소프트웨어 정의 네트워크에서 데이터 평면 애플리케이션을 활용하여 전송 경로를 설정하는 장치 및 방법

The above features and effects of the present invention in conjunction with the following detailed description appended drawing more clear to be at the predetermined level, thereby the present invention is embodiment of the present invention is provided to be hereinafter for the technical idea in person with skill in the art are disclosed. The present invention refers to various modification may take the form an IGBT can apply various, specific embodiment example drawing and examples detailed herein to broadcast receiver. However this particular disclosure be but is defined with respect to the present invention form, all changing range of idea and techniques of the present invention, including the water to replacement should understood to evenly. A term used in a particular embodiment used to explain examples only specification that, defining the present invention intending to be is endured.

Each drawing while similar references in a similar components are described used for substrate.

Number 1, number 2 describes various components such as term can be used but, in terms of said components are defined by said back like. Components are mounted to one of said terms are used only distinguished from other components of the object.

For example, number 1 number 2 outside of the range of the present invention rights without components can be termed components, similarly number 2 of the elements can be termed component number 1. ". ". "". The term "and/or" that the items or the combination of a plurality of associated substrate any of the item comprises an item stored in a plurality of associated substrate.

Not defined differently, scientific or technical terms so that all terms in the present invention thus is provided to the person with skill in the art will generally by the same reconciliation is disclosed.

Generally dictionary used for providing language translators such as defined terms have the meanings associated technology must be consistent semantics and having interprets, the application manifest in defined not, or should not be interpreted as meaning ideal excessively on the moving substrate.

Components which are used in the description hereinafter for suffix "module", "block" and "part" is given the specification database or as mixed into consideration ease only, not the having the meanings or distinguished from each other serves itself.

Hereinafter, a preferred embodiment of the present invention with reference to the drawing objects is for example embodiment described hereinafter can be equal to the field person with skill in the art. Described in example of the present invention embodiment to do not, publicly known or publicly known to operate with specific description of related subject matter of invention the breach can be decided to omit description if the analogy to each other.

Hereinafter, the present invention according to software defined network application data elements utilized to heat transmission path setting device and method for establishing a routing path to less than 1000. Wherein, as well as transmission path setting device comprises a source/destination node, including a plurality of nodes on a network can be any node. The, transmission path setting user terminal device in addition to the network device, one be a router.

Figure 2 shows a general outline of the present invention according to road also software defined transmission in a network device are disclosed.

Figure 3 the present invention according to software defined data network method in relation to the road, indicating the intensified learning basic operational procedure are disclosed.

The present invention according to the road in conjunction with software defined data network method, reinforced learning (Reinforcement Learning) transmission method based on routing can be performed. Thus, the intensified learning (Reinforcement Learning) transmission method based on the residual path setup as follows.

The system reinforced learning method belonging to one category of machine learning, applied in circumstances existing State Agent (state) to observe the behavior is able to perform one or Penalty Reward (compensation values) to maximize (190) to select the silicon minimizing behavior or behavior order are disclosed. This Markov Decision Process (MDP) model is based on Agent Action performed by a State Agent can be obtained, and to define a gradual learning procedure via interaction with Environment Action according to Reward/Penalty subjected to substrate. On the other hand, Figure 2 shows a general outline of the present invention according to software defined data network device also road degree which, in conjunction with the present invention according to the road 3 software defined data network method, indicating the intensified learning basic operational procedure are disclosed. Also 2, 3 also reference surface, said intensified learning basic operational fpr transmission path setting device (1000) is application module (100) open-flow switch (200) having a predetermined wavelength. Wherein, application module (100) and be referred to as the agent (agent), open flow switch (200) is (runtime) environment module ((Runtime) Envirionment module) can be referred to.

As shown in fig. 3, application module (100) is open flow switch (200) receives information from state (State, St), based on the specific actions (action, At) performs. This specific actions based on (action, At), open flow switch (200) has a compensation value (reward, Rt) application module (100) delivers to. This action drives based on compensation value based on the values of the plurality to each reinforced learning wants target corresponding to the substrate.

With respect to such reinforced learning, learning algorithm one SARSA (State provided Action provided Reward provided State provided Action) algorithm is carried out, and performing in each State Action, updating algorithm receives Q table according to Reward thereof are disclosed.

In this regard, the present invention according to SARSA table 1 presents a general outline the basic algorithm.

