CN108881028B - SDN network resource scheduling method for realizing application awareness based on deep learning - Google Patents
SDN network resource scheduling method for realizing application awareness based on deep learning Download PDFInfo
- Publication number
- CN108881028B CN108881028B CN201810573044.1A CN201810573044A CN108881028B CN 108881028 B CN108881028 B CN 108881028B CN 201810573044 A CN201810573044 A CN 201810573044A CN 108881028 B CN108881028 B CN 108881028B
- Authority
- CN
- China
- Prior art keywords
- network
- sdn
- node
- data packet
- path
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/50—Queue scheduling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/74—Address processing for routing
- H04L45/745—Address table lookup; Address filtering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/80—Actions related to the user profile or the type of traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/25—Routing or path finding in a switch fabric
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
An SDN network resource scheduling method for realizing application awareness based on deep learning comprises the following steps: based on the network characteristics of the SDN, a deep neural network DNN is deployed on a virtual network function VNF located on a data plane, the DNN learns and classifies the application data flow forwarded by the switch and reports the classification result to an SDN controller, the SDN controller performs network resource scheduling according to the classification result to generate routing information meeting the network resource requirement of the application data flow and sends the routing information to the switch.
Description
Technical Field
The invention relates to an SDN network resource scheduling method for realizing application awareness based on deep learning, belonging to the technical field of information, in particular to the technical field of SDN networks.
Background
There are a variety of access devices in the internet of things, and different access devices may carry different types of network applications that have different requirements for network resources. For example, for internet-supported audio (VOIP), such applications have high requirements on network latency, and we should try to allocate low-latency paths for such applications. For video monitoring application, the real-time transmission of video data can be realized by requiring a network path with low time delay and occupying enough bandwidth. In the scene of the internet of things, the key to realize the on-demand scheduling of the network is to acquire specific type information of traffic.
Most of The traditional traffic scheduling is based on packet header information to classify traffic, The Internet Engineering Task Force (IETF) allocates fixed port numbers for some standard protocols, and network traffic can be classified into different application categories through corresponding port numbers, but port number-based classification has many problems: firstly, with the increase of network applications, a plurality of application layer protocols are not allocated with special port numbers, and the application protocols cannot be distinguished through the port numbers; secondly, some application layer protocols may carry a plurality of different types of application contents, and the requirements of different application contents on the network are quite different. For example, the http protocol is the most widely applied application layer protocol at present, when the http protocol is used for web browsing, the http protocol has certain requirements on latency, when the http protocol is used for carrying video traffic, the http protocol is bandwidth sensitive, a large-bandwidth link can sufficiently reduce the loading time of a video, the user experience is improved, and the high requirements on latency are not met. In combination with the above analysis, it is far from sufficient to simply rely on packet header information for traffic classification. Another common approach for traffic identification is Deep Packet Inspection (DPI). DPI technology is an application-layer based traffic inspection technology called "deep packet inspection". In addition to parsing the header information, DPI also identifies the specific payload content of each application. When an IP data packet, TCP or UDP data stream passes through a bandwidth management system based on DPI technology, the system reassembles application layer information in OSI seven-layer protocol by reading the content of the IP packet payload in depth, so as to obtain the content of the whole application program, and then performs a shaping operation on the traffic according to the management policy defined by the system. The DPI technology solves the problem of traditional flow identification based on packet header fields to a certain extent, but the DPI technology also has many problems: (1) poor expandability: since the method has hysteresis in identifying traffic of a new P2P application, that is, the new application cannot be detected before the feature library is not upgraded, the payload features of the new application must be found before effective detection of the application can be performed. This becomes a bottleneck limiting the process. (2) Lack of encrypted data analysis functionality: some application loads adopt encryption transmission, and hide the protocol and data characteristics of the application, so that the detection capability of the deep packet inspection technology for the encryption application is very limited. (3) The cost is high: because the operations such as protocol analysis reduction, feature matching and the like need to be completed, the calculation and storage overhead is large, and the performance of the flow detection algorithm is low. The more complex the load characteristics are, the higher the detection cost is, and the worse the algorithm performance is.
