JP2005159551A - Network attack countermeasuring method, network apparatus thereof, and program thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a network attack countermeasuring method or the like capable of detecting an attack without the need for advance settings of flow information or the like and attaining proper countermeasures against a network attack without the need for advance settings. <P>SOLUTION: A node 8 periodically measures a packet quantity distribution with respect to a plurality of attribute types of packets transmitted from terminals 1 to 4, records the packet quantity distribution with respect to a plurality of the attribute types of the measured packets as a traffic profile (traffic profile DB), and a traffic profile extract section automatically extracts a feature quantity of the packet quantity distribution with respect to a plurality of the attribute types in a normal state from the recorded traffic profile. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a network attack countermeasure method and a network apparatus thereof that prevent a network attack that disables the network itself or a specific server by sending a large amount of packets in an open network environment such as the Internet. , As well as the program.

Wide-area networks such as the Internet are open environments, and anyone can connect freely. For this reason, an attacker can easily cause an attack that disables a specific server or network by sending a large amount of packets using such an environment.
For network attacks, the network cannot be used by sending a large number of packets with unspecified destinations, and "fixed destination network attack" that disables the server by sending packets to a specific server. There is a “destination ad hoc network attack”. As examples of fixed destination network attacks, DoS attacks and DDoS attacks are known. Also, as an example of an unspecified destination network attack, an attack that wastes network bandwidth by self-propagation by a worm is known.

By the way, many methods for taking countermeasures against attack packets when a network attack occurs have been proposed and filed. For example, “Distributed denial of service attack prevention method and gate device, communication device and program” (Japanese Patent Application No. 2002-81901), “Attack packet countermeasure system, attack packet countermeasure method, attack” filed by the inventors of the present application, etc. "Packet countermeasure program and recording medium" (Japanese Patent Application No. 2003-151578).
In the former, when a node near the defense target detects an abnormality in the packet amount for each flow, it detects this as an attack, and at the same time, transmits abnormal flow information to the nodes on the attack path. This is a proposal relating to a technique for limiting the bandwidth of a packet matching this information to a predetermined amount or less (primary countermeasure). Also, each node measures the abnormal flow bandwidth for each source address range, and if an illegal pattern such as continuity over a long period of time is found in those nodes, the abnormal flow bandwidth from that range is determined. Limit to a smaller amount (secondary measure).

  The latter analyzes the packet during the attack, identifies the attack packet by source address, improves the identification accuracy, implements an effective countermeasure against the attack, and reduces the impact on the packet of the legitimate user. This is a technology that can be reduced. In addition, by analyzing only packets from a source address with a large number of transfer sheets, analysis processing can be performed using a limited memory.

Many methods for taking measures against attack packets when a network attack occurs are disclosed not only in the above-mentioned application but also in the literature.
For example, a nearby node in the defense target, such as a server, monitors the queue for each interface, detects when a large number of queues overflow, detects this as an attack, and after detection detects the destination address of the packet that is input to the queue By analyzing, the destination network address of the attack packet is extracted, and this is used as attack packet information. At the same time, a technique is disclosed in which this information is transmitted to nodes on the attack path and the bandwidth of the packet corresponding to this information is limited by these nodes (see, for example, Non-Patent Document 1).

Also, statistical analysis is performed to determine the attribute values of the attack packet, packets with those attribute values are blocked at the node, and at the same time, this information is transmitted to other nodes distributed and arranged on the network. A technique for blocking a packet corresponding to this information at the node is also disclosed (for example, see Non-Patent Document 2).
ACC (Aggregate-based Congestion Control) (R. Mahajan, S. Bellovin, S. Floyd, J. Ioan-nidis, V. Paxson and S. Shcnker: “Controlling High Bandwidth Aggregates in the Net-work”, SIGCOMM Computer Communication Review, Vol 32, No.3, p.62-73, Jul. 2002.) "Performance evaluation of Distributed Active Firewall against SQL Slammer worm virus" (Makoto Sugita, Masaru Katayama, Kohei Shiomoto, Naoaki Yamanaka) Information Processing Society of Japan Research Report, 2003-CSEC-22 Vol.2003, No74, p105-112