I.e., based on the compensation value based on the present invention according to specific action in reinforcing the learning process, all state (s) Q (or compensation value table) as compensation values that function on action (a) begins by with respect with each other. Then, for each episode, set of initializing (S), policy (policy) from Q using (S) selects a specific action A set of conditions. Then, by performing a particular action A, updated set of compensation values (R) (S') to obtain substrate. Next, Q that the updated set of policies stored in from (S') (A') selects from a specific action. As aforementioned, and selected action, compensation value based on the process, can be repeated until the completion of the set of conditions and for (terminal). This, each episode state required actions which have been performed about 10 obtained compensation value is converged or if one or more of specific value can be a standby state.

On the other hand, Figure 4 the present invention according to software defined network application data elements establishing a routing path utilizing components of transmission path setting device exhibits. I.e., in the present invention number-in route function exhibits Application structure data elements. As shown in the present invention transmission path 4 are symmetrically 2 also to set includes the controller (10) in a transmission path for a data-planar nodes according to deliver a defined for each State Flow Table Action Action Reward State Action of subsequent development of determining scheme is disclosed.

On the other hand, the changed SDN implementing open flow switch on monitoring function and setting all node according to collect information about the transmission path can be scheme is used. However, controller (10) when a game is started to work or managing part that is high, because node on a network database Data Plane overload occurs on each node can be distributively placing Data Plane monitoring function.

As shown in fig. 4, transmission path setting device (1000) is application module (100) open-flow switch (200) having a predetermined wavelength. Wherein, application module (100) open-flow switch (200) and each can be referred to as DP Application Environment. In particular, in a runtime environment as one example of Environment (Runtime Environment) corresponding to this can be.

In addition, application module (100) is condition monitor (S-a Monitor) module (110), compensation value calculation module (R-a Calculator) module (120) and action conductor (Action conductor) module (130) comprises. Wherein, state monitor module (110), compensation value calculation module (120) and action conductor module (130) each S Monitor, R Calculator Action conductor can be referred to.

In addition, open flow switch (200) the port state number ball module (210) and flow table module (220) comprises. Wherein, port state number ball module (210) and flow table module (220) each Physical port status (providing module) and Flow Table (module) can be referred to. On the other hand, flow table module (220) including flow is table can be also referred to as a flow switch open. In addition, corresponding to the port States Physical port status (providing module) number ball module (210) of in one example, network interface card (NIC: Network Interface Card) is can be corresponding to the data.

On the other hand, as shown in the left of Figure 4, application module (100) open-flow switch (200) and each Applicaiton OVS (Open VSwitch) can be referred to. State OpenVSwitch SR Monitor module is of one of the monitoring port through. SR Monitor Host Port Tx bytes in a number installed in the remaining ports is linked with a time t + 1, Rx bytes t Tx bytes in the time base, when (in when not in use) the same Rx bytes 0, 1 and 0 or 1 (during use) are represented when increased Output Port State after obtaining a set of defined value, upon request by monitoring action conductor (Action conductor) module (130) switch substrate. In addition, the conductor module action said compensation value (Reward) (130) (Action) measured in Port finger action when carrying out transmission delay (Transmission Delay) corresponding to the reciprocal of the value limited thereto. In addition, the compensation value (Reward) said action is an action corresponding to the port when said conductor module (Action conductor) measured in transmission delay (Transmission Delay) weights (weight) depending on product of inverse number of maximum bandwidth value can be calculated. The heat with respect to prevent under-sensors other.

With respect to the aforementioned contents, in Figure 4 of application module (100) open-flow switch (200) in terms of, residual provided as follows.

First, also 2, 4 also relative, packets are transmitted location information transmitted path number before the next residual stroke and disclosed. On the other hand, such transmission path setting order next order rather than limited to, changes in the range of the present invention pivotably.

1) Packet from the host to the open flow switch (200) when entering, open flow switch (200) is SDN controller (10) for transmitting the packets to multipath flow table request as follows.

2) Request message receiving the SDN (Packet_IN Message) the controller (10) is equal to the shortest distance can be sent the packet selecting multiple paths. The, shortest distance multi-path for flow table using a predetermined multipath simply Hop Count are disclosed. After, SDN controller (10) is the shortest distance multiple paths each open flow switch (200) switch substrate.