In summary, in the scene of the internet of things, how to quickly and effectively identify the specific type of the data traffic becomes an urgent technical problem to be solved in the technical field of the internet of things.
Disclosure of Invention
The SDN is a novel network architecture, realizes the separation of a network data plane and a control plane, provides the programmability of data plane equipment, and can realize the intelligent management of the network. The SDN controller is a hub of a control plane, and can acquire all information of a network, where the information includes topology of the network, bandwidth of a link, delay, and the like. In combination with this information, the controller can allocate transmission paths for different types of application traffic that meet its network requirements. But the SDN controller issues a control policy to the SDN switch through a specific control channel, and once the controller fails, the network loses control and even crashes.
In view of this, the present invention is directed to a network traffic type identification method based on an SDN network, so that an SDN controller can sense the type of an application, and further implement on-demand scheduling of network resources.
In order to achieve the purpose, the invention provides an SDN network resource scheduling method for realizing application awareness based on deep learning, which comprises the following steps: based on the network characteristics of the SDN, a deep neural network DNN is deployed on a virtual network function VNF located on a data plane, the DNN learns and classifies the application data flow forwarded by the switch and reports the classification result to an SDN controller, the SDN controller performs network resource scheduling according to the classification result, routing information meeting the network resource requirements of the application data flow is generated, and the routing information is issued to the switch.
The method comprises the following operation steps:
(1) the client host application sends a data packet to the server host, and the data packet enters the SDN network; the client host is connected with an edge node SA of the SDN network, and the server host is connected with an edge node SB of the SDN network;
(2) after receiving the data packet, the edge node SA of the SDN network inquires a flow table, and if a corresponding flow table is matched, the data packet is forwarded according to a flow table rule; if the flow table is not matched, uploading the data packet to the SDN controller through a packet _ In message;
(3) the SDN controller receives the packet _ In message, analyzes the reported data packet, and acquires a source node and a destination node of the data packet according to network topology, wherein the source node and the destination node are respectively SA and SB;
(4) the SDN controller calculates a transmission path from a node SA to a node SB by using a shortest path algorithm, converts the path into an OpenFlow flow table, and then issues the flow table to all switching nodes on the path; all data packets sent by the application of the client host machine are matched with the flow table and are finally forwarded to the server host machine according to the action of the flow table;
(5) the SDN controller calculates a path with a source as a node SA and a destination as an SDN network edge node SC connected with the VNF by using a shortest path algorithm, and the path is converted into a flow table and issued to all nodes on the path; at SA, the data packet sent by the client host application is copied into two copies, one of which is forwarded to the node SB, and this part of data packet is finally forwarded to the server host, and the other part of data packet is forwarded to the node SC connected to the VNF;
(6) the VNF samples the data packet applied by the client host according to the set sampling duration, and after sampling is completed, the VNF calculates the feature vector of the flow data and sends the feature vector into the DNN for classification;
the characteristic vector refers to: a feature vector calculated from the timing features of the applied data stream; the data flow of the application is defined as a series of continuous data packets with the same { Source IP, Destination IP, Source Port number, Destination Port number, Protocol (TCP or UDP) }, namely { Source IP, Destination IP, Source Port, Destination Port and Protocol (TCP orUDP) }.
(7) The classification result is marked In a DSCP field In an IP packet header of a data packet, and then the data packet is reported to an SDN controller through a packet _ In message;
(8) the SDN controller receives a classification result reported by the VNF and maps the classification result into a preset resource demand, wherein the resource demand mainly refers to bandwidth requirements and time delay requirements;
(9) the SDN controller searches a path meeting the resource requirement in all paths with the source of SA and the destination of SB by using a depth First search algorithm DFS (depth First search), converts the path into a flow table, and sends the flow table to all nodes on the path; the flow tables with higher priority can cover the flow tables issued before, and at the moment, the subsequent flow applied by the client host can be forwarded along the path meeting the resource requirement of the client host.
The DNN needs to be trained in advance, and the training method comprises the following steps: and collecting flow data of different types of applications in advance through the switch, and training the DNN by adopting a supervised learning method.
