KR20170142513A - Apparatus for multinet aggregation transmission, and packet scheduling method thereof - Google Patents

Abstract

A packet scheduling method of a device for merging and transmitting multi-networks comprises the following steps of: collecting network state parameters with respect to a plurality of networks, connected to a receiving device, respectively; determining a control network among the plurality of networks based on the network state parameters of each of the plurality of networks; and adjusting a data amount per unit time to be transmitted to the control network based on at least one network state parameter with respect to the control network.

Description

TECHNICAL FIELD [0001] The present invention relates to a multi-network merging transmission apparatus and a packet scheduling method,

The present invention relates to multi-network merging transmission.

Aggregation transmission is a technique of transmitting data using a plurality of communication networks at the same time, and processes data transmitted through each path into one session. Through a merge transmission technique, a terminal can be connected to a plurality of communication networks at a time, and one service / application merges a plurality of networks as a single network, regardless of the number of networks or networks. Therefore, the merge transmission apparatus can quickly transmit and receive a large amount of data using a plurality of usable network resources. It can be called MultiNet Aggregation in the sense of merging a plurality of networks.

There are multi-path TCP (MPTCP) technologies that combine multiple TCP flows in a merge transmission technique. MPTCP is an L4 technology for simultaneously using multiple IP interfaces. A terminal having a plurality of physical interfaces can be connected to a plurality of communication networks at a single point through MPTCP technology, and creates a session on a subflow basis to perform end-to-end communication.

Conventional network devices use a flow control, a congestion control, and an error control method used in a single network to perform packet scheduling even when a packet is transmitted to multiple networks. That is, the conventional network apparatus makes an ideal assumption that the packet transmission delay time, the round trip time (RTT), and the packet loss rate are the same for each route generated in each of the multiple networks. However, multi-networks can be composed of heterogeneous networks with different characteristics, and even if they are composed of homogeneous networks, the network status is not always the same. That is, the packet transmission delay time, RTT, and packet loss rate vary with time depending on the wireless / wired section of the network. Therefore, if a plurality of paths having the same network state are used, the merge transmission efficiency is not lowered even if packet scheduling is performed by a round-robin method. However, if the round-robin method is used for multi- The efficiency is inevitably lowered.

SUMMARY OF THE INVENTION The present invention provides a multi-network merging transmission apparatus that dynamically packetizes packets based on states of multiple networks, and a packet scheduling method therefor.

A packet scheduling method of a multi-network merging transmission apparatus according to an exemplary embodiment of the present invention includes collecting network status parameters for each of a plurality of networks connected to a receiving apparatus, And adjusting the amount of data per unit time to be transmitted to the control network based on the at least one network status parameter for the control network.

The network status parameters may include at least two of a packet round-trip time, a flight status data size, a reception window size, a congestion window size, and a retransmission data size that have not yet received an acknowledgment from transmission data.

The step of adjusting the amount of data per unit time may adjust the amount of data per unit time when the size of the data transmitted through the control network and not received an acknowledgment is larger than the first reference value.

Wherein adjusting the amount of data per unit time comprises determining an amount of transmission data per unit time that is greater or less than a reference value based on at least one network state parameter for the control network, Size or the number of segments that can be transmitted in the available congestion window size.

The packet scheduling method may further include transmitting the determined amount of data to be transmitted from the data waiting to be transmitted in the original data buffer to the control network.

Wherein the packet scheduling method further comprises determining whether to change from a first scheduling method to a second scheduling method based on a scheduling change condition, wherein the step of determining the control network comprises: The first scheduling method may be a scheduling method used in a single network, and the second scheduling method may be a multi-network merge scheduling method.

The scheduling change condition may be a condition including a number of subflows generated in the plurality of networks and a size of a data stream transmitted to the subflows (Data Sequence Signal, DSS).

Wherein the scheduling change condition is a condition that when the transmission quality of any one of the plurality of networks is improved to a first reference value or more and the transmission quality of any one of the plurality of networks becomes worse than a second reference value, And a case where the difference in quality of at least two networks is equal to or greater than a third reference value.

The packet scheduling method may further include returning to the first scheduling method when the scheduling change condition is released after the first scheduling method is changed to the second scheduling method.

The packet scheduling method may further include a step of returning to the first scheduling method after the data transmission to the receiving terminal is completed after the first scheduling method is changed to the second scheduling method.

In a packet scheduling method of a multi-network merging transmission apparatus according to another embodiment of the present invention, there is provided a packet scheduling method of a multi-network merging transmission apparatus that requires segment size adjustment to be transmitted to a subflow based on a data size, Determining a total transmission segment size to be transmitted to the sub-flow based on at least one network state parameter related to the sub-flow when segment size adjustment is required, and transmitting the determined total transmission segment size data .

The step of determining whether the segment size adjustment is necessary may determine that segment size adjustment is necessary when the data size for which the acknowledgment is not received is larger than the first reference value.

Wherein determining the aggregate transmission segment size comprises comparing the network state parameters associated with the sub-flow to determine a total transmission segment size that is greater or less than a reference value, the reference value being an available congestion window size of the control network, The number of segments that can be transmitted in one congestion window size.

The network status parameters related to the sub-flow may include at least two parameters among data transmitted through the sub-flow, such as a data size, a reception window size, a congestion window size, and a retransmission data size which have not yet received an acknowledgment.

A packet scheduler operated by at least one processor according to another embodiment of the present invention, the packet scheduler comprising: a network status monitoring unit for collecting network status parameters of each of a plurality of networks connected to a receiving apparatus; A packet transmission rate adjuster for adjusting the transmission rate of each sub-flow generated in each of the plurality of networks, and dividing the original data to be transmitted to the receiver on the basis of the transmission amount of each sub-flow, And the data stored in the data buffer of each sub-flow is transmitted to the network through which the sub-flows are generated through the multi-path TCP (MPTCP) layer and the IP (Internet Protocol) layer.

The network status parameters include a round trip time of a packet transmitted to each network, a size of data not yet acknowledged among data transmitted to each network, a size of a reception window of each network, a size of a congestion window of each network , And retransmission data size in each network.

Wherein the packet transmission amount determination unit determines whether to start the transmission amount control for each subflow based on the network state parameter of each of the plurality of networks and, when starting the transmission amount control for each subflow, The amount can be adjusted up or down based on the reference value.

The packet transmission amount determination unit determines whether to start the transmission amount control for each subflow based on a determination condition composed of the number of subflows generated in the plurality of networks and the size of a data stream (Data Sequence Signal, DSS) transmitted to the subflows can do.

When the transmission quality of any one of the plurality of networks is improved to a first reference value or more and the transmission quality of any one of the plurality of networks is degraded to a second reference value or more, It is possible to determine whether to start the transmission amount control for each sub-flow based on a determination condition including at least one of the cases where the quality difference of at least two networks is equal to or greater than the third reference value.

The packet transmission amount determination unit may vary the number of segments transmitted or the total transmission segment size per unit time based on the network status parameter of each of the plurality of networks.