3) Flow table is delivered to an output each open flow switch (200) is utilized to switch transmission ports each flow table (Output port) (State) determines a current state. This, switch the current state of the state monitor (State) (S-a Monitor) module (110) which are monitored status through effected.

4) In addition, flow table for the same input port based on the compensation value table (Reward Table) transmission port information is defined. This compensation value table compensation value calculation module (R-a Calculator) module (120) compensation values that are calculated based on through its isomorphs. On the other hand, this compensation value table in one example of compensation values that are also shown in table 6 may be described subsequent, in respect of the form under heat-sensors other.

5) The, to an input port of one transmission direction among multiple port when, during each transmission port Reward value largest Reward value selected flow table port reflecting substrate. Wherein, selecting a largest Reward value corresponding flow entry port formed by corresponding to the priority of transmission ports for updating corresponding to the substrate. Such, the timer and the second timer action conductor module (130) through the combustion chamber.

6) If transmission port to only one input port if there, action conductor module (130) priority is increased to thereby update corresponding flow entry in addition a plurality of hierarchies.

On the other hand, number one first flow entry flow table is delivered to an output of an aggregate of open flow switch (200) is as above process application through a plurality of hierarchies. The, in compensation value based on the current state (Based on Reward Value), verifies the efficiency of each port, last transmission path is equal to selecting.

On the other hand, the above-described process according to transmission path setting in relation to the element carrying out based on residual more specifically as follows.

Application module (100) for data elements - the present destination (Src provided Dst) links between flow table (Flow Table) to deliver consists of advance. The, one input port (In_port) among multiple transmission direction when port (Output port), the aforementioned can be perform an operation. Wherein, whenever the packet flow table needs to searching requests which are not being ball number, the pre-received packet prior to transmission (and/or learning) and flow in response to the action, the transmits packets. The, open flow switch (200) is - origin destination (Src provided Dst) priority to each transmission path is set up consists of a flow table (Flow Table).

On the other hand, application module (100) is based on whether said flow table each compensation value (Reward) (State) (Action) consists of the largest actions to perform. In addition, application module (100) said compensation value is the largest transmits packets by applying the action, said action consists of values corresponding to said compensating value update. On the other hand, said packet edge (Edge) as well as they are transmitted at the node, any node the engine transmitted from the entire network are disclosed. In addition, an origin node to a destination node on a path between said packet with said destination through intermediate nodes can be up to flow table.

State monitor module (110) receiving a number of input ports (In_port) increased number of bytes associated with the transmission of the remaining port /monitors a status of whether to consists of. Specifically, state monitor module (110) is installed hardware with the OpenVSwitch of Ethernet port monitoring (Ethernet Port) can be. In addition, the monitoring of which is connected to a continuous, open flow switch (200) can be structured such that only if Packet_IN Message in receiving. I.e., forwarding the information about each OpenVSwitch until the action conductor module (130) and the operated by request, transmitting port is determined and gets into a standby state. This by monitoring resources at each node is represented Data Plane for reducing overall by load are disclosed.

In this regard, controller (10) packets from the host is open flow switch (200) is introduced, said open flow switch (200) for transmitting said packet from said flow tables of request message (Packet_IN Message) receives a. In addition, controller (10) is the shortest distance from the serving said receives said packets can be multiple paths can be selected. The, shortest distance multi-path for flow table using a predetermined multipath simply Hop Count are disclosed. After, controller (10) is the shortest distance multiple paths each open flow switch (200) switch substrate.

Said open flow switch (200) is, each flow table to determine a current state of said switch (State) transmission port can be utilized. In addition, said application module (100) through said flow compensation value calculation module (R-a Calculator, 130) for the same input port transmission port information table based on the compensation value table (Reward Table) can produce. In one example, compensation value calculation module (130) of the respective transmission/receiving bytes corresponding to the calculated rate, can be configured to monitor the associated with increased rate of whether said state. The, compensation value calculation module (130) is received and port information and rate information based on compensation value table (Reward Table) can produce.

In this regard, flow table module (220) is OpenVSwitch (State) corresponding to the state of state monitor module (110) port through one of the monitoring. State monitor module (110) of 1 in expressions (1), (2) to the input port through a number Tx bytes remaining ports in time t and t + 1 (Input Port), monitors a status of Rx bytes associated with the action conductor module (130) delivers to.

