CN106911526A - Method and system for realizing pressure test - Google Patents
Method and system for realizing pressure test Download PDFInfo
- Publication number
- CN106911526A CN106911526A CN201510968168.6A CN201510968168A CN106911526A CN 106911526 A CN106911526 A CN 106911526A CN 201510968168 A CN201510968168 A CN 201510968168A CN 106911526 A CN106911526 A CN 106911526A
- Authority
- CN
- China
- Prior art keywords
- message
- processing unit
- central processing
- parallel
- graphic process
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
-
- 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
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1441—Countermeasures against malicious traffic
- H04L63/1458—Denial of Service
Abstract
The invention discloses a kind of method and system for realizing pressure test, it is related to field of information security technology.Wherein method includes：First processing units parallel generation message, and message is put into socket buffering area；Second processing unit is by the message asynchronous transmission in socket buffering areas to destination server.Method and system for realizing pressure test of the invention, by graphic process unit parallel generation message, asynchronous transmission message is controlled by central processing unit, improve unit performance utilization rate, the threshold of test is reduced, is easy to dispose concentrating type equipment, lift measurement scope, testing cost is reduced, for test provides possibility extending transversely.
Description
Technical field
The present invention relates to field of information security technology, more particularly to one kind is for realizing pressure test
Method and system.
Background technology
At present, in DDoS (Distributed Denial of Service, distribution refusal clothes
Business) in testing tool, use central processing unit (Central Processing Unit, referred to as：
CPU) carry out the generation of message and send, carry out the simulation of multithreading with software, generation with
The speed for sending message depends on the performance of central processing unit with efficiency.As shown in figure 5, calculating
Central processing unit in machine 501 generates offensive counterfeit message, and then control is located at computer
501 network interface card 5011 sends to destination server 502 message, completes to attack, wherein center
The message of processor is generated as software multithread simulation, rather than producing, it is necessary to repeatedly follow parallel
Ring could complete the message generation of requirement.
Launched a offensive, it is necessary to generate substantial amounts of message in the pressure test of super large system, in
The method of central processor generation message cannot meet test data demand.Current major part
DDoS testing tools depend on the performance of test machine, lack concentrating type deployment and energy extending transversely
Power, effective pressure test, provider server cannot be disposably launched to large-scale service system
Also the performance bottleneck of its own system cannot be known, experience mistaken ideas and potential safety hazard is easily produced.
Graphic process unit (Graphics Processing Unit, referred to as：) and centre GPU
Reason device is compared, with more powerful computation capability, and the logical operation energy of central processing unit
Power is stronger.(Compute Unified Device Architecture, unified calculation sets CUDA
Standby framework) model using central processing unit as main frame, graphic process unit as equipment, by center
Processor is responsible for the strong logical calculated of logicality in host side, and graphic process unit is performed in equipment end
Highly threading parallel processing task.Wherein, central processing unit and graphic process unit have each
From independent memory address space, main frame end memory and equipment end video memory call CUDA to carry
The interface management function of confession can operate video memory to open up, discharge and initialize and in host side
Data transfer and equipment end between etc. is operated.And current CUDA is applied to scientific algorithm
Field is more, lacks the use in pressure test, also fails to play figure in pressure test
The parallel processing capability of processor.For example forged in typical SYN Flood (flood attack)
During TCP SYN messages, not most of field (example of the TCP SYN messages of homologous address
Such as destination address, agreement) it is identical, a source address, source port and verification and difference,
Computational complexity is low, is adapted to be carried out using graphic process unit the generation of message.Therefore how to utilize
The computation capability counterfeit message of graphic process unit is made full use of, it is isobaric to meet super large system
Power test request, is one of technical problems to be solved by the invention.
The content of the invention
The inventors found that above-mentioned have problem in the prior art, and therefore for upper
State at least one of problem problem and propose a kind of new technical scheme.
According to an aspect of the invention, there is provided a kind of method for realizing pressure test,
Including：
First processing units parallel generation message, and it is slow that message is put into socket (socket)
Rush area；
Second processing unit is by the message asynchronous transmission in socket buffering areas to destination server.
In one embodiment, first processing units are the graphics processing unit of CUDA, the
Two processing units are CPU.
In one embodiment, the step of first processing units parallel generation message includes：
Graphics processing unit by set Grid (grid) structures and Block (block) structures come
Parallel line number of passes is set；
Graphics processing unit utilizes parallel thread, circulation generation message.
In one embodiment, message includes heading and corresponding verification and heading
Including the source address forged and the source port of forgery, and meet the phase of message communication agreement inside the Pass
Hold.
In one embodiment, CPU passes through the message in socket buffering areas
Network interface card asynchronous transmission is to destination server.
In one embodiment, graphics processing unit is according to test port number, CPU
Grid structure settings are two-dimensional structure or three-dimensional structure by disposal ability and network interface card disposal ability.
According to another aspect of the present invention, there is provided a kind of system for realizing pressure test,
Including：
First processing units, for parallel generation message, and are put into socket bufferings by message
Area；
Second processing unit, for by the message asynchronous transmission in socket buffering areas to target
Server.
In one embodiment, first processing units are the graphics processing unit of CUDA, the
Two processing units are CPU.
In one embodiment, graphics processing unit, specifically for by setting Grid structures
Parallel line number of passes is set with Block structures；Using parallel thread, circulation generation message.
In one embodiment, message includes heading and corresponding verification and heading
Including the source address forged and the source port of forgery, and meet the phase of message communication agreement inside the Pass
Hold.
In one embodiment, CPU, specifically for by socket buffering areas
Message by network interface card asynchronous transmission to destination server.
In one embodiment, graphics processing unit, specifically for according to test port number, in
Central Processing Unit disposal ability and network interface card disposal ability by Grid structure settings be two-dimensional structure or
Three-dimensional structure.
Method and system for realizing pressure test of the invention, it is parallel by graphic process unit
Generation message, asynchronous transmission message is controlled by central processing unit, improves unit performance utilization rate,
The threshold of test is reduced, is easy to dispose concentrating type equipment, lift measurement scope, reduce test
Cost, for test provides possibility extending transversely.
Brief description of the drawings
Technical scheme in order to illustrate more clearly the embodiments of the present invention, below will to embodiment or
The accompanying drawing to be used needed for description is briefly described, it should be apparent that, it is attached in describing below
Figure is only some embodiments of the present invention, for those of ordinary skill in the art, is not being paid
On the premise of going out creative labor, other accompanying drawings can also be obtained according to these accompanying drawings.
Fig. 1 is the present invention for realizing the flow chart of one embodiment of the method for pressure test.
Fig. 2 is the present invention for realizing the flow chart of another embodiment of the method for pressure test.
Fig. 3 is the present invention for realizing the schematic diagram of one embodiment of the system of pressure test.
Fig. 4 is the present invention for realizing the schematic diagram of another embodiment of the system of pressure test.
Fig. 5 is prior art pressure test schematic flow sheet.
Fig. 6 is pressure test schematic flow sheet of the invention.
Fig. 7 is the design sketch of one embodiment of prior art.
Fig. 8 is the design sketch of one embodiment of the present of invention.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, to the technical scheme in the embodiment of the present invention
It is clearly and completely described, it is clear that described embodiment is only a real part of the invention
Example is applied, rather than whole embodiments.Based on the embodiment in the present invention, ordinary skill
The every other embodiment that personnel are obtained under the premise of creative work is not made, belongs to this
Invent the scope of protection.
Fig. 1 is the flow chart for realizing one embodiment of the method for pressure test of the invention.
As shown in figure 1, the method and step of the present embodiment is as follows：
Step 101, first processing units parallel generation message, and message is put into socket bufferings
Area.
For example, graphic process unit of the first processing units for CUDA frameworks, such as typical
SYN Flood forge TCP SYN messages when, C language source code it is as follows (simplicity for the sake of,
Omit parameter definition and robustness detection of serial code etc.):
It can be seen that, not the TCP SYN messages of homologous address most of field (such as destination address,
Agreement etc.) it is identical, a source address, source port and verification and difference, computational complexity are low,
It is adapted to use graphic process unit parallel generation message, wherein message includes heading and corresponding
Verification and, heading include forge source address and forge source port, and meet message lead to
Believe agreement related content, wherein related content be the affiliated communication protocol of message under make message complete
Required content.
Step 102, second processing unit is by the message asynchronous transmission in socket buffering areas to target
Server.For example, second processing unit is for that can be central processing unit, and for controlling network interface card, will
Message asynchronous transmission in socket buffering areas is to destination server.
In the present embodiment, as shown in fig. 6, the graphic process unit parallel generation report in computer 601
Text, socket buffering areas are put into by message, will be by second processing unit, such as by central processing unit
Central processing unit, by the message in socket buffering areas by the asynchronous transmission of network interface card 6011 to target
In server 602, so as to realize attack test.