According to the embodiment of the present invention, the multi-network merging transmission apparatus manages the states of each of the multiple networks and performs integrated scheduling according to the states of the respective networks, so that the transmission efficiency of the high-quality network is not hindered by the state of the low-quality network. Therefore, according to the embodiment of the present invention, efficiency of multi-network merging transmission can be improved by efficiently using resources of each of the multiple networks.

FIG. 1 is a diagram illustrating a configuration of a multi-network merging system according to an embodiment of the present invention. Referring to FIG.
2 is a conceptual diagram illustrating multi-network merging transmission according to an embodiment of the present invention.
3 is a configuration diagram of a packet scheduler of a transmission apparatus according to an embodiment of the present invention.
4 is a conceptual illustration of a network stack in which a packet is transmitted in a transmitting apparatus according to an embodiment of the present invention.
FIG. 5 is a view for explaining a merging transmission scheduling method when the present invention is not applied.
6 is a diagram illustrating a merging transmission scheduling method according to an embodiment of the present invention.
7 is a flowchart of a packet scheduling method according to an embodiment of the present invention.
8 is a flowchart of a packet scheduling method according to another embodiment of the present invention.
9 is a flowchart of a packet scheduling method according to another embodiment of the present invention.
10 is a hardware block diagram of a multiplexing transmission apparatus according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise. Also, the terms " part, "" module," and " module ", etc. in the specification mean a unit for processing at least one function or operation and may be implemented by hardware or software or a combination of hardware and software have.

In the present invention, a transmitting apparatus and a receiving apparatus are connected by a multi-network, and when a transmitting apparatus sends data to a receiving apparatus, multi-network merging transmission is performed. Next, a description will be given of a scheduling method for determining how much data is to be transmitted to each sub-flow generated in each of the multiple networks when the transmitting apparatus, which is a multi-network merging transmission apparatus, transmits data to the receiving apparatus.

In the present invention, the transmitting apparatus or the receiving apparatus may be a user terminal. A user terminal includes a mobile station (MS), a mobile terminal (MT), a subscriber station (SS), a portable subscriber station (PSS), a user equipment An access terminal (AT), and the like, and may include all or some of functions of a mobile station, a mobile terminal, a subscriber station, a mobile subscriber station, a user equipment, an access terminal, and the like.

The user terminal includes a base station (BS), an access point (AP), a radio access station (RAS), a node B, an evolved NodeB (eNodeB) A base transceiver station (BTS), a mobile multihop relay (MMR) -BS, and the like.

The user terminal may be a mobile terminal such as a smart phone, a tablet terminal such as a smart pad and a tablet PC, a communication terminal such as a computer and a television, and may have a plurality of communication interfaces. The communication interface may vary. For example, the communication interface may include a short-range wireless network interface such as WiFi / WLAN / Bluetooth and a mobile communication network interface such as 3G / Long Term Evolution (LTE) / Long Term Evolution-Advanced And the terminal manufacturer may add various communication interfaces. In this specification, the 3G / LTE interface and the WiFi interface are described as an example, but the communication interface is not limited thereto.

In the present invention, the transmitting apparatus or the receiving apparatus may be a network apparatus. A network device may be a gateway or a server that merges sub-flows that are transmitted in a multi-path, or splits a flow that is transmitted in a single path into sub-flows of a multi-path and transmits the sub-flows. The network device is located at the contact point of the multi-network, for example, at the contact point of the LTE network and the WiFi network. The network device may be referred to as a MultiNet Aggregation-Gateway (MA-GW).

FIG. 1 is a diagram illustrating a configuration of a multi-network merging system according to an embodiment of the present invention. Referring to FIG.

Referring to FIG. 1, MultiNet Aggregation transmission is a technique of transmitting data by merging a plurality of communication networks. The multi-network aggregation transmission is a technique of dividing transmission data into a plurality of routes of the same kind of networks or a plurality of heterogeneous networks, The data transmitted through the path can be bundled and transmitted. Multi-network merging transmission may be referred to as multi-path transmission in the sense of simultaneously transmitting data to a plurality of paths.

The multi-network merging system may include a user terminal 10 and a network device 20 connected to the user terminal 10 and a plurality of networks (e.g., a 3G / LTE network and a WiFi network).

The user terminal 10 has multiple communication interfaces and can be connected to a plurality of networks at one time via multiple communication interfaces. The user terminal 10 may be equipped with a management application that allows a user to access and set up or manage a multi-network connection.

The user terminal 10 includes a network agent for performing authentication, state management, and traffic processing for merging multiple networks, and the network agent can be implemented as internal logic of the terminal. In the user terminal 10, a network agent and various applications communicate with each other through a socket. The network agent cooperates with the network device 20 in accordance with the setting information of the management application for network management.

The network device 20 merges the sub-flows transmitted in the multi-path or splits the flows transmitted in the single path into the multi-path sub-flows and transmits the sub-flows. The network device 20 is located at the contact point of the multi-network, and can be located, for example, at the contact point of the LTE network and the WiFi network.

The network device 20 divides the data to transmit the received data to the terminal 10. [ Then, the network device 20 transmits some data to the user terminal 10 through the sub-flow of the first network (e.g., LTE network), and transmits the remaining data to the second network (e.g., WiFi network) To the user terminal 10 via the sub-flow. The user terminal 10 merges data received via a plurality of communication interfaces. In the same way, the network device 20 can merge the data transmitted by the user terminal 10 using the multiple communication interfaces and transmit the merged data to the server 30.

The multi-network merging technique can be classified as follows according to the merging point.

L2 / link layer merging creates a dedicated tunnel to the WiFi AP at the boundary between the LTE core and the access network (ie, the eNB).

L3 / network layer merging creates virtual IP tunnels to integrate IP addresses independently used in LTE and WiFi networks.

L4 / transport layer merging can create a session over a single access network and then participate in data transmission regardless of the IP address scheme, if additional access networks are available. At this time, the application-level communication entity supports a structure capable of single-session-based data communication using one or more access networks.

L7 / Application layer merge reassembles data received by dedicated application / network agent through LTE network and WiFi network or transmits application protocol data separately.

In this way, it is possible to perform various merge transmissions according to the merge transport layer. In the following, merge technology using L4-based multipath TCP (MPTCP) will be described as an example.

2 is a conceptual diagram illustrating multi-network merging transmission according to an embodiment of the present invention.

2, the network device 20 transmits the original data 1, 2, 3, 4 and 5 in this scheduling round and the original data 6, 7, 8, 9 and 10 in the next scheduling round, (Not shown).

The network device 20 divides the original data 1, 2, 3, 4 and 5 in the first scheduling round to transmit some data 2 and 3 to the first network and transmits the remaining data 1 , 4, 5). The network device 20 divides the original data 6, 7, 8, 9, 10 in the second scheduling round to transmit some data 7 and 8 to the first network and transmits the remaining data 6 , 9, 10). At this time, the network device 20 displays an order in the data so that the receiver can confirm the order of the data even if the data is transmitted to the multi-network (In-order delivery).

The user terminal 10 merges the data 2, 3 received in the first network and the data 1, 4, 5 received in the second network in order of the original data 1, 2, 3, 4 , 5).

On the other hand, the state of the second network is poor and the user terminal 10 may not properly receive data through the second network. Then, the user terminal 10 requests retransmission to the network device 20. Or the network device 20 may not receive an acknowledgment (ACK) for the transmission data from the user terminal 10. [ In this case, the first network is to transmit the data 7, 8 of the second scheduling round, and the second network must retransmit the data 1, 4, 5 of the first scheduling round. Thus, in order to prevent out-of-order delivery, the network device 20 may abandon the second scheduling round opportunity through the first network even though there is a resource of the first network to transmit data . As described above, TCP, which is an end-to-end connection-based transport layer protocol, guarantees reliability by retransmitting data when it does not receive an acknowledgment (ACK) for transmission data or is requested to retransmit . However, reliability can be guaranteed by data retransmission because of a single network. However, data retransmission in any one of multiple networks is a major cause of deteriorating the merge transmission efficiency.

Alternatively, before the user terminal 10 receives the data (1, 4, 5) transmitted in the first scheduling round through the second network, the data 7, 8 transmitted in the second scheduling round through the first network (Head-of-line blocking). If the data transmitted first arrives later than the data transmitted next, the user terminal 10 may not receive the data in order and may not be able to perform the data merge process.

As described above, when the states of the multiple networks are different, the high-quality network may not be able to use the resources of the highest quality due to the influence of the low-quality network. Or if the user terminal 10 can not sequentially receive data due to the difference in quality of the multi-network, retransmission may be required even in a high-quality network in which data is correctly transmitted. In this way, even if the network state (for example, packet transmission delay time, RTT, packet loss rate, etc.) differs for each route generated in each of the multiple networks, if each of the multiple networks is individually scheduled as a single network, The efficiency is inevitably lowered.

Multi-network merging Another cause of poor transmission efficiency is the receiving window. The user terminal 10 as a receiver transmits a receive window indicating the number of bytes it can receive to the network device 20 as a sender. The receiving window may be a buffer size that is not used in the buffer. The network device 20 transmits data corresponding to the number of bytes corresponding to the reception window. This is referred to as flow control in TCP. When the user terminal 10 restricts the reception window, the packet transmission may be congested even if the transmission network is a high-quality network.

3 is a configuration diagram of a packet scheduler of a transmission apparatus according to an embodiment of the present invention.

Referring to FIG. 3, the packet scheduler 100 of the transmission apparatus includes a network status monitoring unit 110, a packet transmission rate adjustment unit 130, and a buffer delivery unit 150. Here, it is assumed that the transmitting apparatus and the receiving apparatus have generated sub-flows through a plurality of networks (for example, a first network and a second network). The packet scheduler 100 divides the original data based on the network state in which each sub-flow is generated, and adds the divided data to the data buffer of the corresponding sub-flow. The data added to the data buffer is transmitted to the MPTCP layer, the IP layer, and the network connected via L2 / L1.

The network status monitoring unit 110 collects status parameters of each network from a lower layer. The network status monitoring unit 110 can estimate the status of each network based on the network status parameters. The network status parameters that can be collected by the network status monitoring unit 110 may vary. For example, the network state parameters may include RTT, flight in-flight, receive window (rwnd) size, congestion window (cwnd) size, The number of subflows, the size of retransmission data, the size of a DSS (Data Sequence Signal), and the like.

The RTT is a parameter that can be obtained in the TCP handshake process through each network at the initial stage of session establishment.

The flight status data size is data information that has not yet received acknowledgment (Ack) from the transmission data. Since the data transmitted through the sub-flows of the respective networks are stored in the data buffers 210 and 220 of FIG. 4, the flight status data size is a parameter that can be obtained by referring to the data buffers 210 and 220 of FIG.

The receiving window size may be included in the TCP header of the acknowledgment (Ack) received from the receiver or may be inferred based on the response. The receive window size is a parameter that can be obtained from a variable ("rwnd") indicating the size of the receive window in the Linux kernel or the like.

The congestion window size is a parameter that can be obtained from a variable ("cwnd") indicating the congestion window size in the Linux kernel and so on.

The size of data waiting in the data buffer is the size of the data to be transmitted in this scheduling round, indicating the size of data that is transmitted and left over time. Here, the data buffer is the original data buffer (200 in FIG. 4), and the data size waiting in the data buffer can be obtained from the value indicated by "skb.len ". Here, "skb" means a socket buffer, and "len " means a length.

The sub-flow number, the retransmission data size, the DSS size, and the like are parameters that can be obtained from the network status monitoring unit 110. The DSS size is the size of the stream combining sub-flows by path, and is a term used in MPTCP. For example, if a subflow is generated in each of the LTE network and the WiFi network, the DSS size is the data size transmitted to the LTE subfluid and the WiFi subflow.

The network status can be estimated based on at least one network status parameter. The network state may mean quality or transmission rate and may be quantified as the optimal transmission byte of the network. For example, if the optimal transmission byte of the first network is calculated as 1500 bytes, the first network state may be quantified as 1500 bytes. In the future, quality or transmission speed can be described as an example of network conditions.

For example, the network status monitoring unit 110 can determine the network status based on the transition of the TCP sequence number. The change in the TCP sequence number can be calculated as a sequence number change amount (slope) that is increased per unit time, as information indicating whether the accumulated number of transmission bytes changes rapidly or slowly with time. The transmission rate can be estimated based on the TCP sequence number change trend. Transition of TCP sequence number can be calculated including SACK (Selective Ack) and DSACK (Duplicate SACK).

The TCP sequence number (hereinafter simply referred to as "sequence number") is information included in the TCP header and written in a field specified for each TCP segment including transmission data. The receiving apparatus confirms the receiving number in the receiving TCP segment, and transmits a description of the sequence number to be received next in the acknowledgment. Based on the sequence number included in the acknowledgment, the transmitting apparatus can determine whether to send the next data or retransmit the already-transmitted data based on the sequence number included in the acknowledgment, thereby estimating the state or network state of the receiving apparatus. Alternatively, the transmitting device may transmit multiple TCP segments within the transmission window size and receive a single acknowledgment from the receiving device. According to this sequence numbering scheme, if the increase rate of the sequence number (the sequence number change per unit time) is almost constant, it can be judged that the network status is stable or the resource usage has reached the threshold value. If the increase rate of the sequence number is increased, it can be judged that the network status is improved or the resource usage remains. If the increase rate of the sequence number is slowed down, it can be judged that the network state is deteriorated or the resource usage is insufficient. Or the sequence number change amount (slope) increased per unit time is steeply changed, it can be determined that the transmission speed of the network is fast. If the sequence number change (slope) increases per unit time or if there is no change, it can be judged that the transmission rate of the network is slow or the data transmission is not proceeding any further. In this way, the network status monitoring unit 110 can grasp the transmission status and transmission pattern of the network based on the TCP sequence number change trend.

If the RTT is equal to or greater than the reference value, the network status monitoring unit 110 may determine that a packet transmission delay is expected or may be a low quality network.

The network status monitoring unit 110 can determine the network status, that is, the quality or the transmission rate, based on the flight status data size. The network status monitoring unit 110 determines how active the network is participating in transmission based on the size of the flight status data . The network status monitoring unit 110 can grasp the network status based on a reception window size, a transmission window size, a data size waiting in a data buffer, a retransmission data size, a DSS size, and the like.

The packet transmission rate adjustment unit 130 adjusts a transmission rate to be transmitted to each sub-flow (or a route generated in each network) based on the status parameters of the respective networks obtained by the network status monitoring unit 110. Transmission data is divided into packet transmission unit MTU (Maximum Transmission Unit) or MSS (Maximum Segment Size) and packetized. Therefore, the packet transmission amount of each network can be determined by the number of packets or the number of TCP segments. Alternatively, the packet transmission rate of each network can be determined as a transmission sharing rate (transmission rate) in which each network shares the data transmission. The packet transmission amount adjustment unit 130 can adjust the ratio of the opportunities divided into segments.

The time point at which the packet scheduler 100 controls the amount of transmission may be variously set.

According to one embodiment, the packet transmission rate controller 130 may determine a transmission rate to be transmitted to each sub-flow based on status parameters of each network for each scheduling round.

According to another embodiment, the packet transmission rate controller 130 may determine a transmission rate for each sub-flow based on status parameters of each network acquired at a specific time, and schedule the transmission rate for a predetermined time based on the determined transmission rate for each sub-flow . Here, the predetermined time means a time during which a plurality of scheduling rounds can proceed.

According to another embodiment, the packet transmission rate controller 130 may determine whether to start the transmission rate control for each subflow based on the scheduling change condition. The scheduling change condition may be referred to as a triggering condition, and may be configured by a combination of a network state parameter and a network state. The control of transmission rate by sub-flow means that packet scheduling of each network is integrated based on the states of a plurality of networks. For example, when the at least one network state is improved to be equal to or greater than the reference value, or the state is at least equal to or less than the reference value, or when the state difference of at least two networks is equal to or greater than the reference value, It is possible to enter the transmission amount control for each sub-flow.

The scheduling change condition can be set based on various network status parameters. If the scheduling change condition is satisfied, it is triggered to enter the transmission amount control per sub-flow. The scheduling change condition may be, for example, when the transmission quality value of at least one network has reached a threshold value, or at least when the transmission quality value difference of the two networks has reached a threshold value. The transmission quality value is determined from at least one piece of information capable of evaluating the transmission quality of each network, for example, RTT, transmission rate, transmission sharing ratio (transmission ratio), retransmission data size, reception window size, transmission window size, For example, if the difference between the reception windows of at least two networks is equal to or greater than the reference value, or if the RTT difference is equal to or greater than the reference value, the control may be triggered to enter the transmission rate control for each sub-flow. It may be triggered to enter the transmission amount control per subflow when the acknowledgment not received data (flight status data) transmitted to at least one network occurs more than the reference value or when the acknowledgment not received data continuously occurs beyond the reference value.

The scheduling change condition can take into account the number of sub-flows. That is, when the number of subflows is equal to or greater than the reference value, the control of the amount of transmission for each subflow can be started.

The scheduling change condition can take into account the DSS size. That is, when the size of the data stream transmitted to the sub-flows is equal to or larger than the reference value, the control of the amount of transmission for each sub-flow can be started.

When the scheduling change condition entering the sub-flow-based transmission amount control is released, the transmission amount control for each sub-flow is triggered to be released. When the transmission amount control for each subflow is released, the packet transmission amount adjustment unit 130 performs packet scheduling according to the default scheduling method. This is because it is not necessary to determine the amount of data to be transmitted to each network for each scheduling round when the first network and the second network are in a similar state or when the first network and the second network are both good. The default scheduling method may vary and may be, for example, a round-robin scheme. The default scheduling method may be a method of transmitting according to the size of the receiving window of each network. The default scheduling method may be a conventional scheduling scheme that considers each of the plurality of networks as a single network.

The manner in which the packet scheduler 100 determines the amount of transmission scheduled for each network in the scheduling round may vary.

According to one embodiment, the packet transmission rate controller 130 identifies a high quality network among a plurality of networks based on status parameters of each of a plurality of networks. The packet transmission rate controller 130 schedules a data amount of a size smaller than a reception window or a reception window of the high-quality network in the high-quality network. The packet transmission rate adjustment unit 130 schedules a certain amount of data to the low-quality network based on the reception window of the low-quality network. At this time, instead of transmitting the amount of data corresponding to the reception window of the high-quality network, the packet transmission rate controller 130 may transmit the amount of data corresponding to the reception window difference between the high-quality network and the low-quality network. The reason why the amount of data is reduced as compared with the reception window is that even if data transmitted to the low-quality network is lost, the data order is shifted due to data retransmission in the next scheduling round, It is possible to prevent the problem of re-transmission of the data. That is, if the quality of the low-quality network is not good, the packet transmission rate controller 130 receives the transmission amount of the high-quality network so that the lost data can be transmitted in the next scheduling round It is less than Windows.

According to another embodiment, the packet transmission amount adjustment unit 130 can determine the optimal data transmission ratio to a plurality of networks based on the state parameters of each of the plurality of networks. B, c, ..., n, and the weight of each state parameter is Wa (i, j), for example, , Wn, Wc, ..., Wn, and if a plurality of networks exist (N1, N2, N3, ..., Nn), the state values of the respective networks are calculated as shown in Equation (1). If the weight is a positive value or a value that raises the state parameter value, and if it is a negative state parameter (e.g., RTT, packet loss rate, etc.), then the weight is negative or the state parameter May be a value that reduces the value. In Equation (1), N is any one of N1 to Nx.

The packet transmission rate controller 130 normalizes the ratio of the status values of the respective networks among a plurality of networks and determines the ratio as a data transmission rate of each network. For example, the packet transmission rate controller 130 may have a transmission rate calculated from the state value of the first network is 75%, and a transmission rate calculated from the state value of the second network may be 25%. The packet transmission rate controller 130 divides the transmission rate into transmission ratios. For example, if the transmission rate is 2000 bytes, the packet transmission rate adjuster 130 may transmit 1500 bytes to the first network and 500 bytes to the second network. The amount of transmission is represented by the number of packets, and the number of transmission packets can be divided by the transmission rate.

The buffer transfer unit 150 divides the original data on the basis of the transfer amount per sub-flow determined by the packet transfer amount control unit 130, and adds the divided data to the data buffer of the corresponding sub-flow. The data added to the data buffer of each sub-flow is transmitted through the MPTCP layer and the IP layer, and is transmitted to the receiver through the L2 / L1 network layer.

4 is a conceptual illustration of a network stack in which a packet is transmitted in a transmitting apparatus according to an embodiment of the present invention.