Wherein, T is on between time t-a 1 packet reception quantity and corresponds to, T t-a 1 packet transmission amount on time is between corresponding to the substrate. In addition, and T is time on each packet corresponding to a reception quantity t-a 1 in 2000. In addition, and T is time on each packet transmission amount t-a 1 corresponding to the substrate.

Specifically, Tx bytes in time t and t + 1, when the same Rx bytes 0 (in when not in use), when increased 1 (during use) can be represented. I.e., the number of bytes received transmission/receiving/transmission time t or t + 1 to the number of previously accumulated bytes, transmission/receiving bytes when changes in sense that it can be equal to the time t t + 1 so as to sense that it can be received/newly transmission are disclosed. On the other hand, state monitor module (110) receiving said time t + 1 is the number of bytes transmitted/transmission/receiving bytes when increasing the number of said time t, 1 said corresponding state defined as follows. I.e., the number of bytes received transmission/receiving/transmission time t or t + 1 to the number of previously accumulated bytes, transmission/receiving time t t + 1 to increase the number of bytes in the meanings that changes received/transmitted when that data is present novel means are disclosed. The, output port 0 or 1 value (Output Port) are defined by a set of state (State) after obtaining, upon request by monitoring action conductor module (130) switch substrate.

As such, the data newly transmission/presence or other criteria defined according to whether the various available, whether such method defining limited but freely deformable disclosed.

In addition, action conductor module (Action conductor, 130) whether associated with the increased number of transmission/receiving bytes associated state (State) receive the largest each compensation value (Reward) (Action) consists of actions to perform. In addition, action conductor module (130) whether said transmission/receiving the increased number of bytes and whether said each receiving the associated with the increased rate of the largest compensation values each associated state (State) (Action) can be configured to perform actions. The, and whether said transmission/receiving said increased and determines the number of bytes each increased rate of each state monitor module (110) and compensation value calculation module (120) from conditions associated with the monitored actions conductor module (130) can be delivered.

Wherein, the compensation value (Reward) said action is an action corresponding to the port when said conductor module (Action conductor, 120) measured in transmission delay (Transmission Delay) can be calculated as the reciprocal of the value.

In this regard, the present invention according to Application OpenVSwitch (open flow switch) is also Figure 5 shows a (or application module) which is obtained from the surface each state are disclosed. For example, as shown in fig. 5, s (1) to s (8) with respect to the States of 8, Application OpenVSwitch (or application module (100)) (open flow switch (200)) is obtained from the indicating each state (State) are disclosed.

On the other hand, action conductor module (130) in a defined state prior (State), action (Action), compensation value (Reward) on combined with the State Action Reward table for a natural number. When the aforementioned SARSA algorithm is also associated 3, Pub end point after learning table corresponding Q, and policy (Policy) use other. However, in such a way that the power-in port is also 4 and in Figure 5 number defined by the thread number because the State State Action even from such as Reward is s402. An adhesive spacer remains efficient Action through select the Q table updating process from to to the error.

Specifically, this algorithm is an algorithm to locate OpenVSwitch in transmission ports are disclosed. In addition, other existing reinforced learning algorithm that provides, ε-a greedy point of view that learning resources are disclosed. In such a way that state-in number in each port controller (State) defined by the thread number because such state (State) even behavior (Action) (Reward) from each Packet_IN s402. for each compensation value. An adhesive spacer is disposed between the stator (State) state of packets on behavior (Action) then sense through transmission ports can be applied to other adjusting screw.

In this regard, table 2 and in Figure 5 number 4 is also applied along the manner in which the algorithm pseudo code indicating into (pseudo code) are disclosed.

I.e., action conductor module (130) in the Q (s, a) said compensation value in accordance with a value table, packet (s) (a) inlet is detected the state from action number 1 is a positive integer, (a) said number 1 action performed, said compensation value (r) observing the substrate. In addition, action conductor module (130) from the state (s) Q (s, a) selecting said compensation value table number 2 action (a ') Q (s, a') to update the other. In addition, action conductor module (130) associated with the compensation value tables (s) state said compensation value table (reward table) until completed (complete) observed and said compensation value table updating said compensation value repeatedly transmitting the other. the repeated actions through said compensation value by selecting the largest observation compensation value, determining path can be performing (Path decision). I.e., action conductor module (130) said ports and said controller based on the received flow table, said each said compensation value (Reward) (State) whether the largest (Action) can be configured to perform actions. The, said controller state are defined by said ports, said ports depending on the compensation values from said state an action for the controller can be. On the other hand, input port if there only one transmission direction port, a priority level of the message can be performing height corresponding flow entries.

