Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L63/00—Network architectures or network communication protocols for network security
  • H04L63/02—Network architectures or network communication protocols for network security for separating internal from external traffic, e.g. firewalls
  • H04L63/0227—Filtering policies
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L63/00—Network architectures or network communication protocols for network security
  • H04L63/02—Network architectures or network communication protocols for network security for separating internal from external traffic, e.g. firewalls
  • H04L63/0227—Filtering policies
  • H04L63/0263—Rule management
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L63/00—Network architectures or network communication protocols for network security
  • H04L63/10—Network architectures or network communication protocols for network security for controlling access to devices or network resources
  • H04L63/102—Entity profiles

Definitions

• Firewalls are a type of system designed to protect unauthorized access to or from a private network and can be implemented in hardware, software, or a combination of both hardware and software.
• the recent trend in firewall protection is toward "personal firewalls.”
• the security benefits of this trend have been positive and have provided an improvement in configurability, utility, and (in the case of mobile devices) portability of firewalls.
• This is referred to as the "every node is a firewall" model, which presumes the underlying fundamental economic assumption that the cost of delivery of unwanted packets is negligible. This economic assumption is not always correct, especially in the realm of wireless communication.
• firewalls should mitigate the volume of unwanted traffic; even a small reduction of such unwanted traffic is a net gain.
• a common type of firewall is a packet filter that passes or blocks packets, but otherwise leaves the traffic flow untouched. At the core of each packet filter is a mechanism that classifies packets according to a supplied policy. Stateful packet filters (such as OpenBSD's ⁇ /) possess scalable mechanisms for processing packets that belong to established traffic flows.
• Packets that do not belong to an established flow are classified according to a policy, which is expressed as a set of rules. Rules are generally processed in sequence order to assess each packet.
• Some packet classifiers employ optimization techniques to their rule sets in order to speed up packet processing. Facilities for early termination of rule processing under specified circumstances are common. A more sophisticated example is pfs skipsteps, which enable predictive skipping when contiguous rule blocks could never match a packet. Such techniques can be very effective if the rule set is highly ordered and exhibits strong commonality in rule criteria. However, in a highly dynamic environment, where there are ongoing incremental updates to the rule set, these conditions are not generally met. [007] Traditionally, classifier rule sets tend to be quite static in nature, and are often updated through a manual process.
• Nodes protected by a centralized packet filter may wish to extend service (typically by listening for packets that initiate a flow) at any time. Similarly, they may wish to retract previously offered services. This is consistent with the Internet end-to- end model. If the maximum number of unwanted packets is to be blocked while allowing ad hoc service extension and retraction, the filtering policy must be dynamically updated by nodes as changes occur. The filter should also have a mechanism (such as keep-alives) to discover when a node departs the network abruptly, so that obsolete rules can be removed from the policy in a timely fashion.
• Embodiments describe a method and/or system for efficient classification of network packets.
• a method for classifying a packet is provided. The method includes describing a packet as a feature vector and mapping the feature vector to a feature space.
• the feature vector can be an n-dimensional feature and the feature space can be an ⁇ -dimensional feature space.
• the feature vector can comprise features represented by a number wherein the number is within a predetermined range and can be generated based on at least one feature of the packet.
• the method can include defining a feature prism, classifying the packet relative to the feature prism, and determining if the feature vector matches the feature prism. The packet is classified based on the result of this matching process. For instance, if the feature vector matches the feature prism, the packet is passed to a recipient; otherwise it is blocked. [011]
• According to yet another embodiment is an apparatus for classifying a packet.
• the apparatus includes an identification component that defines at least one feature of the packet and a classification component that classifies the packet based at least in part upon the at least one defined feature.
• the identification component can further define the at least one feature of the packet as a number that is included within a predetermined range.
• a prognosis component can also be included that generates a stateful feature based at least in part on information from previous packets.
• Also included can be a comparison component that applies matching techniques to facilitate categorizing the data access of the packet.
• the packet feature can be an included feature that is present in the packet, a generated feature that is synthesized from values in the packet, and/or a stateful feature.
• a computer readable medium having computer-executable instructions for inserting prisms into a spatial index. A packet is matched against these prisms by performing a point queries on the index with the packet's feature vector.