The feature vector is specifically composed of the following features: the arrival time interval of the forward datagram, including the maximum, minimum, mean and standard deviation; the arrival time interval of the backward datagram, including maximum, minimum, average, standard deviation; the arrival time interval of the bidirectional datagram, including maximum, minimum, mean, standard deviation; the number of messages and the number of bytes sent by forward datagram per second; the number of messages and the number of bytes sent to the datagram per second; the ratio of the packet number and the byte number of the forward datagram and the backward datagram arriving in each second time; the forward datagram refers to uplink traffic sent by the client to the server, and the backward datagram refers to downlink traffic replied by the server to the client.
The invention has the beneficial effects that: the method of the invention deploys the classification task of the flow in the VNF, the VNF works on the data plane of the SDN network, the data sampling is directly forwarded by the switch without occupying a control channel, and the fault of the VNF does not affect the basic function of the controller, so the method of the invention greatly improves the flexibility and robustness of the system, realizes the reasonable scheduling of the network resources according to the applied resource requirements, and improves the service quality of the network.
Drawings
Fig. 1 is a schematic diagram of an SDN network according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the accompanying drawings.
Based on the network characteristics of the SDN, a Deep Neural Network (DNN) is deployed on a Virtual Network Function (VNF) (virtual network function) located on a data plane, the DNN learns and classifies the application data stream forwarded by the switch and reports the classification result to an SDN controller, and the SDN controller performs network resource scheduling according to the classification result, generates routing information meeting the network resource requirements of the application data stream and sends the routing information to the switch.
Referring to table 1, to verify the feasibility of the method of the present invention, the inventors categorized the traffic data collected from 18 different applications into 8 types, web browsing, instant messaging, video streaming, audio streaming, mail, IP telephony, P2P and file transfer, respectively. The present invention focuses on applying the resource requirements for the network, and the requirements for the network for some of the above 8 types are similar, so the inventors cluster the above 8 types into 4 categories as shown in table 1.
TABLE 1
| Type (B) | Application classes | Resource demand |
| 1 | Instant messaging/IP telephone | Low delay and |
| 2 | Video stream/audio stream | High |
| 3 | Web page browsing | The time delay is lower, the bandwidth is lower |
| 4 | P2P \ file transfer | High bandwidth |
Referring to fig. 1, the method comprises the following operation steps:
(1) the client host (h1) application sends out a data packet to the server host (h3), and the data packet enters the SDN network; the client host is connected with an edge node S1 of the SDN network, and the server host is connected with an edge node S7 of the SDN network;
(2) the edge node S1 of the SDN network inquires a flow table after receiving the data packet, and if a corresponding flow table is matched, the data packet is forwarded according to the flow table rule; if the flow table is not matched, uploading the data packet to the SDN controller through a packet _ In message;
(3) the SDN controller receives the packet _ In message, analyzes the reported data packet, and obtains a source node and a destination node of the data packet according to the network topology, wherein the source node and the destination node are S1 and S7 respectively;
(4) the SDN controller calculates a transmission path PathA from a node S1 to a node S7 by using a shortest path algorithm, converts the path into an OpenFlow flow table, and then issues the flow table to all switching nodes on the path; all data packets sent by the application of the client host machine are matched with the flow table and are finally forwarded to the server host machine according to the action of the flow table;
(5) the SDN controller calculates a path PathB with a source of a node S1 and a destination of a node S3 connected with the VNF by using a shortest path algorithm, wherein the path is converted into a flow table and issued to all nodes on the path; at this point, at node S1, the packet sent by the client host application is copied into two copies, one of which is forwarded to node S7, and this portion of the packet is eventually forwarded to the server host, and the other is forwarded to node S3 to which the VNF is connected;
(6) the VNF samples the data packet applied by the client host according to a set sampling duration (for example, 15 seconds), and after sampling is completed, the VNF calculates a feature vector of the flow data and sends the feature vector into the DNN for classification;
the characteristic vector refers to: a feature vector calculated from the timing features of the applied data stream; the data flow of the application is defined as a series of continuous data packets with the same { Source IP, Destination IP, Source Port number, Destination Port number, Protocol (TCP or UDP) }, namely { Source IP, Destination IP, Source Port, Destination Port and Protocol (TCP orUDP) }.