Referring to FIG. 4, when the data to be transmitted to the receiving apparatus is filled in the original data buffer 200, the packet scheduler 100 determines how much data is to be transmitted to each of the plurality of networks. Or the packet scheduler 100 obtains a kernel context from the processor of the transmitting apparatus and acquires an opportunity to perform socket communication, the packet scheduler 100 performs a pointer movement operation (descriptor designation) through the data buffer 200 allocated by the socket ) To start packet scheduling

The original data buffer 200 is a storage for storing data to be transmitted to the receiving apparatus. The original data may be data received from the outside, or data generated in the application of the transmitting apparatus.

The packet scheduler 100 collects status parameters of each network. The packet scheduler 100 determines the state of each network based on the state parameters of the respective networks. The packet scheduler 100 can determine whether to enter the throughput control for each subflow based on the state of each network. The packet scheduler 100 can release the transmission rate control for each sub-flow and perform packet scheduling according to the default scheduling method when the condition for entering the transmission rate control for each sub-flow is canceled based on the state of each network. Packet scheduler 100 may determine the amount of data to be delivered to each sub-flow per scheduling round. The amount of data to be delivered to each sub-flow may be the ratio of data to be delivered to each sub-flow. The packet scheduler 100 divides the data stored in the original data buffer 200 based on the data amount / data ratio for each sub-flow. The packet scheduler 100 adds the divided data to the data buffers 210 and 220 of the corresponding sub-flows. The data buffers 210 and 220 may be transmit socket buffers.

The MPTCP layer splits the data added to the data buffers 210 and 220 of each sub-flow into TCP segments. The IP layer associated with each sub-flow creates a packet with an IP header attached to the TCP segments. The packet is transmitted to the receiving device via the L2 / L1 network layer.

FIG. 5 is a diagram for explaining a merging transmission scheduling method when the present invention is not applied, and FIG. 6 is a view for explaining a merging transmission scheduling method according to an embodiment of the present invention.

First, TCP, an end-to-end connection-based transport layer protocol, operates based on flow control, congestion control, and error control. TCP is characterized in that, when an acknowledgment (ACK) for transmission data is not received, retransmission of acknowledgment / non-acknowledgment data ensures reliability.

Flow control means that the transmitting device transmits the amount of data allowed by the receiving window of the receiving device. The receiving window informs the receiving device.

Congestion control means to limit the amount of data that flows into the network to prevent congestion in the network. Congestion control allows transmission of the amount of data allowed by the congestion window. Congestion control is implemented by the transmitting device.

Congestion control adjusts the amount of transmission at the point of time when congestion occurs, gradually increasing the amount of data transmitted by the transmitting device to the receiving device. Congestion control can control the communication speed by occupying the maximum possible network resources. At this time, the TCP congestion control gradually increases the transmission amount by the transmitting apparatus according to the acknowledgment of the receiving apparatus. Transmission volume increases exponentially from session connection to slow start. It is a congestion avoidance (ca) interval from the slow start threshold (ssthresh) interval defined by the setting. In the congestion avoidance period, the amount of transmission increases linearly with receipt of the acknowledgment. The amount of transmission used at this time is called the congestion window. The size of the congestion window increases as the TCP sequence number increases. If the data can not be transmitted to the expected degree, or if the sequence number is abnormally received and normal data communication is not possible, the congestion window size is adjusted, Lt; / RTI >

However, when data is transmitted through multiple networks, if the congestion control mechanism designed based on a single network is used as it is, retransmission or congestion control caused by the first network state affects the second network. Therefore, as described with reference to Fig. 2, the order of transmission of the original data received by the receiving apparatus may be out of order.

Meanwhile, the data transmission amount / transmission ratio / transmission opportunity transmitted to each network is determined by the packet scheduler, not the TCP layer of the transmitting apparatus. However, since the conventional packet scheduler schedules only a predetermined manner without knowing the network status, it is possible to change the scheduling dynamically even though the state of one network affects the other network, it's difficult. That is, the conventional network device performs congestion control only at the TCP layer, and it is difficult to expect the transmission rate / transmission rate / transmission opportunity control by the packet scheduler at the time of congestion or before congestion occurs.

Referring to FIG. 5, it is assumed that the first network is a high-quality network having a fast RTT and a low packet loss rate, and the second network is a low-quality network having a slow RTT and a high packet loss rate than the first network.

In the case of scheduling a packet transmitted to multiple networks in a conventional TCP congestion control scheme, the initial transmission amount of the first network increases due to congestion control, but the packet transmission delay or retransmission of the first network is delayed The network may be affected and the throughput may be reduced.

For example, the transmitting apparatus transmits certain data to the receiving apparatus in the first scheduling round, and receives a reply from the receiving apparatus that the receiving window of the first network is 2000 bytes and the receiving window of the second network is 500 bytes. Then, the transmitting apparatus transmits 2000 bytes to the first network in the second scheduling round, and transmits 500 bytes to the second network. However, the state of the second network may be poor and the transmitting apparatus may not receive an acknowledgment of the data transmitted from the receiving apparatus to the second network. At the third scheduling round, the transmitting apparatus can wait for the next data to be transmitted to the second network or retransmit the previous data, and can send the next data to the first network or abandon the transmission opportunity. As a result, if the TCP congestion control or the retransmission is performed in any network, the order of data received by the receiving apparatus may be changed and the transmitting apparatus may retransmit data sent to the first network of high quality.

As described above, according to the conventional scheduling scheme, although the first network is of high quality, the packet transmission amount and the merge transmission efficiency may be lowered due to the loss of the packet scheduling opportunity due to the influence of the second network.

Referring to FIG. 6, the packet scheduler 100 of the present invention manages the state of each network. In particular, the packet scheduler 100 manages the reception windows of each network. The packet scheduler 100 dynamically determines a transmission amount / transmission rate to be shared by each network based on the reception window.

For example, the transmitting apparatus transmits certain data to the receiving apparatus in the first scheduling round, and receives a reply from the receiving apparatus that the receiving window of the first network is 2000 bytes and the receiving window of the second network is 500 bytes. At this time, if the first network and the second network are both in good network quality, the packet scheduler 100 of the transmitting apparatus transmits scheduling information such as 2000 bytes to the first network and 500 bytes to the second network, can do.

If the packet scheduler 100 determines that the second network is a lower-quality low-quality network based on the state parameters of the second network, it can be predicted that the packet loss rate is equal to or higher than the reference value. That is, the packet scheduler 100 determines that the transmission rate control for each sub-flow is necessary when the acknowledgment is delayed or the acknowledgment is not received with respect to the data transmitted to the second network. The transmitting apparatus must transmit the original data in order. If the receiving apparatus does not receive the acknowledgment of the data already sent to the second network, it can not transmit the next data to the first network.