In the present invention number Application into a specific scenario is a through packets are transmitted as follows. By a message to each OpenVSwitch Packet_IN calculated transfer paths has transferred. The, flow entry (Flow Entry) Output In port shape of the port on the link having one of paired receive a flow entry is equal to total 2. The, efficient transmission ports to locate each compensation value table (Reward table) based on port information on each match (Match) In field cause the production of a with each other. Open flow switch (200) for calculating a transmission hop destination links be - in the contents of each transmission path is provided on the OpenVSwitch by differently setting the priorities. Wherein, in conjunction with the contents of said origin - be a transmission hop destination, can be day [khu the [su sprout (Dijkstra) algorithm is employed. I.e., the [su sprout algorithm given on day [khu direction weighted graph (weighted graph) programmed inputs G s subjected. Graph G V all subset of unit length, on the order of arrival point pair (u, v) programmed with side graph u v represented in a substrate. G E subset of all sides of unit length, sides of the weight function w: [0, ∞] E→represented in a substrate. The weight w (u, v) is used in moving into the costly u v (time, distance recorded) is disclosed. Path cost is the sum of the path between the weighting of all sides are disclosed. T the [su sprout algorithm is certainly act point on any day [khu V s s then go to t when in pair comprises a cost (shortest path) can be utilized to find a path.

On the other hand, table 3 below for illustrative of one OpenVSwitch origin - Flow Entry Table exhibits disposed between destinations.

I.e., said flow table, flow entry (Flow Entry), and action (Action) field match field (Match Field) can be. Wherein, flow entry consists of a plurality of flows to each other. In addition, match field said flow entries, each transmission path said priority, input port (in_port), MAC address of the destination IP address on said links and said (src_mac, dst_mac) (dst_ip, src_ip) without using a tool. In addition, action field said flow entries, corresponding to said input port (in_port) output port (output port) without using a tool.

On the other hand, action conductor module (130) input port of said each flow entry in the flow table (In port) extracting information, said input port corresponding to the at least one output port (output port) based on said compensation value calculation module available bandwidth (120) can be configured to determine said compensation value applied from the last action.

Specifically, action conductor module (130) disposed in flow table In port information extracted in each flow table. This behavior based on port information (Action) In Rule Reward Table process performed for making are disclosed. Specifically, the reference table 3, in_port=1, in_port=2, in_port=3, for Table in_port=4 is made. The, for the output of the behaviors (Action) Rule Table in_port=1:2, output: 3 and output: 4 (State) because a current state for each Output port compensation value calculation module based on available widths in in_port=1 (R Calculator) (120) (Reward) (Action) transmitted to the compensation value from the last behavior to determine the other. In addition, in_port=2, in_port=3, for in the case of Table in_port=4, output: 1 corresponding to one behavior (Action) corresponding Flow Entry nickel only to select the other. In each Reward Table action the velocity compensation value (Reward) value (Action), application module (110) having a high priority of corresponding behavior (Action) Flow Entry update can be applied. In addition, in the case of applying each Entry Timeout time after time to utilize corresponding Entry if not said Timeout can be back number. Wherein, said Timeout time is 5 S. can be determined.

On the other hand, in the embodiment according to each input port in one of the present invention also Figure 6 shows a compensation value table etched upon. 6 Also are shown as aforementioned, each input port and at least one output port, e.g., an action for the rule in_port=1 the plurality of output port (output: 2, output: 3 and output: 4) respect to each other can be defined differently. I.e., each behavior (Action) to the motor (Reward) based finally flow entry (Flow Entry) is determined for a given substrate. The final flow of the Entry Data Plane OpenVSwitch update an entry in the corresponding paths are made based on the packet transmitted.

In_port=2, in_port=3, in the case of a determined through Reward Table of Flow Entry in_port=4, upholstery or number utilized according packet transmission from neighboring nodes with each other. Wherein the meanings is 2 times if session from neighboring nodes communicating with a flow entry (Flow Entry) field (Match Field) cover of fingerprints and in_port=4 in_port=3 is set back by a number Timeout Match field and transferred to the destination packet having flow entry only utilized in in_port=2 to be coated. The scheme is open flow switch at a predetermined destination number based multi paths through a departure Hop OpenVSwitch Application of current state (State) most efficient in a transmission is equal to man decide for himself. The SDN environment since the largest number sub-in resource and more flexible than transmission path setting method utilizing Data Plane impeded by search are disclosed.