(7) The classification result is marked in the dscp (differentiated services code point) field in the IP header of the packet, for example: if the classification result is type 1, setting the DSCP field to be 00100000, and then reporting the data packet to the SDN controller through a packet _ In message;
(8) the SDN controller receives a classification result reported by the VNF and maps the classification result into a preset resource demand, wherein the resource demand mainly refers to bandwidth requirements and time delay requirements;
(9) the SDN controller searches a path PathC meeting the resource requirement in all paths with the source of a node S1 and the destination of a node S7 by using a DFS algorithm, converts the path into a flow table, and sends the flow table to all nodes on the path; the flow tables with higher priority can cover the flow tables issued before, and at this time, the subsequent flow of the client application can be forwarded along the path meeting the resource requirement of the client application.
The DNN needs to be trained in advance, and the training method comprises the following steps: and collecting flow data of different types of applications in advance through the switch, and training the DNN by adopting a supervised learning method.
The feature vector is specifically composed of the following features: the arrival time interval of the forward datagram, including the maximum, minimum, mean and standard deviation; the arrival time interval of the backward datagram, including maximum, minimum, average, standard deviation; the arrival time interval of the bidirectional datagram, including maximum, minimum, mean, standard deviation; the number of messages and the number of bytes sent by forward datagram per second; the number of messages and the number of bytes sent to the datagram per second; the ratio of the packet number and the byte number of the forward datagram and the backward datagram arriving in each second time; the forward datagram refers to uplink traffic sent by the client to the server, and the backward datagram refers to downlink traffic replied by the server to the client.
The inventor builds a simulation experiment environment based on Mininet, wherein the switch adopts an OVS soft switch of secondary development, and the controller is secondarily developed based on OpenDayLight. The inventor carries out simulation experiments on the method of the invention, and the simulation result shows that the method of the invention is completely effective.
Claims (3)
1. An SDN network resource scheduling method for realizing application awareness based on deep learning is characterized in that: the method comprises the following steps: based on the network characteristics of the SDN, deploying a deep neural network DNN on a virtual network function VNF positioned on a data plane, wherein the DNN learns and classifies the application data stream forwarded by the switch and reports the classification result to an SDN controller, and the SDN controller performs network resource scheduling according to the classification result to generate routing information meeting the network resource requirements of the application data stream and sends the routing information to the switch;
the method comprises the following operation steps:
(1) the client host application sends a data packet to the server host, and the data packet enters the SDN network; the client host is connected with an edge node SA of the SDN network, and the server host is connected with an edge node SB of the SDN network;
(2) after receiving the data packet, the edge node SA of the SDN network inquires a flow table, and if a corresponding flow table is matched, the data packet is forwarded according to a flow table rule; if the flow table is not matched, uploading the data packet to the SDN controller through a packet _ In message;
(3) the SDN controller receives the packet _ In message, analyzes the reported data packet, and acquires a source node and a destination node of the data packet according to network topology, wherein the source node and the destination node are respectively SA and SB;
(4) the SDN controller calculates a transmission path from a node SA to a node SB by using a shortest path algorithm, converts the path into an OpenFlow flow table, and then issues the flow table to all switching nodes on the path; all data packets sent by the application of the client host machine are matched with the flow table and are finally forwarded to the server host machine according to the action of the flow table;
(5) the SDN controller calculates a path with a source as a node SA and a destination as an SDN network edge node SC connected with the VNF by using a shortest path algorithm, and the path is converted into a flow table and issued to all nodes on the path; at SA, the data packet sent by the client host application is copied into two copies, one of which is forwarded to the node SB, and this part of data packet is finally forwarded to the server host, and the other part of data packet is forwarded to the node SC connected to the VNF;
(6) the VNF samples the data packet applied by the client host according to the set sampling duration, and after sampling is completed, the VNF calculates the feature vector of the flow data and sends the feature vector into the DNN for classification; the characteristic vector refers to: a feature vector calculated from the timing features of the applied data stream; the data flow of the application is defined as a series of continuous data packets with the same { Source IP, Destination IP, Source Port number, Destination Port number, Protocol (TCP or UDP) }, namely { Source IP, Destination IP, Source Port, Destination Port and Protocol (TCP or UDP) };
(7) the classification result is marked In a DSCP field In an IP packet header of a data packet, and then the data packet is reported to an SDN controller through a packet _ In message;
(8) the SDN controller receives a classification result reported by the VNF and maps the classification result into a preset resource demand, wherein the resource demand mainly refers to bandwidth requirements and time delay requirements;
(9) the SDN controller searches a path meeting the resource requirement in all paths with the source as SA and the destination as SB by using a depth-first algorithm (DFS) algorithm, converts the path into a flow table, and sends the flow table to all nodes on the path; the flow tables with higher priority can cover the flow tables issued before, and at the moment, the subsequent flow applied by the client host can be forwarded along the path meeting the resource requirement of the client host.