The packet scheduler 100 schedules the first network with a transmission amount smaller than the reception window (2000 bytes) of the first network in the second scheduling round. Here, the transmission amount less than the reception window may be a value (1500 bytes) excluding the amount of data (for example, 500 bytes) transmitted from the reception window (2000 bytes) of the first network to the second network. If the packet scheduler 100 fails to acknowledge the 500 bytes sent to the second network, the packet scheduler 100 may send 500 bytes not yet sent in the second scheduling round to the first network when the third scheduling round is reached. Alternatively, the packet scheduler 100 may send a transmission amount including 500 bytes that were sent from the second scheduling round to the second network but failed to receive an acknowledgment to the first network.

As described above, the packet scheduler 100 determines whether or not the acknowledgment of the packet transmitted to the second network of the lower quality in the second scheduling round is received, the next packet of the original data, which is scheduled to the first network in the second scheduling round, , It reinjects the next data remaining in the third scheduling round to the first network. In this case, since the next data scheduled in the first scheduling round in the second scheduling round includes the data scheduled in the second network during the second scheduling round, the receiving apparatus sequentially transmits through the first and second networks Received packets, and merges them to obtain original data.

When the packet scheduler 100 controls the amount of transmission of the first network considering the state of the second network, the packet scheduler 100 can quickly combine the data sequence transmitted from the receiving apparatus to the multi-network, It can speed up. This is because even if the acknowledgment is not received from the second network of low quality, it is possible to avoid the situation in which the first network is forced to retransmit the first network in order to lose the transmission opportunity or to arrange the order of the data.

On the other hand, the packet scheduler 100 determines a transmission amount of each network based on the size of the reception window included in the acknowledgment. At this time, the packet scheduler 100 determines the number of bytes less than the number of bytes corresponding to the reception window as the transmission amount.

In fact, the size of the receiving window that the receiving device is informed of may be somewhat higher than the actual receivable amount. Accordingly, the packet scheduler 100 determines whether the size of the reception window indicated by the reception apparatus is higher than the actual receivable amount based on the trend of increase of the TCP sequence number and the change of the size of the reception window received from the reception apparatus . Based on the determination result, the packet scheduler 100 determines the number of bytes less than the number of bytes corresponding to the reception window as the transmission amount.

As described above, if the packet scheduler 100 transmits a smaller amount of data than the receiving window of each network, packet retransmission / reinjection due to data sequence error can be avoided in the entire original data configuration.

7 is a flowchart of a packet scheduling method according to an embodiment of the present invention.

Referring to FIG. 7, in a state where a sub-flow is generated through each of a plurality of networks for multi-network merging transmission, the packet scheduler 100 schedules data to be transmitted to each sub-flow if there is data to be transmitted to the receiving apparatus.

 The packet scheduler 100 collects status parameters of each of a plurality of networks connected to the receiving apparatus (S110). The network status parameters may include RTT, flight status data size, receive window size, congestion window size, data size waiting in the data buffer, number of subflows, retransmission data size, DSS size, and the like.

The packet scheduler 100 determines a point in time when the integrated scheduling for a plurality of networks is required based on the status parameters of each of the plurality of networks (S120). That is, the packet scheduler 100 determines whether to enter the transmission rate control for each sub-flow. The scheduling integration criteria may be when the reception window of at least one network has reached a threshold value or when the RTT has reached a threshold value, as described in FIG. Or the scheduling integration criteria may be at least when the reception window difference of the two networks is above the threshold or when the RTT difference is above the threshold. For example, if the reception window of the second network reaches a threshold value or the RTT reaches a threshold value, it can be determined to be the integrated scheduling time point. If the difference between the reception windows of the first network and the second network is equal to or greater than the threshold value or the RTT difference is equal to or greater than the threshold value, it can be determined as the integrated scheduling time point.

In the case of the integrated scheduling time, the packet scheduler 100 determines a first network of high quality among a plurality of networks based on the state parameters of the respective networks (S130).

The packet scheduler 100 determines the amount of transmission of the second network based on the reception window of the second network that is not the first network among the plurality of networks (S140). The transmission amount of the second network is equal to or less than the reception window of the second network.

The packet scheduler 100 determines a data amount smaller than the reception window of the first network by a predetermined amount as the transmission amount of the first network (S150). Here, the predetermined size may be the transmission amount of the second network.

The packet scheduler 100 adds the data corresponding to the transmission amount of the first network and the data corresponding to the transmission amount of the second network to the sub-flow data buffer of the corresponding network (S160). The data stored in the data buffer of the sub-flow is divided into TPC segments in the MPTCP layer and is generated as IP packets in the IP layer and transmitted to the receiving device.

The packet scheduler 100 determines a time point at which unified scheduling is released based on the state parameters of each of the plurality of networks (S170). If the scheduling integration criterion is not satisfied, it can be judged that the integrated scheduling is canceled.

In the case of the unified scheduling release point, the packet scheduler 100 schedules data to be transmitted to each of the plurality of networks according to the default scheduling method (S180).

On the other hand, the packet scheduler 100 can calculate the transmission amount of the first network and the transmission amount of the second network for each scheduling round. Alternatively, the packet scheduler 100 can use the transmission amount of the first network and the transmission amount of the second network, which are calculated once, in a plurality of scheduling rounds. The transmission rate of the first network and the transmission rate of the second network can be calculated from the transmission ratios of the first and second networks and the total transmission amount in the scheduling round.

8 is a flowchart of a packet scheduling method according to another embodiment of the present invention.

Referring to FIG. 8, the packet scheduler 100 receives acknowledgments from the receiving apparatus for data transmitted to the first network and the second network in the first scheduling round (S210).

The packet scheduler 100 confirms the reception window of the first network and the reception window of the second network based on the acknowledgments (S220).

The packet scheduler 100 determines whether the amount of traffic to be transmitted to the first network and the second network is necessary based on the status parameters and the triggering conditions of each network (S230). 3, when the reception window of at least one network reaches the threshold value or when the transmission quality value (for example, RTT, transmission delay time, packet loss rate, etc.) reaches the threshold value Lt; / RTI > Or the determination criterion may be at least when the reception window difference between the two networks is equal to or greater than the reference value or when the transmission quality value difference is equal to or greater than the reference value.

The packet scheduler 100 determines a first network of high quality among the first and second networks and transmits a data amount smaller than the reception window of the first network by a predetermined amount to the first one of the first and second networks in the second scheduling round, Is determined as the network transmission amount (S240). Here, the predetermined size may be the transmission amount of the second network.

The packet scheduler 100 determines a second network transmission amount in a second scheduling round based on a reception window of a second network of a lower quality among the first and second networks (S250).

The packet scheduler 100 confirms that an acknowledgment transmitted by the receiving apparatus has been received, to the data transmitted to the first network in the second scheduling round (S260).

At the time when the third scheduling round arrives, the packet scheduler 100 confirms that the acknowledgment is not received for the data transmitted to the second network in the second scheduling round (S270).

The packet scheduler 100 adds the data transmitted to the second network in the second scheduling round to the data buffer to be transmitted to the first network in the third scheduling round (S280).

9 is a flowchart of a packet scheduling method according to another embodiment of the present invention.

Referring to FIG. 9, the packet scheduler 100 varies a scheduling method based on state parameters of each of a plurality of networks connected to a receiving apparatus. The packet scheduler 100 changes the scheduling method from the first method to the second method when the state parameters of each of the plurality of networks satisfy the scheduling change condition. The packet scheduler 100 returns the scheduling method to the first method in the second method when the scheduling change condition is released. It is assumed that the first method is a default scheduling method and that the second method is a Multinet Aggregation Packet Scheduling (MA-PS) method for controlling the throughput per sub-flow, and the name of the scheduling method is of course And can be variously changed. In the following, the total transmission segment size may have a meaning corresponding to the number of segments transmitted with a certain size. Next, a method of controlling the amount of transmission with the segment size control will be described.

When a subflow is created through at least one network (session establishment), the packet scheduler 100 starts scheduling by the default scheduling method (S310). The default scheduling method is predetermined to operate regardless of the status parameters of each of the plurality of networks.

The packet scheduler 100 determines whether the scheduling change condition (triggering condition) is satisfied based on the state parameters of each of the plurality of networks (S320). The scheduling change condition is information for triggering to enter the transmission rate control per sub-flow, and can be configured as a combination of various conditions. For example, the scheduling change condition can use the number of subflows and the size of the DSS as condition parameters. If the number of subflows exceeds the reference value A, the packet scheduler 100 determines whether the size of the DSS exceeds the reference value B (S322, S324). If the number of subflows is less than or equal to the reference value A or the DSS size is less than or equal to the reference value B, the packet scheduler 100 maintains the default scheduling method (unsatisfactory scheduling change condition) (S322, S324). If the DSS size exceeds the reference value B, the packet scheduler 100 determines that the scheduling change condition is satisfied.

If the scheduling change condition is satisfied, the packet scheduler 100 starts the multi-network merge scheduling (MA-PS) (S330).

The packet scheduler 100 determines at least one surf flow as a transmission rate control object (simply referred to as a "control object") (S340). The control object can be variously determined, and for example, a high-quality sub-flow among a plurality of sub-flows can be determined as a control object. Next, the transmission amount control for the sub-flow generated in the LTE network will be described. The transmission rate control for the generated sub-flows in the WiFi network can also be performed separately. On the other hand, when the multi-network merging scheduling fails to receive the acknowledgment of the data transmitted to the WiFi network, the acknowledgment non-receipt data can be reinjected into the LTE network. Do not.

The packet scheduler 100 determines whether to adjust the total transmission segment size to be transmitted per unit time through the control object based on the network status parameter of the control object (S350). Here, the segment size can mean the size of a single segment if a segment larger than the MTU can be generated in the TCP stack by TSO (TCP Segment Offload), otherwise the total size (total transmission amount) . ≪ / RTI > The unit segment may be an MTU or an MSS, which is a TCP transmission unit. The packet scheduler 100 uses the state parameters of the control object, for example, the flight state data (in-flight) size, the reception window size, the congestion window size, and the retransmission data (retx) size Determine if the segment size should be adjusted. For example, in step S350, the packet scheduler 100 may determine whether to adjust the total segment size based on the flight state data size.

If the flight state data occurs beyond the reference value C, the packet scheduler 100 determines the total transmission segment size (S360). That is, when the flight state data occurs beyond the reference value C, the packet scheduler 100 determines the segment size to be transmitted with increasing or decreasing the total transmission segment size based on the network state parameter of the control object. The amount of transmission data per unit time may be varied through step S370.

 When the total transmission segment size is determined, the packet scheduler 100 compares the total transmission segment size with the data size (data size indicated by skb.len) waiting in the data buffer 200 (S370).

If the data size (skb.len) waiting in the data buffer 200 is larger than the total transmission segment size, the packet scheduler 100 allocates the total transmission segment size from the data waiting in the data buffer 200 to the control target (LTE network (S372).

If the data size (skb.len) waiting in the data buffer 200 is smaller than the total transmission segment size, the packet scheduler 100 transmits data (skb.len) waiting in the data buffer 200 to the control target (LTE network) (S374).

On the other hand, if the flight state data is less than or equal to the reference value C in step S350, the packet scheduler 100 moves to step S370. At this time, the packet scheduler 100 does not go through the total transmission segment size adjusting step of step S360, but follows the general method. That is, if the total transmission segment size (MTU size * segment number) determined based on the available congestion window size (the difference between the cwnd size and the in-flight size) is larger than skb.len, skb.len And transmits the total transmission segment size if it is smaller than skb.len.

The packet scheduler 100 determines whether multi-network merging scheduling is released based on the network status parameter of the control object after transmitting at least some data waiting in the data buffer (S380). The packet scheduler 100 may determine whether to release the scheduling change condition based on the data size (skb.len) stored in the original data buffer 200. [ If the data size stored in the original data buffer 200 becomes a reference value (for example, 0), the packet scheduler 100 moves to the default scheduling method (moves to S310). If the size of the data stored in the original data buffer is larger than the reference value, the packet scheduler 100 maintains the transmission amount control for the control object (moves to S350).

On the other hand, in step S360, various network status parameters may be used to determine the total transmission segment size of the current scheduling round. For example, flight state data (in-flight) size, reception window (rwnd) size, congestion window size (cwnd), retransmission data (retx) size can be used.

The packet scheduler 100 determines whether the total transmission segment size is adjusted up or down based on the network status parameters of the control object (S362). The judgment conditions can be set as shown in Table 1. At this time, the total transmission segment size is adjusted upward or downward from the reference value, which may be an available congestion window size (available cwnd) or the number of segments that can be transmitted in an available congestion window size. Here, the available congestion window size can be defined as a value excluding the flight state data size (in-filght) in the congestion window size.

Condition Judgment result in-flight <rwnd (or cwnd)
and / or
retx <threshold Total transmission segment size or total number of transmission segments)
Upwards than the available congestion window (cwnd) size in-flight> = rwnd (or cwnd)
and / or
retx> = threshold The total transmission segment size or the total number of
Lower than the available cwnd size

The packet scheduler 100 determines the total transmission segment size adjusted upwards than the available congestion window size based on the determination result (S364).

Alternatively, the packet scheduler 100 determines the total transmission segment size that is lower than the available congestion window size based on the determination result (S366).

For example, assume that the available congestion window size is 5000Bytes and the unit segment size is 1500Bytes. The number of segments that can be transmitted per unit time in the available congestion window size is three.

However, if the determination result in Table 1 is the upward adjustment, the packet scheduler 100 adjusts the total transmission segment size to a value larger than the available congestion window size (6000 bytes). Then, the LTE network can transmit more than four segments before segment size adjustment.

If the determination result in Table 1 is a downward adjustment, the packet scheduler 100 adjusts the total transmission segment size to a value smaller than the available congestion window size (3000 bytes). Then, the LTE network can transmit less than two segments before segment size adjustment.

At this time, it is possible to adjust the total transmission segment size based on the unit segment size to be adjusted to 6000Bytes larger than the available congestion window size or down to 3000Bytes smaller than the available congestion window size. That is, the packet scheduler 100 can perform at least one transmission or at least one less transmission control than the number of segments that can be transmitted at the current available congestion window size based on the state parameters of the LTE network to be controlled .

Packet scheduler 100 may adjust the number of segments based on the generic segmentation offload (GSO) of TCP. Accordingly, the packet scheduler 100 can obtain the effect of varying the number of transmission segments (or the number of packets) per unit time. On the other hand, a code level indicating whether to apply the GSO segment size (<64 KB) to a specific path (for example, a sub-flow path created in the WiFi network) may be used.

10 is a hardware block diagram of a multiplexing transmission apparatus according to an embodiment of the present invention.

10, a multi-network merge transmission apparatus 1000 in which a packet scheduler 100 is implemented includes a processor 1100, a memory device 1200, a storage device 1300, a plurality of communication devices 1400, and the like And stores a program that is executed in combination with hardware at a specified location. The hardware has a configuration and performance capable of executing the method of the present invention. The program includes instructions implementing the method of operation of the present invention described with reference to FIGS. 1-9, and is implemented in conjunction with hardware, such as processor 1100 and memory device 1200, to implement the present invention. The operation of the packet scheduler 100 is driven by the processor 1100 and the original data buffer 200 and the data buffers 210 and 220 of each subflow can be implemented at designated locations in the memory device 1200. The MPTCP / IP algorithm can be implemented in the operating system kernel domain.

The embodiments of the present invention described above are not implemented only by the apparatus and method, but may be implemented through a program for realizing the function corresponding to the configuration of the embodiment of the present invention or a recording medium on which the program is recorded.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

Claims (20)

A packet scheduling method of a multi-network merging transmission apparatus,
Collecting network status parameters for each of a plurality of networks coupled to the receiving device,
Determining a control network among the plurality of networks based on the network state parameters of each of the plurality of networks, and
Adjusting a data amount per unit time to be transmitted to the control network based on at least one network status parameter for the control network
Lt; / RTI &gt; The method of claim 1,
The network state parameters
A packet round-trip time, a flight status data size, a reception window size, a congestion window size, and a retransmission data size that have not yet received an acknowledgment from transmission data. The method of claim 1,
The step of adjusting the amount of data per unit time
And adjusting the amount of data per unit time when a size of data that has not yet received an acknowledgment from the data transmitted through the control network is larger than a first reference value. The method of claim 1,
The step of adjusting the amount of data per unit time
Determining a transmission data amount per unit time that is more or less than a reference value based on at least one network status parameter for the control network,
Wherein the reference value is a number of transmittable segments in an available congestion window size of the control network or the available congestion window size. 5. The method of claim 4,
Transmitting the determined amount of data to be transmitted from the data buffer waiting in the original data buffer to the control network
Lt; / RTI &gt; The method of claim 1,
Further comprising determining whether to change from a first scheduling method to a second scheduling method based on a scheduling change condition,
The step of determining the control network
Wherein when the first scheduling method is changed to the second scheduling method, a control network is determined among the plurality of networks,
The first scheduling method is a scheduling method used in a single network,
Wherein the second scheduling method is a multi-network merge scheduling method. The method of claim 6,
The scheduling change condition
And a determination condition configured by a number of subflows generated in the plurality of networks and a size of a data stream transmitted to the subflows (Data Sequence Signal, DSS). The method of claim 6,
The scheduling change condition
When transmission quality of any one of the plurality of networks is improved to be equal to or higher than a first reference value and transmission quality of any one of the plurality of networks is deteriorated to a second reference value or more, And the difference is equal to or greater than a third reference value. The method of claim 6,
Wherein when the scheduling change condition is released after the first scheduling method is changed to the second scheduling method, the step of returning to the first scheduling method
Lt; / RTI &gt; The method of claim 6,
Wherein the first scheduling method is changed to the second scheduling method, and when data transmission to the receiving terminal is completed, returning to the first scheduling method
Lt; / RTI &gt; A packet scheduling method of a multi-network merging transmission apparatus,
Determining whether segment size adjustment to be sent to the subflow is necessary based on a data size that has not yet received an acknowledgment from among data transmitted through an arbitrary subflow;
Determining a total transmission segment size to transmit to the sub-flow based on at least one network status parameter associated with the sub-flow if segment size adjustment is required; and
The step of transmitting data of the determined total transmission segment size
Lt; / RTI &gt; 12. The method of claim 11,
The step of determining whether the segment size adjustment is necessary
And determining that segment size adjustment is necessary if the size of the data that does not receive the acknowledgment is greater than a first reference value. 12. The method of claim 11,
The step of determining the total transmission segment size
Comparing the network state parameters associated with the sub-flow to determine a total transmission segment size greater or smaller than the reference value,
Wherein the reference value is a number of transmittable segments in an available congestion window size of the control network or the available congestion window size. The method of claim 13,
The network state parameters associated with the sub-
Wherein the at least two parameters include at least one of a data size, a reception window size, a congestion window size, and a retransmission data size that have not yet received an acknowledgment from the data transmitted through the sub-flow. A packet scheduler operating by at least one processor,
A network status monitoring unit for collecting network status parameters of a plurality of networks connected to the receiving apparatus,
A packet transmission rate controller for controlling a transmission rate of each sub-flow generated in each of the plurality of networks based on the network status parameters of the plurality of networks,
A buffer transfer unit for dividing the original data to be transferred to the receiving apparatus based on the transmission amount for each sub-flow and adding the divided data to the data buffer of each sub-
Lt; / RTI &gt;
Wherein data stored in a data buffer of each sub-flow is delivered to a network in which each sub-flow is generated through a multi-path TCP (MPTCP) layer and an IP (Internet Protocol) layer. 16. The method of claim 15,
The network state parameters
The round trip time of packets transmitted to each network, the size of data not yet acknowledged among the data transmitted to each network, the size of the receiving window of each network, the size of the congestion window of each network, The size of the retransmission data of the packet scheduler. 16. The method of claim 15,
The packet transmission amount determination unit
Determining whether to start the transmission rate control for each subflow based on the network status parameter of each of the plurality of networks, and when starting the transmission rate control for each subflow, calculating a transmission data amount in at least one network among the plurality of networks, To the packet scheduler. The method of claim 17,
The packet transmission amount determination unit
And determines whether to start the transmission rate control for each subflow based on a determination condition including a number of subflows generated in the plurality of networks and a size of a data stream transmitted to the subflows (Data Sequence Signal, DSS). The method of claim 17,
The packet transmission amount determination unit
When transmission quality of any one of the plurality of networks is improved to be equal to or higher than a first reference value and transmission quality of any one of the plurality of networks is deteriorated to a second reference value or more, And the difference is equal to or greater than a third reference value. The packet scheduler according to claim 1, further comprising: 16. The method of claim 15,
The packet transmission amount determination unit
Wherein the packet scheduler varies the number of segments transmitted or the total transmission segment size per unit time based on the network status parameter of each of the plurality of networks.

Images