However, the above-described conventional techniques have the following drawbacks. That is, according to the technology disclosed in Japanese Patent Application Nos. 2002-81901 and 2003-151578, it is necessary to designate a server to be protected in advance and input flow information and abnormal conditions for the server to the system. There is. Here, “flow” refers to a group of packets having the same attributes, and the attributes include a source / destination address, a source / destination port number, a protocol number, and the like.
For example, when the server S is a protection target, information such as flow 1: {destination address = S, protocol = TCP}, abnormal condition 1: {packets of 100 Mbps or more are continuously transferred for 30 seconds or more}, and the like. Set. Such information needs to be derived by monitoring the network for a certain period of time by an administrator, and is not only labor intensive and time consuming because of the manual intervention, which increases the possibility of improper settings.
In addition, this method is a countermeasure against a fixed destination network attack, but is not a countermeasure against a destination indefinite network attack. Furthermore, it is not a countermeasure against an attack using a packet deviating from the flow information (in the above example, an attack using the UDP protocol). In this method, the bandwidth is limited for abnormal flows, but the limit value is manually set by the administrator. Therefore, the administrator must determine how to determine this value by monitoring the network for a certain period of time. In addition, since manual intervention is required, it takes time and effort, and the possibility of improper setting increases. As described above, both of the two cases are problematic in that the setting takes time and time, and there is a high possibility of improper setting, and the attacks targeted for countermeasures are limited.

On the other hand, the technique disclosed in Non-Patent Document 1 is also a countermeasure against a fixed destination network attack as described above, but is not a countermeasure against a destination indefinite network attack. Moreover, since only the destination address is used as the identifier of the attack packet, the identification accuracy is low. For this reason, there is a high possibility that the packet of the legitimate user is included in the packet to be identified and the communication of the legitimate user is restricted, and the influence on the legitimate user by the countermeasure is great.
Furthermore, as described above, the bandwidth is limited for abnormal flows, but the limit value is manually set by the administrator. Therefore, it is necessary for the administrator to determine how to determine this value by monitoring the network for a certain period of time. In addition, since human intervention is involved, not only is it time-consuming and time-consuming, but also the possibility of improper settings increases. In this way, countermeasure attacks are limited, attack packet identification accuracy is low, countermeasures have a large impact on legitimate users, and it may take time and effort to make settings incorrectly. High points are each a problem.

  Further, according to the technology disclosed in Non-Patent Document 2, since the attack packet is blocked by the network device, when the identifier of the attack packet includes even a legitimate user packet, the communication of the legitimate user is performed. Will be shut off. This is a problem because the influence on the regular user by the countermeasure is further increased as compared with the techniques disclosed in Japanese Patent Application Nos. 2002-81901 and 2003-151578.

The present invention has been made in view of the above circumstances, and records a packet amount distribution related to one or a plurality of attribute types of a packet as a traffic profile, automatically extracts a characteristic value of the packet amount distribution at normal time, and A network attack countermeasure method, its network device, and its program that can detect an attack without prior setting of flow information etc. by periodically measuring the distribution and detecting as an attack when there is a deviation from the traffic profile The purpose is to provide.
In addition, by automatically extracting multiple abnormal attribute values with divergence as attack packet identifiers at this time, it is possible to identify attack packets with high accuracy, and automatically according to the packet distribution before the occurrence of an abnormality It is another object of the present invention to provide a network attack countermeasure method, a network device thereof, and a program thereof that enable appropriate countermeasures against an attack without prior setting by restricting the bandwidth of attack packets.

  In order to solve the above-described problems, the present invention provides a network environment in which a plurality of terminals are connected, and when an attack that makes a specific server or network unusable due to a large amount of packets being sent occurs. A network attack countermeasure method for preventing, comprising: periodically measuring a packet amount distribution regarding a plurality of attribute types of a packet transmitted from the terminal; and a packet amount distribution regarding a plurality of attribute types of the measured packet. A step of recording as a traffic profile; and a step of automatically extracting feature values of a packet amount distribution relating to a plurality of attribute types at normal times from the recorded traffic profile.

  In the present invention, the attribute type is any one or more of a destination address, a destination port number, a protocol number, and a packet length.

  In order to solve the above-described problems, the present invention provides a network environment in which a plurality of terminals are connected, and when an attack that makes a specific server or network unusable due to a large amount of packets being sent occurs. A network device for protection, a packet amount distribution measuring unit that periodically measures a packet amount distribution related to one or more attribute types of packets transmitted from the terminal, and one or more of the measured packets A traffic profile recording unit that records a packet amount distribution related to the attribute type as a traffic profile, and a traffic profile extraction unit that automatically extracts feature values of packet amount distributions related to a plurality of attribute types at normal times from the recorded traffic profile; It is provided with.