• a processor that executes instructions for applying packet matching. The instructions include constructing a spatial index and inserting prisms into the spatial index. The packets are matched against prisms by performing point queries on the spatial index.
• FIG. 1 illustrates a block diagram of a communication system utilizing firewall technology.
• FIG. 2 illustrates a block diagram of a system for classifying packets.
• FIG. 3 illustrates packet classification system that defines packets according to features associated with the packets.
• FIG. 4 illustrates a system that applies matching techniques to facilitate denying and/or allowing data access.
• FIG. 5 illustrates a flow chart of a methodology for applying packet classification and matching.
• FIG. 6 illustrates a flow chart of a methodology for applying packet matching.
• FIG. 7 illustrates a flow chart of a methodology for classifying a packet and applying filtering techniques.
• FIG. 8 illustrates a communication system that includes an artificial intelligence-based component that can automate functionality with respect to packet filters.
• FIG. 9 illustrates a conceptual block diagram of a configuration of a terminal.
• TERMINOLOGY [024] Affine space - A vector space in which axes are not necessarily mutually perpendicular, nor have the same unit measure.
• Complexity A mathematical measure of the way in which an algorithm scales.
• Cuboid A prism (convex solid), in which all faces are rectangles.
• Feature prism An n dimensional axes aligned cuboid in n dimensional feature space.
• Feature space A finite n dimensional affine space, in which the nth axis represents the range of the nth feature.
• Feature vector A vector of specific feature values.
• Firewall A device that applies a security policy to traversing network traffic.
• ICMP Internet Control Message Protocol. Utilized to send control messages between Internet hoses.
• Variants include ICMPv ⁇ (for use with IPv6).
• IP Internet protocol.
• Variants include IPv4 (version 4) and IPv6 (version 6).
• Packet The transmission unit in a network.
• Packet filter A mechanism that selects specific packets to forward or discard.
• ком ⁇ онент As used in this application, the terms "component,” “system,” and the like are intended to refer to a computer-related entity, either hardware, firmware, a combination of hardware and software, software, or software in execution.
• a component may be, but is not limited to being, a process running on a processor, a processor, an object, an executable, a thread of execution, a program, and/or a computer.
• both an application running on a computing device and the computing device can be a component.
• One or more components can reside within a process and/or thread of execution and a component may be localized on one computer and/or distributed between two or more computers.
• these components can execute from various computer readable media having various data structures stored thereon.
• the components may communicate by way of local and/or remote processes such as in accordance with a signal having one or more data packets (e.g., data from one component interacting with another component in a local system, distributed system, and/or across a network such as the Internet with other systems by way of the signal).
• a signal having one or more data packets (e.g., data from one component interacting with another component in a local system, distributed system, and/or across a network such as the Internet with other systems by way of the signal).
• FIG. 1 illustrates a block diagram of a communication system 100 utilizing firewall technology that can be implemented in a portable device or terminal, a portable (mobile) phone, a personal data assistant, a personal computer (desktop or laptop), or other electronic and/or communication devices.
• System 100 includes a packet filter 102 that filters incoming and/or outgoing data, referred to as a data or network packet 104.
• a packet 104 can be any type of communication, including a group of data, sent and/or communicated from one device to another device.
• Packet filter technology inspects each packet (incoming data), classifies each packet, and performs one or more actions as a consequence of such inspection and/or classification. Typical actions are to pass, block, and/or route the packet in a specific manner. Stateful packet filters may also take into account previously seen packets when performing classification. [047]
• packet filter 102 may allow a data packet(s) 104 sent from a sender 106, located on one side of packet filter 102, to be transmitted to a recipient 108, located on the other side of packet filter 102.
• Packet(s) 104 conveyed by sender 106 that are intended and/or authorized to reach recipient 108 are relayed or allowed to pass through packet filter 102. Packet(s) 104 not intended and/or not authorized for such recipient 108 are blocked by packet filter 102 and not relayed to recipient 108. In such a way, recipient 108 is unaware of and does not receive unwanted packets and/or packets unintended for such recipient 108.
• Packet filter(s) 102 are typically configured by specifying a set of classification rules. It is possible to construct very simple packet filters that exhibit O(log N) complexity relative to the size of the rule set, generally by filtering on a very small number of criteria.
• a packet filter with 0(N) performance may be acceptable when the number of rules remains relatively small, particularly if each rule allows for rich expression of classification criteria. However, for large numbers of rules, such filters are not viable.
• a packet filter that protects a large number of systems, and that allows each system to specify a rich security policy, is a good example of an application that needs better performance than extant packet filtering technology.
• FIG. 2 illustrates a block diagram of a system 200 for classifying packets.