2. The SDN network resource scheduling method based on deep learning application awareness, according to claim 1, wherein: the DNN needs to be trained in advance, and the training method comprises the following steps: and collecting flow data of different types of applications in advance through the switch, and training the DNN by adopting a supervised learning method.
3. The SDN network resource scheduling method based on deep learning application awareness, according to claim 1, wherein: the feature vector in step (6) is specifically composed of the following features: the arrival time interval of the forward datagram, including the maximum, minimum, mean and standard deviation; the arrival time interval of the backward datagram, including maximum, minimum, average, standard deviation; the arrival time interval of the bidirectional datagram, including maximum, minimum, mean, standard deviation; the number of messages and the number of bytes sent by forward datagram per second; the number of messages and the number of bytes sent to the datagram per second; the ratio of the packet number and the byte number of the forward datagram and the backward datagram arriving in each second time; the forward datagram refers to uplink traffic sent by the client to the server, and the backward datagram refers to downlink traffic replied by the server to the client.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810573044.1A CN108881028B (en) | 2018-06-06 | 2018-06-06 | SDN network resource scheduling method for realizing application awareness based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810573044.1A CN108881028B (en) | 2018-06-06 | 2018-06-06 | SDN network resource scheduling method for realizing application awareness based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108881028A CN108881028A (en) | 2018-11-23 |
| CN108881028B true CN108881028B (en) | 2020-06-26 |
Family
ID=64337123
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810573044.1A Active CN108881028B (en) | 2018-06-06 | 2018-06-06 | SDN network resource scheduling method for realizing application awareness based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108881028B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111355670B (en) * | 2018-12-24 | 2023-05-02 | 中移（杭州）信息技术有限公司 | Traffic identification method and device, electronic equipment and storage medium |
| CN110298381B (en) * | 2019-05-24 | 2022-09-20 | 中山大学 | Cloud security service function tree network intrusion detection system |
| CN112311578B (en) * | 2019-07-31 | 2023-04-07 | 中国移动通信集团浙江有限公司 | VNF scheduling method and device based on deep reinforcement learning |
| CN110601916A (en) * | 2019-08-14 | 2019-12-20 | 天津大学 | Flow sampling and application sensing system based on machine learning |
| CN110380940B (en) * | 2019-08-22 | 2022-05-24 | 北京大学深圳研究生院 | Router and data packet evaluation method thereof |
| CN110581802A (en) * | 2019-08-27 | 2019-12-17 | 北京邮电大学 | fully-autonomous intelligent routing method and device based on deep belief network |
| CN110708260A (en) * | 2019-11-13 | 2020-01-17 | 鹏城实验室 | Data packet transmission method and related device |
| CN113452559B (en) * | 2021-06-24 | 2022-03-08 | 同济大学浙江学院 | Network resource allocation method, system and medium based on deep learning |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104518984A (en) * | 2014-12-22 | 2015-04-15 | 北京航空航天大学 | SDN controller for carrying out traffic classification on data package based on multiple classifiers |
| CN104994033A (en) * | 2015-05-13 | 2015-10-21 | 南京航空航天大学 | Method for guaranteeing QoS (quality of service) of SDN (software defined network) by means of dynamic resource management |
| CN107317758A (en) * | 2017-06-12 | 2017-11-03 | 西安交通大学 | A kind of fine granularity SDN traffic monitoring frameworks of high reliability |