  Further, in the present invention, the attack is detected from the difference between the packet amount distribution regarding the plurality of attribute types periodically measured and the traffic profile at the normal time, and the attack packet is detailed when the attack is detected. And an attack packet feature value extraction unit for automatically extracting feature values.

  The present invention further includes a bandwidth limit value calculation unit that calculates a limit value based on the normal traffic profile when performing bandwidth limitation on the attack packet after the attack detection. .

  Here, every time a packet is input, the network device extracts an attribute value given to the packet. The attribute value here refers to an IP packet value such as a destination address, a destination port, a protocol number, and a packet length. For each attribute type, the network device measures a distribution for each attribute value of “amount of packets input within a unit time (hereinafter referred to as a unit time packet amount)” for each “measurement period” set in advance. Record as a traffic profile. When the measurement data is accumulated for the preset “measurement period”, the network device calculates, for example, the average value, maximum value, and standard deviation of the period and records them in the traffic profile. Update the value. As a result, the traffic profile represents the distribution of packet amount by attribute value at the normal time.

At the time of measurement, the alert mode is entered when the measurement data is significantly different from the packet amount distribution of the traffic profile. For example, in a certain attribute type, when the measurement data of a specific attribute value satisfies the condition that the maximum value in the measurement period exceeds N1 times or exceeds the average value N2 × standard deviation in the measurement period, the alert mode to go into. Further, when this alert mode is continuously satisfied C times or more for the same attribute value, it is detected as an attack. These values of Nx and C represent “sensitivity” of attack determination and can be changed by setting.
According to the present invention, the traffic profile representing the packet characteristic value at the normal time is automatically updated, and the content and the packet amount distribution for each measurement period are compared in real time without setting the abnormal condition in advance. Enables attack detection. In addition, when an attack is detected, a plurality of abnormal attribute values are extracted from a plurality of attribute types in the traffic profile, so that an accurate attack packet identifier can be extracted. Furthermore, according to the present invention, the network device limits the bandwidth of the attack packet and automatically calculates the limit value based on the normal traffic profile, thereby affecting the packet of a legitimate user who does not have an abnormal identifier. Can be lowered.

  In order to solve the above-described problems, the present invention provides a network environment in which a plurality of terminals are connected, and when an attack that makes a specific server or network unusable due to a large amount of packets being sent occurs. A program used for a network device to be protected, which periodically measures a packet amount distribution related to a plurality of attribute types of a packet transmitted from the terminal, and a packet amount related to a plurality of attribute types of the measured packet It is characterized by causing a computer to execute a process of recording a distribution as a traffic profile and a process of automatically extracting feature values of packet amount distributions related to a plurality of attribute types at normal times from the recorded traffic profile.

According to the present invention, the network device (each node on the WAN) automatically updates the traffic profile representing the packet characteristic value at the normal time, and compares it with the packet amount distribution for each measurement period in real time. It is possible to detect attacks without setting abnormal conditions in advance. Further, when an attack is detected, a plurality of abnormal attribute values are extracted from a plurality of attribute types in the traffic profile, so that it is possible to extract an attack packet identifier with high accuracy.
In addition, the network device limits the bandwidth of attack packets and automatically calculates the limit based on the normal traffic profile, thereby reducing the impact on packets of legitimate users who do not have an abnormal identifier. it can. This makes it possible to take appropriate countermeasures against network attacks without manual intervention, and keep the entire network in a safe state.

The present invention will be described below with reference to the drawings. FIG. 1 is a diagram quoted for explaining a network environment to which a network device of the present invention is connected.
In FIG. 1, reference numerals 1 to 4 denote terminals such as PCs or servers. Reference numeral 9 denotes a WAN (Wide Area Network) to which the terminals 1 to 4 are connected. The WAN 9 has a plurality of nodes 5 to 8 on which the network device of the present invention is mounted.