• the packet classification system 200 represents each packet as a point and each rule as a prism in ⁇ -dimensional space. Fast matching of packets against prisms is achieved by utilizing a spatially indexed data structure.
• System 200 includes an identification component 202 that interfaces with a classification component 204.
• Identification component 202 and classification component 204 can be employed separately, as illustrated, or as a single component and can be included as components of a packet filter separately or integrated with a communication device.
• Identification component 202 receives a packet 206 and associated feature(s) relayed by a sender 208 that appear to be intended for recipient 210.
• Sender 208 and/or recipient 210 can be a user and/or entity (e.g., the Internet, another system, a computer, ).
• Packet 206 possesses a predetermined set of n interesting features that identification component 202 can utilize to define each feature, allowing each feature to be represented by a number that falls within a predetermined range of numbers.
• Features can be represented by floating point numbers, but are most often integral in nature.
• Identification component 202 interfaces with and transmits the defined features to classification component 204.
• Classification component classifies such defined features according to predefined classification rules. Classification of the features includes a determination whether packet 206 is intended and/or authorized for recipient 210 or if packet 206 is unintended and/or unwanted, and thus blocked before reaching recipient 210.
• classification component 204 can employ packet matching techniques and/or spatial access methods (SAMs), such as R-trees, R+-trees and/or R* -trees. Such techniques will be discussed in connection with further aspects disclosed herein.
• SAMs spatial access methods
• System 300 includes an identification component 302 that interfaces with a classification component 304, and a prognosis component 306.
• Identification component 302 receives a data packet 308 intended to be relayed to a recipient 310 and defines packet 308 according to its associated features. The defined associated features are utilized by system 300 to determine if packet 308 is intended for recipient 310 and/or if packet 308 is unwanted by recipient 310.
• Packet 308 can have a set of n interesting features that can be referred to as included feature(s) 312, generated feature(s) 314, and/or stateful feature(s) 316.
• included feature(s) 312, generated feature(s) 314, and/or stateful feature(s) 316 For instance, the source and destination address of an IP packet 308 can be utilized directly as included features 312 as they are each representable by an integer with a predetermined range of 0 to 2 -1, or 0 to 2 -1, in the case of IPv4, or IPv6, respectively.
• the upper layer protocol number is another example of a typical included feature 312, being an integer in the range of 0 to 255.
• information in a packet 308 can be utilized either directly as a feature 312, or to algorithmically construct a feature 312 and 314. In either case, such information generates the feature 312 and 314.
• Information that may or may not be present in the packet 308 may also be utilized to generate features 312 and 314.
• a typical example of such information is optional data (such as IPv4 header options and/or IPv6 optional headers).
• Information from the packet 308 may also be utilized to generate feature(s) 314, such as fields from encapsulate upper layer protocol headers. Typical examples of such information is TCP or UDP port numbers, and ICMP types and codes.
• a generated feature 314 takes on a distinguished "undefined" value (which is an element of the feature's range). In other words, when the information is not present, the feature 314 is still defined.
• Stateful features 316 can be generated utilizing information recalled from previous packets, through utilization of prognosis component 306. In other words, feature generation maybe stateful.
• Prognosis component 306 can store, record, perform a look up, etc. of packet information and associated features 312-316. Based on such data, prognosis component 306 can infer a stateful feature 316 for a current packet 308 based upon the prognosis stateful feature 316.
• Each packet 308 can be represented as a fixed length feature vector v, consisting of n feature values ⁇ .
• Each vector v describes a point in an ra-dimensional affine feature space F. Accordingly, ⁇ -dimensional feature vectors are mapped to points in an w-dimensional feature space.
• An axes aligned ⁇ -dimensional cuboid ⁇ in feature space F can be defined by specifying a contiguous sub-range for each feature.
• Each feature prism represents a set of geometrically coherent classification criteria.
• Prism P encloses vector v if:
• FIG. 4 illustrates a system 400 that applies matching techniques to facilitate denying and/or allowing data access.
• System 400 includes a comparison component 402 that receives a packet 404 from a data originator 406.
• Comparison component 402 interfaces with a packet filter 408. While matching component 402 and packet filter 408 are illustrated as separate components, it is to be understood that they can comprise the same component.
• Packet classification techniques are utilized where P is defined as an arbitrary set of feature prisms, and prisma is defined to be any element of P.
• Comparison component 402 determines if a vector v of packet 404 matches an arbitrary set of feature prisms P. A feature vector v matches feature prism P if there exists a prisma that encloses vector v.
• comparison component 402 finds a match, the associated packet 404 is permitted through the filter 408 and can reach its destination 410. In such a situation, feature prism P represents a positive rule set 412. If comparison component 402 interprets feature prism P as a negative rule set 414, there is not a match. Not having a match results in the packet 404 being blocked by the packet filter 408 and not reaching destination 410.
• Classification that is more complex is possible by matching vector v against a sequence, or even a decision tree, of distinct feature prisms P. Accordingly, the packet classification criteria is described as cuboids in an ⁇ -dimensional feature space, and feature vectors are matched against criteria by geometrical enclosure.
• R-trees are an extension of the well-known B+- tree data structure, in which the keys are multidimensional rectangles. Interior nodes hold the minimum- bounding rectangle (MBR) for each child.
• a classifier rule set ⁇ may be represented by an R-tree whose leaf MBRs are isomorphic with ⁇ 's elements.
• FIG. 5 illustrates a flow chart of a methodology for applying packet classification and matching. While, for purposes of simplicity of explanation, the following methodologies are shown and described as a series of acts, it is to be understood and appreciated that the methodologies are not limited by the order of acts, as some acts may, in accordance with these methodologies, occur in different orders and/or concurrently with other acts from that shown and described herein.
• the method starts, at 502, when a packet is received at firewall that appears to be intended for a recipient protected by such firewall.
• a firewall technology that can be utilized is a packet filter that passes or blocks packets, but otherwise leaves the traffic flow untouched.
• the received packet is analyzed, at 504, to determine the intended recipient and to analyze features associated with the packet. For example, there can be a number of features (n features) associated with a packet. These features can be included features, generated features, and/or stateful features.
• the included features can be the source and destination address of the packet, for example.
• Generated features are those features algorithmically constructed based on information that may or may not be present in the packet, such as optional data (e.g., IP v4 header options, IPv6 optional headers). Generated features are stateful and can be generated utilizing historical information from previous received packet(s). [068]
• the analyzed features are utilized, at 506, to classify the packet.
• the features are represented by a number (e.g., floating point, integral, ...) that falls within a predetermined range of numbers. It should be noted that the features need not be orthogonal.
• the features are classified according to classification rules that are predefined.
• the classification rules can employ packet matching techniques and/or spatial access methods (SAMs), such as R-trees, R+- trees, and/or R* -trees. It is to be understood that while R-trees and their variants are discussed, the systems and/or methods disclosed herein are not limited as such and are equally applicable to any spatial index methods.
• SAMs spatial access methods
• the classified packet is, at 508, either blocked and not transmitted to the target recipient or allowed to pass to the recipient.
• a packet is blocked if the identified recipient is not the intended recipient and/or if the packet is not desired by the recipient. For example, a recipient may not want communication from a particular source, subject matter, or other defined criteria.
• Classified packets falling within the defined criteria are not communicated to recipient, and recipient may remain unaware of the existence of such packets. Classified packets not falling within the defined criteria as allowed to pass through and communicated to the recipient.
• a spatial index is constructed.
• the packet can be described as a fixed length feature vector v.
• the feature vector can be an ⁇ -dimensional feature and the feature space can be an ⁇ -dimensional feature space.
• the feature vector can comprise features represented by a number wherein the number is within a predetermined range and can be generated based on at least one feature of the packet.
• the method continues, at 604, where a prism P is inserted into the spatial index.
• the prism is an axis-aligned ⁇ -dimensional cuboid in feature space and is defined by specifying a contiguous subrange for each axis.
• Each feature prism represents a set of geometrically coherent classification criteria.
• the packet is then matched against the prism, at 606. For example, a feature vector v of the packet matches a feature prism P if there exists a prism/? that encloses vector v.
• feature prism P represents a positive rule set. With a match access to the data is permitted and can reach its intended destination. If there is not a match, feature prism P is a negative rule set and data access is blocked.
• the matching can alternatively or in addition be performed by utilizing point queries ⁇ , which is performed utilizing a random point from inside each prism.
• the ⁇ point queries cab also be performed utilizing randomly generated "typical" vectors. After the ⁇ point queries are performed a determination is made whether the point queries successfully matched a prism.
• FIG. 7 illustrates a flow chart of a methodology for classifying a packet.
• the method starts, at 702, when a packet is received at a packet filter, for example, intended for a recipient that is protected from unwanted and/or unauthorized packets.
• the packet can be received from a user and/or entity (e.g., the Internet, another system, a computer, ).
• entity e.g., the Internet, another system, a computer, .
• the packet is described as a feature vector, which is a vector of specific feature values.
• Each packet can be represented as a fixed length feature vector v, consisting of n feature values ⁇ .
• Each feature vector v describes a point in an ⁇ -dimensional affine feature space F.