In the case of the aforementioned compensation value (Reward), each flow port entries based on information extracted In Output port. In addition, when performing each Output port to the motor (Reward), corresponding to the corresponding port available bandwidth where action conductor module (130) delivers to. 1 Angles of the aforementioned (1) and (2) of whether packet through the Ethernet port, i.e. can be checking usage of a port. The, action conductor module (130) request by transmitting bytes (Bytes) SMonitor module monitoring substrate. The bytes transmitted and received increased current port is used status (State) determination if judged that a reflecting substrate. In addition, each Rate range is increased transmission delay calculation calculates a compensation value utilizing available mixers (Reward).

In this regard, 2 of expressions (3) to (5) is each Reward Table Application (State) whether the action is for performing compensation value when (Action) (Reward) value for calculating equation are disclosed. The respective port maximum bandwidth-available currently available have their widths can be based on transmission delay values to be coated. 2 Of expressions (3) and (4) product 8 in main reason bandwidth units by a further value for Bit Rate Bit per Second Byte thus making a wiring unit are disclosed.

Wherein, and T exhibits maximum bandwidth available bandwidth port port each time. In addition, and T time delay of the transmission of each packet (Packet) (Size) size of inlet port value exhibits. In addition, T is time compensation value corresponding port by a goniophotometer.

Equation (3) and (4) whether the action is for the Application (State) when carrying out compensation value of each Reward Table (Action) (Reward) value for calculating equation are disclosed. The respective port maximum bandwidth-available currently available have their widths can be based on transmission delay values to be coated. Equation (3) and (4) product 8 in main reason bandwidth units by a further value for Bit Rate Bit per Second Byte thus making a wiring unit are disclosed.

(Reward) value is defined by the inverse number of transmission delay compensation value which can be the same or different super-when each port available by performing packet transmission can be achieved. In addition in the case of weight (Weight), maximum bandwidth determined value, compensation values that is preventing the level and properly (Reward) is multiplied in order to compare each port's desire. The molten obtained transmission delay value corresponding to ports perform Output Application based on each behavior (Action), compensation value (Reward) value for selecting the largest Port packet memory to be coated.

In this regard, according to a mathematical equation 2, compensation value (Reward) this action conductor module (Action conductor, 120) when carrying out this action (Action), as shown by the inverse number of transmission delay which is measured on the corresponding port (Port) can be know. On the other hand, application module (100) is, said compensation value the largest transmits packets by applying the action, can be configured to update said compensation values corresponding to said action. I.e., transmission delay which is measured on the corresponding port packet transmission is minimized along path is compensated by configured to select values can be.

One in the embodiment according to Figure 7 of the present invention, based on the flow table OpenVSwitch Packet_IN message controller exhibits high-frequency waves. Controller a predetermined multipath regard, total 5 two paths can be generated. I.e., SW1 - SW2 - SW3 - SW4 on SW1 - SW4 - SW5 - SW6 B A through path through the path, path through the SW1 - SW5 - SW6 C, D path through the SW1 - SW2 - SW6, finally SW1 - SW2 - SW5 - SW6 of path is produced through the path to the total 5 E for the routing Flow Table OpenVSwitch transferred to on. The controller after selecting multiple paths arranged in each Entry destination links by SW1, SW2, SW5, SW6 In port of one streams of Action Rule is generated. Transmitting a packet coming from current state of OpenVSwitch (State) can be performed to action (Action) for transmitting a packet of Output Port Entry based on files based on which each derived (Reward) is equal to value.

Each Flow Entry etc. has a different priority. This same Match field in various field is generated to have final Action Action serve as priority higher than other Entry Match In utilizing the temperature of the port when the packet field set to be coated.

On the other hand, one in the embodiment according to Figure 8 of the present invention, exhibits transmission path is shape. As shown in fig. 8, each done in SW1 and SW2 during compensation value (Reward) value for selecting the largest Flow Entry behavior (Action) are disclosed. In addition, in the case of setting the number and back surfaces are not utilized in each OpenVSwitch Flow Entry Timeout Flow Entry flange supports the current packet transmission only by remaining disclosed. After a routing, packet transmitting upon another Host, wants to be formed with status (State) can be carried out by obtaining a compensation value according to a predetermined path behavior (Action) (Reward) value are disclosed. In the present invention most efficient packet transmission route setup OpenVSwitch Data Plane without intervention of a current state of each port open flow switch further checks the transmission link spectrum at a predetermined substrate. The, controller at a predetermined transmission without further intervention techniques since Self Routing Organization specify link can be provided.

Or more data elements in the present invention according to software defined network application utilizing transmission path setting device for establishing a routing path for flaws. Below, according to another aspect of the present invention software defined network application data elements utilized to heat transmission path setting method for establishing a routing path to less than 1000. On the other hand, the aforementioned setting device setting a menu to a transmission path of a transmission method can be utilized even negative disclosed.

In this regard, the present invention according to software defined network application data elements also Figure 9 shows a transmission path setting method of establishing a routing path utilizing etched during flow. Transmission path setting method performed by setting a transmission path device and, in particular, application module (100) and the open flow switch (200) can be performed by at least one.

As shown in fig. 9, transmission path setting method monitoring state (S100), rate (S150) monitoring and action steps (S200) without using a tool. On the other hand, action steps (S200) is number 1 (S210) action selection step, step (S220) observation compensation value, and number 2/compensation value table updating step (S230) comprising the selected action. In addition, action steps (S200) has a compensation value table (S240) and path decision step further includes a step for determining the complete (S250) can be.

In monitoring state (S100), the transmission of a remaining port number input port (In_port) associated with increased number of received bytes/ whether monitors a status of other.

Rate (S150) in monitoring, said transmission/receiving bytes of the respective rate corresponding to the calculated, whether associated with the increased rate of said monitoring substrate.

On the other hand, in monitoring state (S100), the number of bytes transmitted/received time t t + 1 time equal to the transmission/receiving bytes 0 said corresponding state defined as follows. In addition, the number of bytes transmitted/received said time t + 1 number of bytes when increasing the transmission/receiving said time t, said corresponding state 1 through an, said ports monitors the status of the other.

In action steps (S200), whether associated with the increased number of transmission/receiving bytes associated state (State) receive the largest each compensation value (Reward) actions (Action) performs.

Specifically, number 1 in action selection step (S210), said compensation value in accordance with a value table in Q (s, a), packet (s) (a) optionally selects from inlet is detected the state number 1 action.

In addition, compensation values observed in step (S220), (a) said number 1 action performed, said compensation value (r) observing the substrate. The, number 2 in step (S230)/compensation value table updating selected action, said state (s) Q (s, a) selecting from said compensation value table number 2 action (a ') Q (s, a') to update the other.

Next, in step for determining the complete compensation value table (S240), said compensation value tables said state (s) associated with the compensation value table (reward table) finished (complete) determining whether or not the substrate. The, said compensation value table is completed (S220) until said selected compensation value table updating step (S230)/compensation value observation step and said number 2 action repeatedly transmitting the other.

Said state (s) associated with the compensation value table (reward table) is completed when determines, in decision step (S250) path, said repeated compensation value by selecting the largest observation through actions said compensation value, determining path can be performing (Path decision).

According to the present invention, path setting number must perform other functions in the present invention number through the network on the path of the discharge controller can be-in manner. In addition, even large networks including many nodes can be Flow Table request according to color.

With respect to the expected effect of the present invention, loose so as to transmission path setting function in the present invention number it assists the controller must reduce a load controller may perform other functions can be smoothly number environment.

In relation to the business perspective of the present invention, if applied in the present invention number networks into way software defined developed load controller network environment with respect to a smoother users by reducing the number ball subjected hif2e..

In the embodiment according to the method of the present invention setting at least one transmission path, for transmitting packet in software defined network was responsible only for data elements transmission path updating function such as Application responds to a request by treating, controller capable of reducing a load is concentrated on the pin is necessary.

In addition, in the embodiment according to the method of the present invention setting at least one transmission path, for transmitting packet in software defined network was responsible only for data elements transmission path updating function such as Application responds to a request by treating, load is concentrated on a controller according to reduce delay of the transmission of user data is an tranfectants disclosed.

The software implementation, the specification described procedure and function as well as sensing device can be implemented as software modules each component. Said software module performs an operation for each of the one or more function and can be described specification. Suitable program language written into its software application software code can be implemented. Said software code is stored in a memory, number control unit (controller) or processor (processor) can be executed by (or application).