  A network attack occurs when a terminal is infected with a virus or worm, or control is taken over by an attacker. For example, when the terminals 1 to 3 take over control by an attacker and a large amount of attack packets are transmitted toward the terminal 4, a fixed destination network attack occurs. Further, when the terminal 4 takes over control by the worm and transmits a self-propagating packet having an arbitrary destination address including the terminals 1 to 3, destination indefinite or a network attack occurs. When these attacks occur, the bandwidth and resources of the nodes 5 to 8 and the WAN 9 are wasted, and a phenomenon that normal operation cannot be performed occurs.

Hereinafter, the network attack countermeasure of the present invention will be described focusing on the node 8 in FIG.
FIG. 2 is a block diagram showing an expanded function of the internal configuration of the node 8 on which the network device of the present invention is mounted. As shown in FIG. 2, the node 8 includes a packet amount distribution measuring unit 81, a traffic profile DB 82, a traffic profile extracting unit 83, an attack detecting unit 84, an attack packet feature value extracting unit 85, and a bandwidth limit value. The calculation unit 86 is configured.

The packet amount distribution measuring unit 81 has a function of periodically measuring a packet amount distribution related to one or a plurality of attribute types of packets transmitted from the terminals 1 to 4, and a traffic profile DB 82 as a traffic profile recording unit The packet amount distribution relating to a plurality of attribute types of the packet measured by the packet amount distribution measuring unit 81 is recorded as a traffic profile.
In addition, the traffic profile extraction unit 83 has a function of automatically extracting feature values of a packet amount distribution related to a plurality of attribute types at normal times from the traffic profile recorded in the traffic profile DB 82.

The attack detection unit 84 has a function of detecting a network attack based on a difference between a packet amount distribution related to a plurality of attribute types periodically measured and a traffic profile in a normal state. The attack packet feature value extraction unit 85 When the detection unit 84 detects a network attack, it has a function of automatically extracting detailed feature values of the attack packet.
Further, the bandwidth limit value calculation unit 86 has a function of calculating the limit value based on a normal traffic profile when performing bandwidth limitation on an attack packet after detecting a network attack.

FIG. 3 shows an example of the data structure of the traffic profile 20 recorded in the traffic profile DB 82. Since the traffic profile 20 is composed of a traffic table of each attribute type attached to an IP packet, such as a destination address, a destination boat, a protocol number, and a packet length, it will be hereinafter referred to as a multidimensional traffic profile. .
The destination boat traffic table 21 stores a packet amount distribution for each destination port number in the multidimensional traffic profile. The packet length traffic table 22 stores a packet amount distribution for each packet length in the multi-dimensional traffic profile. Each table includes a packet amount, a packet amount per unit time, an average value, a maximum value, and a standard value. It consists of each field of deviation and warning mode flag. In the destination port traffic table 21, for example, an attribute value 211, a packet amount 212, a unit time packet amount 213, an average value 214, a maximum value 215, a standard deviation 216, and a warning mode flag 217 indicate each field.

FIGS. 4 to 6 are flowcharts cited for explaining the operation of the network device of the present invention, and partially show the processing procedure of the program of the present invention.
The operation of the network device of the present invention shown in FIGS. 2 and 3 will be described in detail below with reference to the flowcharts shown in FIGS.

FIG. 4 shows a processing procedure of each data of the multidimensional traffic profile 20 at the node 8. When the node 8 receives the packet, the node 8 extracts the attribute value of the attribute type described in the “measurement attribute type” (not shown) set in advance and executes the subsequent processing.
First, the node 8 selects an attribute type from preset measurement attribute types (S41). If all attribute types have been selected, the process ends.
If the selection is successful in step S41, the node 8 extracts an attribute value (S42), and checks whether or not a record corresponding to the attribute value already exists in the corresponding traffic table (S43). . If it does not exist, try to create a record. Whether or not a record can be created is determined based on the amount of memory that the node 8 can use at that time. If record creation is possible (S46 Yes), a record is created (S47), and 1 is added to the packet amount of the created record (S48). If it is confirmed in step S43 that a record corresponding to the attribute value already exists, 1 is added to the packet amount of the corresponding record in step S44 (S44). Next, 1 is added to the packet amount of the total record (S45), and the process returns to step S41. The processes of steps S41 to S45 are executed for the items set in the measurement attribute type every time a packet is received.