| CN107547379A (en) * | 2016-06-23 | 2018-01-05 | 华为技术有限公司 | The method and relevant device of route test action are generated in software defined network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9967188B2 (en) * | 2014-10-13 | 2018-05-08 | Nec Corporation | Network traffic flow management using machine learning |
-
2018
- 2018-06-06 CN CN201810573044.1A patent/CN108881028B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104518984A (en) * | 2014-12-22 | 2015-04-15 | 北京航空航天大学 | SDN controller for carrying out traffic classification on data package based on multiple classifiers |
| CN104994033A (en) * | 2015-05-13 | 2015-10-21 | 南京航空航天大学 | Method for guaranteeing QoS (quality of service) of SDN (software defined network) by means of dynamic resource management |
| CN107547379A (en) * | 2016-06-23 | 2018-01-05 | 华为技术有限公司 | The method and relevant device of route test action are generated in software defined network |
| CN107317758A (en) * | 2017-06-12 | 2017-11-03 | 西安交通大学 | A kind of fine granularity SDN traffic monitoring frameworks of high reliability |
Non-Patent Citations (1)
| Title |
|---|
| "vTC：Machine Learning Based Traffic Classification as a Virtual Network Function";Lu He，等;《SDN-NFV Security "16: Proceedings of the 2016 ACM International Workshop on Security in Software Defined Networks & Network Function Virtualization》;20160331;第2-4部分 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108881028A (en) | 2018-11-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108881028B (en) | SDN network resource scheduling method for realizing application awareness based on deep learning | |
| US11356869B2 (en) | Preservation of policy and charging for a subscriber following a user-plane element failover | |
| Hamdan et al. | Flow-aware elephant flow detection for software-defined networks | |
| US8149705B2 (en) | Packet communications unit | |
| JP6162337B2 (en) | Application-aware network management | |
| CN101399749B (en) | Method, system and device for packet filtering | |
| WO2021052379A1 (en) | Data stream type identification method and related devices | |
| CN104320358A (en) | QoS (Quality of Service) business control method in power telecommunication net | |
| Chung et al. | Machine learning based path management for mobile devices over MPTCP | |
| Xu et al. | IARA: An intelligent application-aware VNF for network resource allocation with deep learning | |
| CN106550241A (en) | Video traffic identifying system and virtualization dispositions method | |
| Mongkolluksamee et al. | Combining communication patterns & traffic patterns to enhance mobile traffic identification performance | |
| US9444680B2 (en) | Performing value and context aware communications networking | |
| Alshammari et al. | How robust can a machine learning approach be for classifying encrypted VoIP? | |
| CN113382039A (en) | Application identification method and system based on 5G mobile network flow analysis | |
| US20160248652A1 (en) | System and method for classifying and managing applications over compressed or encrypted traffic | |
| Amelyanovich et al. | Centralized control of traffic flows in wireless LANs based on the SDN concept | |
| JP2007228217A (en) | Traffic decision device, traffic decision method, and program therefor | |
| Al-Saadi et al. | Unsupervised machine learning-based elephant and mice flow identification | |
| Shadad et al. | Efficient and Reliable Management of 5G Network Slicing based on Deep Learning | |
| Obasi et al. | An experimental study of different machine and deep learning techniques for classification of encrypted network traffic | |
| Calderón et al. | Predicting traffic through artificial neural networks | |
| Zaw | Delay-Aware Elephant Flow Rerouting in Software-Defined Networking (SDN) | |
| An et al. | Evaluating SIP-based VoIP communication quality and network security | |
| RU181257U1 (en) | Data Clustering Firewall |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |