, due to the small size of the data to be transmitted and received, the DDS packet can be quickly recovered, and the transmission timing can be adjusted by adding a delay in the transmission.

In case of (b) of FIG. 4, the cost of additional work at the application level is not incurred, but the cost incurred in recovering lost DDS packets is large because the user data is transmitted and received at a time. It is disadvantageous that other DDS packets cannot be processed for a long time for transmitting and receiving large data, rather than a method for transmitting and receiving using a small topic.

The above-described mass data transmission / reception unit 130 performs transmission / reception of large data by using an application level fragment profile for transmitting and receiving user data having a large data size.

In this case, the application level fragment profile may be generated based on the DDS QoS policy used for transmitting and receiving large data.

Meanwhile, the background large-capacity data transceiver 140 transmits and receives low-priority large-capacity data (background large-capacity data) classified by the data analyzer 110 using an asynchronous DataWriter and a flow controller.

Basically, since DDS processes data in the order of transmission and reception, even when transmitting and receiving user data that does not require the requirement of real-time, it affects the exchange of user data that has high priority or real time.

In order to process high-priority or real-time user data, it is difficult for a user to calculate a proper timing and control a large amount of low-priority data in the background and transmit it slowly.

In this case, you can use asynchronous DataWriter and flow controller QoS policies.

Figure 5 shows the operation of the synchronous DataWriter, asynchronous DataWriter, and flow controller QoS policy.

The DataWriter of the DDS basically operates synchronously as in the method of FIG. When API (write ()) is called for user data transmission at the user level, the result is returned to the user level after completion of the transmission from the DDS to the network, and the next work continues.

The method (b) of FIG. 5 is a diagram illustrating a case where the DataWriter operates asynchronously. When the API (write ()) is called to transmit user data at the user level, the method writes the user data to the DDS internal queue and immediately. Show what is returned to the user level. Stored user data is handled by a separate thread inside DDS.

The flow controller is a DDS QoS policy that can regulate network traffic by using a separate thread in asynchronous operation. Tokens_added_per_period value has more impact on performance than other property values. This is because a large tokens_added_per_period value increases the number of user data transmitted at one time and a small number of data transmitted at a time.

Accordingly, setting tokens_added_per_period value is more important than other attribute values in consideration of requirements, node and network status.

The smaller tokens_added_per_period setting increases the total time required for sending / receiving, but the network utilization rate used for background large-capacity data transmission and reception decreases the effect on other user data transmission and reception.

The background large-capacity data transceiver 140 may transmit / receive data such that tokens_added_per_period is adjusted so as not to be affected by real-time and high-priority data transmission by adjusting setting values of relevant attributes according to network conditions.

The above-described background large data transmission / reception unit 140 has a large data size but a low priority, so that the background large data profile for the background large data is transmitted using a background large data profile for transmitting / receiving user data that needs to be transmitted / received by minimizing the effect on system performance. Perform the transmission and reception.

In this case, the background large data profile may be generated based on a DDS QoS policy used for transmitting and receiving a large amount of low priority data.

The general data transceiver 150 may transmit and receive general data classified by the data analyzer 110 according to a basic DDS QoS policy.

The general data transmission / reception unit 150 performs transmission and reception on general data using a general profile, which is a profile for transmitting and receiving general data having no special characteristic.

Here, the general profile may be generated based on a basic DDS QoS policy used for general data transmission and reception.

The above-mentioned batching profile for high frequency user data transmission and reception, application level fragment profile for high volume data transmission and reception, background high volume data profile for low priority high volume data transmission and reception, general profile for user data transmission and reception without special characteristics, etc. It is desirable that it is done.

Meanwhile, in the embodiment of the present invention, although the high frequency data transceiver 120, the mass data transceiver 130, the background mass data transceiver 140, and the general data transceiver 150 are independently implemented, One transceiver may be implemented, and one transceiver may be configured to transmit and receive user data classified according to characteristics by applying a profile according to each characteristic.

6 is a flowchart illustrating a data transmission / reception method for a data distribution service according to an embodiment of the present invention.

First, the data analysis unit 110 of the data transmission and reception apparatus 100 analyzes characteristics (eg, size, transmission / reception period, transmission / reception frequency, etc.) of user data to be transmitted / received (S10).

Thereafter, it is checked whether the user data to be transmitted or received is a large amount of data based on the analysis result of step S10 (S20).

If the user data to be transmitted / received is not a large amount of data as a result of the checking in step S20, it is checked whether the user data to be transmitted / received is data having a high transmission / reception frequency (S30).

When the user data to be transmitted and received as the result of the check in step S30 is data having a high frequency of transmission and reception, the high frequency data transmission / reception unit 120 may transmit and receive high frequency data according to an integrated method by applying a batching profile (S40). .

In step S40, the high frequency data transmission / reception unit 120 may use an integrated method to reduce the number of transmission / reception of common parts such as a message header, and may set a HEARTBEAT cycle quickly after sufficiently setting the buffer size.

In addition, in step S40, the high frequency data transceiver 120 may transmit / receive the high frequency data classified by the data analyzer 110 according to a batch QoS policy.

On the other hand, if the user data to be transmitted and received as the result of the check in step S30 is a user data that does not have a special characteristic that the transmission frequency is not high, the general data transmission and reception unit 150 is applied according to the basic DDS QoS policy by applying a general profile User data having no characteristic may be transmitted and received (S50).

On the other hand, if the user data to be transmitted and received as a result of the check in step S20 is a large amount of data to determine whether the high priority data (S60).

When the user data to be transmitted and received is the high priority data as a result of the checking in step S60, the mass data transmission / reception unit 130 may transmit and receive mass data by applying an application level fragment profile (S70).

In step S70, the large-capacity data transmission / reception unit 130 divides the large-capacity data into topics of an appropriate size in consideration of network performance, transmits the data according to a batch QoS policy, and reassembles the received topics at the receiving side. Restore to the signal.

On the other hand, if the user data to be transmitted and received as a result of the check in step S60 is not the high priority data, the background large data transmission / reception unit 140 may transmit and receive a large amount of low-priority data by applying the background large data profile Can be (S80).

In step S80, the background mass data transmission / reception unit 140 may transmit / receive the low-priority mass data using an asynchronous DataWriter and a flow controller.

In step S80, the background mass data transceiving unit 140 may set a token_added_per_period value to minimize the influence on other real-time user data transceiving in consideration of requirements, nodes, and network conditions.

Hereinafter, according to the present invention, the performance of high frequency data, large data, and low-priority background large data transmission / reception are respectively tested according to user data characteristics.

First, a test for high frequency data transmission and reception may be as shown in FIG. 7.

The sender transmits the user data of the desired size n times and stores the test start time (T1). The receiving side receives n times of data transmitted from the transmitting side, and then transmits "Echo Data" to the transmitting side. The sender stores the time (T2) of receiving “Echo Data”, and then uses the test start time (T1) and test end time (T2) values for the coalescing and batching methods. Change QoS policy values and test until optimal performance is achieved. Using the DDS QoS policy when the optimal performance is achieved, the delay time can be measured by repeating 20 times for each topic size and the number of transmission and reception.

Tables 1, 8 and 9 show the results of changing the size of the topics to 256 bytes, 512 bytes, 1 KB, and changing the number of transmission / reception of 3,000 times and 10,000 times for each integration method and batching method.

In all cases tested, regardless of topic size and number of transmissions and receptions, the batching method is the most stable and has the lowest latency, with a lower standard deviation than the integrated method.

The batching method can process multiple user data at once using one DDS RTPS header. Therefore, the overhead of meta-information processing is reduced at the receiving side, and resource usage such as CPU is reduced. In addition, since the ACK or NACK process is performed for one batch including a plurality of user data in reliable data transmission, throughput is more than other methods of ACK or NACK processing for one user data. It can be seen from the results of the high frequency performance test that much better performance of the

Meanwhile, for large data, the size of user data to be transmitted at a time is set to 5MB. The large data transmission / reception test program operates in the same way as the high frequency data transmission / reception test program, which is divided into small data in advance at the application level and transmitted and received at several times. Can be used.

Using this partitioning and reassembly method, we can compare and analyze the delay time measured when sending and receiving 5MB topic once and 5,120 topic sending and receiving 1KB size when testing 5MB user data. . The large data transmission / reception test measures the delay time for transmitting and receiving user data of 5MB size by repeating 20 times using topics of 512Bytes, 1KB, 256KB, 512KB, 1MB, and 5MB.

Table 2 and FIG. 10 are test results of measuring delay times according to respective topic sizes when transmitting and receiving user data of 5 MB using topics of 512 bytes, 1 KB, 256 KB, 512 KB, 1 MB, and 5 MB.

As the test results show, the average latency is 57.93ms for a topic size of 5MB, 51.93ms for a topic size of 1MB, 53.87ms for a topic size of 512KB, and a topic size of 256KB. Is 58.8ms for the topic, 155.73ms for the topic size of 1KB, and 217.2ms for the topic size of 512B. When transmitting 5MB of user data, it can be seen that the method of transmitting and receiving five times using a topic of 1MB size is the most stable, and the time at which the transmission and reception is completed is the fastest. Relatively small topic sizes and a large number of transmission / receptions have a lower performance in terms of speed than large topic sizes and a small number of transmissions and receptions. This is because the number of packets to be transmitted and received increases, so that the DDS header information to be processed and processing time to split and aggregate are increased. When sending and receiving large amounts of data of the same size, if the number of samples of a topic that is not divided at all or becomes large, it can be confirmed by the results of the large data test that it is not good in terms of time.

On the other hand, the background large data transmission and reception test program divides 100MB user data into small topics and measures the delay time when transmission and reception are completed. Using the DDS QoS policy when the optimal performance is achieved, 20 times of delay time and network utilization are measured and background large data transmission / reception test is performed.

Table 3 shows the test results of measuring latency and network utilization by changing the token_added_per_period attribute value to UNLIMITED, 30, 20, 10, 5, and 1 when 100MB of user data is transmitted and received using a 32KB topic.

With an asynchronous DataWriter and flow controller, when transferring user data, the amount of data that can be sent depends on the number of tokens available. Therefore, the token_added_per_period value, which is the number of tokens added periodically in the background large data transmission / reception test, is adjusted. Unlike other tests, the network utilization is measured along with the latency measurement.

If the number of tokens added periodically is set to UNLIMITED, the average time for transmitting and receiving 100MB of user data is 5,918.4ms, and the network utilization rate is 12%. If the number of tokens added periodically is set to 1, the time required for transmitting and receiving 100MB of user data is 116,765.3ms on average, and the network utilization rate is 0.4%. Reducing the number of tokens added periodically from UNLIMITED to 1 can reduce network utilization from 12% to 0.4%, but the time required to complete sending and receiving data varies by about 20 times.

As can be seen from the background large data transmission / reception test results, if the number of tokens added per cycle is reduced by using an asynchronous DataWriter and flow controller, the time required to complete transmission and reception is increased, but the data size and the number of data to be transmitted and received at once are processed. Decreases. Nodes that send and receive large amounts of data in the background reduce the amount of data that needs to be processed at one time and reduce network utilization for that node, thus reducing the impact on other important real-time data transmissions and processing.

Although described above with reference to the embodiments, those skilled in the art will understand that various modifications and changes can be made within the scope of the invention without departing from the spirit and scope of the invention as set forth in the claims below. Could be.

110. data analysis unit, 120. high frequency data transmission and reception unit,
130. large data transmission and reception unit 140. background large data transmission and reception unit,
150. General Data Transceiver

Claims (3)

Analyzing, by the data analyzing unit, characteristics of the data to be transmitted and received and classifying the data according to the analyzed characteristics of the data;
Transmitting, by a high frequency data transmission / reception unit, user data having a small data size and high transmission / reception frequency among the classified data by applying a batching profile;
Transmitting / receiving, by a large data transmission / reception unit, user data having a high data size and high priority among the classified data by applying an application level fragment profile;
A background mass data transmitting / receiving unit, transmitting and receiving user data having a large data size but low priority among the classified data by applying a background mass data profile; And
And transmitting and receiving, by the general data transmitting and receiving unit, user data having a small data size and a low transmission / reception frequency among the classified data by applying the general profile.
Transmitting / receiving user data having a small data size and high transmission / reception frequency,
A method of transmitting and receiving user data with a small data size and a high frequency of transmission and reception using a coalescing packet structure in which a pair of DDS RTPS headers and user data information are gathered in one IP packet,
Transmitting and receiving user data having a large data size and high priority,
In this step, data to be transmitted is divided into topics of a size determined according to the performance and status of nodes and networks at an application level.
Transmitting and receiving user data having a large data size but low priority,
Among the categorized data, user data having a large data size but low priority is transmitted and received using an asynchronous DataWriter and a flow controller.
A method of transmitting and receiving data for a data distribution service, comprising: setting a token_added_per_period value to minimize the influence on other real-time user data transmission and reception in consideration of requirements and node and network conditions. delete delete

Images