FIG. 5 shows the processing method of each value of the unit time packet amount 213, the average value 214, the maximum value 215, the standard deviation 216, and the alert mode flag 217, and the attack detection determination procedure in the multi-dimensional traffic profile of the node 8. Is.
The node 8 executes the following processing for the number of attribute types described in the measurement attribute type every preset measurement interval (Dc), for example, 10 s. That is, first, an attribute type is selected from the measurement attribute types (S51). If all the attribute types have been selected, the process proceeds to step S56. When the selection is successful, the unit time packet amount 213 is calculated by the following arithmetic expression (1).
Unit time packet amount (213) = packet amount (212) / Dc (1)

  In step S52, the unit time packet amount 213 calculated as described above is recorded as the past Do history. These values are hereinafter referred to as p (0), p (1), ..., p (Do-1). Here, the value of Do represents the measurement period and is set by the administrator, and p (0) is the latest unit time packet amount (213) calculated by equation (1). . In step S53, it is determined whether or not the unit time packet amount information is recorded as a Do history, and if not recorded, it is determined that the traffic profile is not sufficiently configured, and in step S51. Proceed to processing.

In step S54, the alert mode flag (217) of each record is set to “ON” if the latest unit time packet amount po greatly exceeds the distribution of the multidimensional traffic profile, and is set to “OFF” otherwise. Set. For example, the condition that p (0) of a specific record exceeds the maximum value (215) in the measurement period by N1 times, or p (0) exceeds the average value (213) + N2 × variance value (215). Set to “ON” when the condition is satisfied.
Next, in step S55, when there is a record set to "ON" in the alert mode flag (217), the attribute value of the record is registered as "abnormal attribute value". When the warning mode flag (217) of a plurality of records is “ON”, the logical sum (OR) condition of these attribute values is set as an abnormal attribute value. For example, when the record of the attribute value “X1” and the warning mode flag of the attribute value “X2” are set to “ON” in the destination port traffic table, the abnormal attribute value {destination port = X1 OR destination port = X2} is obtained.

In step S55, if an abnormal attribute value is already registered with another attribute type, a new abnormal attribute value is added under a logical product (AND) condition. For example, in the above example, the abnormal attribute value {destination port = X1 OR destination port = X2} has already been registered, and the warning mode flag of the record having the attribute value “Y” in the packet length traffic table is also “ON”. Is set, the abnormal attribute value {destination port = X1 OR destination port = X2} AND packet length = Y.
On the other hand, in step S56, it is determined whether or not there is a registration in the abnormal attribute value. If there is registration, the process proceeds to step S61. In step S56, the average value (214) is updated by the following arithmetic expression (2).

In step S58, the maximum value (215) is updated by the following arithmetic expression.
Maximum value (215) = max (maximum value (215), unit time packet maximum (213)) (3)
Here, max (A, B) is a function that compares A and B and returns a large value.
Further, in step S59, the standard deviation (216) is updated by obtaining the following arithmetic expression (4).

In step S61, the previously registered abnormal attribute value “warning attribute value” is compared with the current abnormal attribute value, and if they match, it is determined that the attack is the same, and the warning counter (ctr) is set to 1. Increase (S62). If they do not coincide with each other, it is determined that the attack is a new attack, and 1 is assigned to "warning counter (ctr)" (S63). In step S64, the abnormal attribute value is copied to the warning attribute value.
Next, in step S65, ctr and threshold C are compared, and if ctr ≧ C, it is determined that an attack has occurred, and countermeasure processing is executed. Otherwise, the process ends. The above-mentioned values of Nx and C represent “sensitivity” at the time of attack determination, and can be changed by setting.

  FIG. 6 is a flowchart showing a detailed procedure of the countermeasure process shown in FIG. When an attack is detected, the node 8 creates an abnormal queue, which will be described later, and classifies packets having abnormal attribute values. In the abnormal queue, the maximum output amount is limited to the average value (214) of the unit time packet amount of the attribute value before the attack occurs. Hereinafter, the flow of processing will be described by taking as an example a case where the abnormal attribute value is {destination port = X1 OR destination port = X2} AND packet length = Y.