• the method continues, at 706, where the feature vector v is mapped to points in an ra-dimensional feature space.
• a feature prism P which is an axis- aligned ⁇ -dimensional cuboid in feature space F, is defined by specifying a contiguous subrange for each axis.
• a set of feature prisms forming a rule set of a packet classifier is binary classified relative to a feature prism P.
• a determination is made, at 714, whether the feature vector v matches the feature prism P.
• a feature vector v of the packet matches a feature prism P if there exists a prism p that encloses vector v. If the determination is "yes,” there is a match and the packet is represented as a positive rule set and permitted access through the packet filter, at 716.
• FIG. 8 illustrates a communications system 800 that employs artificial intelligence (AI), which facilitates automating one or more features associated with a packet filter 802.
• the packet filter 802 receives a packet 804 from a data originator 806 intended for a destination 808 that is protected by packet filter 802.
• Packet filter 802 can work in conjunction with an artificial intelligence component 810 to minimize unauthorized and/or unwanted packets 804 from reaching a protected destination 808.
• a communication system e.g., in connection with classifying and filtering packets
• a process for determining if a packet of data is authentic and/or intended for a particular recipient can be facilitated through utilization of an automatic classifier system and process.
• the classifier can be employed to determine which packet filter to employ in a particular situation.
• Such classification can employ a probabilistic and/or statistical- based analysis ⁇ e.g., factoring into the analysis utilities and costs) to prognose or infer an action that a user desires to be automatically performed.
• attributes can be features, words, phrases or other data-specific attributes derived from the features ⁇ e.g., included, generated, stateful)
• the classes are categories or areas of interest ⁇ e.g., levels of classification and/or matching).
• a support vector machine (SVM) is an example of a classifier that can be employed. The SVM operates by finding a hypersurface in the space of possible inputs, which hypersurface attempts to split the triggering criteria from the non-triggering events.
• Other directed and undirected model classification approaches include, e.g., na ⁇ ve Bayes, Bayesian networks, decision trees, neural networks, fuzzy logic models, and probabilistic classification models providing different patterns of independence can be employed. Classification as used herein also is inclusive of statistical regression that is utilized to develop models of priority. [079] As will be readily appreciated from the subject specification, the system can employ classifiers that are explicitly trained ⁇ e.g., through utilization of a generic training data) as well as implicitly trained ⁇ e.g., by observing user behavior, receiving extrinsic information).
• SVM's are configured by means of a learning or training phase within a classifier constructor and feature selection module.
• the classifier(s) can be used to automatically learn and perform a number of functions, including but not limited to determining according to a predetermined criteria when to block a packet, when to permit a packet to pass though the filter, etc.
• FIG. 9 illustrated is a conceptual block diagram of a possible configuration of a terminal 900. As those skilled in the art will appreciate, the precise configuration of the terminal 900 may vary depending on the specific application and the overall design constraints.
• Processor 902 can implement the systems and methods disclosed herein.
• Terminal 900 can be implemented with a front end transceiver 904 coupled to an antenna 906.
• a base band processor 908 can be coupled to the transceiver 904.
• the base band processor 908 can be implemented with a software based architecture, or any other type of architecture.
• a microprocessor can be utilized as a platform to run software programs that, among other functions, provide control and overall system management function.
• a digital signal processor (DSP) can be implemented with an embedded communications software layer, which runs application specific algorithms to reduce the processing demands on the microprocessor.
• the DSP can be utilized to provide various signal processing functions such as pilot signal acquisition, time synchronization, frequency tracking, spread-spectrum processing, modulation and demodulation functions, and forward error correction.
• Terminal 900 can also include various user interfaces 910 coupled to the base band processor 908.
• the base band processor 908 comprises a processor 902.
• the processor 902 may be a software program running on a microprocessor.
• the processor 902 is not limited to this embodiment, and may be implemented by any means known in the art, including any hardware configuration, software configuration, or combination thereof, which is capable of performing the various functions described herein.
• the processor 902 can be coupled to memory 912 for the storage of data.
• the embodiments described herein may be implemented by hardware, software, firmware, middleware, microcode, or any combination thereof.
• the systems and/or methods When the systems and/or methods are implemented in software, firmware, middleware or microcode, program code or code segments, they may be stored in a machine-readable medium, such as a storage component.
• a code segment may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a class, or any combination of instructions, data structures, or program statements.
• a code segment may be coupled to another code segment or a hardware circuit by passing and/or receiving information, data, arguments, parameters, or memory contents. Information, arguments, parameters, data, etc.

Landscapes

• Engineering & Computer Science (AREA)
• Computer Hardware Design (AREA)
• Computer Security & Cryptography (AREA)
• Computing Systems (AREA)
• General Engineering & Computer Science (AREA)
• Computer Networks & Wireless Communication (AREA)
• Signal Processing (AREA)
• Business, Economics & Management (AREA)
• General Business, Economics & Management (AREA)
• Data Exchanges In Wide-Area Networks (AREA)
• Computer And Data Communications (AREA)
• Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Priority Applications (2)

Applications Claiming Priority (6)

Publications (1)

Family

ID=34972844

Family Applications (1)

Country Status (4)

Cited By (2)

Families Citing this family (48)

Citations (3)

Family Cites Families (12)

• 2005
  • 2005-06-21 JP JP2007518222A patent/JP2008504737A/ja active Pending
  • 2005-06-21 EP EP05762219A patent/EP1762079A1/en not_active Withdrawn
  • 2005-06-21 WO PCT/US2005/022023 patent/WO2006002215A1/en not_active Ceased
  • 2005-06-21 US US11/158,588 patent/US8027330B2/en not_active Expired - Fee Related
• 2010
  • 2010-09-21 JP JP2010210843A patent/JP5362669B2/ja not_active Expired - Fee Related
• 2011
  • 2011-09-26 US US13/245,680 patent/US8750285B2/en not_active Expired - Fee Related

Patent Citations (3)

Also Published As

Similar Documents

Legal Events