First, in step S66, the sum of average values of unit time packet amounts corresponding to {destination port = X1} and {destination port = X2} in the normal state from the average value (213) of the destination port traffic table (20) ( μport = X1 OR port = X2) is extracted. Next, in step S67, the value is set as an abnormal limit value. In step S48, it is determined whether there is another abnormal attribute. If there is no abnormal attribute, the process proceeds to step S72. If there are other abnormal attributes, the abnormal attribute value {packet length = 64} of the next attribute type is referred to in step S69, a traffic table corresponding to the attribute type is selected, and the average value of the corresponding table is extracted.
In the case of the above example, the average value of records corresponding to {packet length = 64} in the packet length traffic table is extracted.

Next, in step S70, the ratio of the average value extracted in step S69 to the total value (average value ratio) is calculated. For example, if the average value having the {packet length = 64} attribute is μlength = 64 and the average value of all packets is μ, the average value ratio = μlength = 64 / μ. Next, in step S71, a value obtained by multiplying the previously obtained abnormality limit value by the average value ratio is set as a new abnormality limit value. The processes in steps S68 to S71 described above are repeated as many times as the number of discovered abnormal attribute values. On the other hand, in step S72, an abnormal queue is created and the maximum transfer amount is set as an abnormal limit value. In step S73, the packet having the abnormal attribute value is classified into an abnormal queue.

FIG. 7 shows a packet amount control model of the node 8. The packet amount control model is a model for classifying input packets by class and controlling the output packet amount for each class.
The filter 31 classifies the input packet into a regular class 32 and an abnormal class 33. The filter 31 checks the attribute values of the input packet, and if they match the abnormality identifier, transmits the packet to the corresponding abnormality class 33. If it does not match the abnormal identifier, the packet is transmitted to the regular class 32. The regular class 32 is a default class, and packets classified into the regular class 32 are connected to the regular queue 321 and output without limiting the amount of output packets. On the other hand, packets classified into the abnormal class 33 are connected to the abnormal queue 331, and the output packet amount is limited to the previously set abnormal limit value.
The abnormal class 33 and the abnormal queue 331 are created every time an abnormal attribute value and an abnormal limit value are generated according to the algorithm shown in FIG. Here, the “queue” is a memory for temporarily storing packets used in the node 8 while waiting for the order in which the input packets are transmitted, and is a multidimensional traffic table. It is stored in a work data area (not shown) different from the above.

As described above, according to the embodiment of the present invention, every time a packet is input, the node 8 extracts an attribute value given to the packet. The attribute value here refers to an IP packet value such as a destination address, a destination port, a protocol number, and a packet length. For each attribute type, the node 8 measures the distribution for each attribute value of “amount of packets input within a unit time (hereinafter referred to as a unit time packet amount)” for each preset “measurement period”. Record as a traffic profile.
When the measurement data is accumulated for a preset “measurement period”, the network device calculates, for example, the average value, maximum value, and standard deviation of the period, and records them in the traffic profile. Update the value. As a result, the traffic profile represents the distribution of packet amount by attribute value at the normal time.

At the time of measurement, the alert mode is entered when the measurement data is significantly different from the packet amount distribution of the traffic profile. For example, in a certain attribute type, when the measurement data of a specific attribute value satisfies the condition that the maximum value in the measurement period exceeds N1 times or exceeds the average value N2 × standard deviation in the measurement period, the alert mode to go into. Further, when this alert mode is continuously satisfied C times or more for the same attribute value, it is detected as an attack. These values of Nx and C represent “sensitivity” of attack determination and can be changed by setting.
According to the present invention, the traffic profile representing the packet characteristic value at the normal time is automatically updated, and the content and the packet amount distribution for each measurement period are compared in real time without setting the abnormal condition in advance. Enables attack detection.

Furthermore, according to the present invention, when an attack is detected, the node 8 assigns a packet having an abnormal attribute value and a packet having no abnormal attribute value to different queues. A queue assigned to a packet having an abnormal attribute value limits the maximum output amount to the average value of the unit time packet amount of the attribute value at the normal time.
For example, if {destination port = X} is extracted as an abnormal attribute value and the average value of the unit time packet length of {destination port = X} at normal time recorded in the traffic profile is Y, an attack occurs. Thereafter, the packet of {destination port = X} is assigned to the “abnormal queue” whose maximum output amount is limited to Y. When abnormal attribute values are extracted into a plurality of attribute types, packets having a logical product (AND) condition are classified into an abnormal queue. At that time, the maximum output amount of the abnormal queue is further limited according to the ratio of each abnormal attribute value at the normal time.

Specifically, consider the case where {packet length = 64} is also extracted as an abnormal attribute value in the above example. Assuming that the ratio of packets having the {packet length = 64} attribute to the total packets for the average value of the unit time packet amount recorded in the multidimensional traffic profile is a, the destination port after the attack occurs = X AND packet length = 64} is assigned to an “abnormal queue” in which the maximum output amount is limited to Y × a.
As described above, according to the present invention, the node performs band limitation on the attack packet, and the limit value is automatically calculated based on the normal traffic profile, whereby the packet of the authorized user having no abnormal identifier is obtained. Can be less affected.

In addition, according to the above-described embodiment of the present invention, the case where the network device of the present invention is mounted on the node 8 has been described, but it is also mounted on other nodes 5 to 7 connected to a network such as a WAN. In this case, it has the same function as above and operates.
Further, the procedure executed by each of the packet amount distribution measuring unit 81, the traffic profile extracting unit 83, the attack detecting unit 84, the attack packet feature value extracting unit 85, and the bandwidth limit value calculating unit 86 shown in FIG. The present invention is realized by recording on a computer-readable recording medium, causing a computer system to read and execute a program recorded on the recording medium. The computer system here includes an OS and hardware such as peripheral devices.

Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design and the like within the scope not departing from the gist of the present invention.

It is the figure quoted in order to demonstrate the network environment to which the network apparatus of this invention is connected. It is the block diagram which expanded and showed the function of the internal structure of the node by which the network device of this invention is mounted. It is a figure which shows an example of the data structure of the multidimensional traffic profile used by embodiment of this invention. It is a flowchart which shows operation | movement of this invention embodiment. It is a flowchart which shows operation | movement of this invention embodiment. It is a flowchart which shows operation | movement of this invention embodiment. It is the figure which showed notionally the packet amount control model which the network apparatus of this invention has.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1-4 ... Terminal, 5-8 ... Node (network apparatus), 9 ... WAN, 81 ... Packet amount distribution measurement part, 82 ... Traffic profile DB, 83 ... Traffic profile extraction part, 84 ... Attack detection part, 85 ... Attack Packet feature value extraction unit, 86... Bandwidth limit value calculation unit

Claims (6)

In a network environment in which multiple terminals are connected, a network attack countermeasure method that prevents an attack that makes a specific server or network unusable by sending a large amount of packets,
Periodically measuring a packet amount distribution related to a plurality of attribute types of packets transmitted from the terminal;
Recording a packet amount distribution related to a plurality of attribute types of the measured packet as a traffic profile;
Automatically extracting feature values of a packet amount distribution related to a plurality of attribute types at normal time from the recorded traffic profile;
A network attack countermeasure method characterized by comprising:   2. The network attack countermeasure method according to claim 1, wherein the attribute type is one or more of a destination address, a destination port number, a protocol number, and a packet length. In a network environment in which a plurality of terminals are connected, a network device that prevents a specific server or an attack that makes a network unusable by sending a large amount of packets,
A packet amount distribution measuring unit that periodically measures a packet amount distribution related to one or more attribute types of packets transmitted from the terminal;
A traffic profile recording unit that records a packet amount distribution relating to one or more attribute types of the measured packet as a traffic profile;
A traffic profile extraction unit for automatically extracting feature values of a packet amount distribution regarding a plurality of attribute types at normal time from the recorded traffic profile;
A network apparatus comprising: The attack is detected from the difference between the packet amount distribution of the plurality of periodically measured attribute types and the traffic profile at the normal time, and the detailed feature value of the attack packet is automatically extracted when the attack is detected. An attack packet feature value extraction unit;
The network device according to claim 3, further comprising: After the attack detection, when performing bandwidth limitation on the attack packet, a bandwidth limitation value calculation unit that calculates the limitation value based on the normal traffic profile;
The network apparatus according to claim 4, further comprising: In a network environment in which a plurality of terminals are connected, a program used for a network device that prevents an attack that makes a specific server or network unusable by sending a large amount of packets,
Processing for periodically measuring a packet amount distribution regarding a plurality of attribute types of packets transmitted from the terminal;
A process of recording a packet amount distribution regarding a plurality of attribute types of the measured packet as a traffic profile;
A process of automatically extracting feature values of a packet amount distribution related to a plurality of attribute types at normal time from the recorded traffic profile;
A program that causes a computer to execute.

Images