Fig. 2 is the flow for realizing another embodiment of the method for pressure test of the invention
Figure.As shown in Fig. 2 the method and step of the present embodiment is as follows：
Step 201, the graphic process unit of CUDA frameworks is by setting Grid structures and Block
Structure sets parallel line number of passes.
Under CUDA models, graphic process unit needs to define a parallel computation function, kernel
(kernel) function undertakes parallel computation.Wherein, kernel functions include the parallel of two levels,
Parallel and Block threads (Thread) is parallel between the Block in respectively Grid.Block
Numbering and thread number are represented with two dimension or three-dimensional array, by setting Grid structures and Block
Structure sets parallel line number of passes.Wherein Block threads are not simultaneously parallel, but by thread
Beam (Warp) is specified, such as in current version, each thread beam is made up of 32 threads, and
Parallel Block numbers can be defined by setting Grid structures.
Step 202, graphic process unit utilizes parallel thread, circulation generation message.
In one embodiment, in one embodiment, in 65535 corresponding to PC
Individual port, when generating the TCP SYN messages of pressure test to each port using CUDA grammers,
It is as follows：
Wherein, 64 Block are provided with a Grid, 32*32 is provided with each Block
Totally 1024 threads, the current CUDA versions thread bundle branch of correspondence holds synchronization each Block
Concurrent thread number is 32.Therefore, can each port of correspondence generation IP reports by 32 circulations
Text, and be stored in socket buffering areas, network interface card is then controlled to socket by central processing unit
Message asynchronous transmission in buffering area is to destination server.Because central processing unit is only used for controlling net
Block and the message in socket buffering areas is sent, so as to reduce the requirement to central processing unit.
Message in socket buffering areas is passed through network interface card asynchronous transmission by step 203, central processing unit
To destination server.
In the present embodiment, 64 Block are only used, each Block runs 1024 lines
Journey.It is 65536 that Grid maximums can accommodate Block numbers in current CUDA3It is individual, if making full use of
Three-dimensional parallel processing capability cook source address produces message, it is clear that test performance can be substantially improved,
But being limited by current socket functions can not parallel be adjusted in CUDA environment by graphic process unit
With, therefore, the message of all generations is intended to converge to principal function, by central processing unit asynchronous transmission.
According to three-dimensional parallel processing manner, the expense of internal memory and video memory, parallel generation can be significantly increased
More massive counterfeit message not only needs to open up bigger socket buffering areas, and parallel time
Also significantly shorten, so that moment message flow will exceed the bearing capacity of single network interface card and bandwidth.Therefore,
In one embodiment, energy is processed according to test port number, central processing unit disposal ability and network interface card
Grid structure settings are two-dimensional structure or three-dimensional structure by power, to play preferably test effect.
Those skilled in the art are by the present invention it will be appreciated that in correlations such as central processing unit, network interface cards
In the case that equipment performance is lifted, parallel Thread Count can be improved, to play graphic process unit
Performance.
Scalability of the invention is strong, after CUDA 5.0, by GPU Direct technologies
Can be by supporting that the network equipment of Infiniband coordinates pipe to concentrating type CUDA equipment
Reason, realize the horizontal lifting of performance, by thread synchronization, can by message transmission frequency with
The Retransmission timeout mechanism of ICP/IP protocol is coupled, and produces the high LDoS of sync rates
(Low-rate Denial-of-Service, low-speed denial of service attack) attack echelons, more can pole
It is big to reduce local resource consumption.
Fig. 3 is the schematic diagram for realizing one embodiment of the system of pressure test of the invention.
As shown in figure 3, including first processing units 301 and second processing unit 302, wherein：
First processing units 301 are used for parallel generation message, and message is put into socket bufferings
Area.For example, graphic process unit of the first processing units for CUDA frameworks.Wherein message includes
Heading and corresponding verification and, heading includes the source address forged and the source forged
Mouthful.The source address of different counterfeit messages, source port and verification and difference.
Second processing unit 302 is used for the message asynchronous transmission in socket buffering areas to target
Server.For example, second processing unit can be central processing unit.
In the present embodiment, by the strong first processing units of parallel processing capability, such as graphic process unit,
Parallel generation message, socket buffering areas are put into by message.By second processing unit, such as center
Processor, by message asynchronous transmission to the destination server in socket buffering areas, so as to realize
Attack test.
Fig. 4 is the schematic diagram for realizing one embodiment of the system of pressure test of the invention.
Including graphic process unit 401 and central processing unit 402, wherein：
Graphic process unit 401 is CUDA frameworks, by setting Grid structures and Block structures
To set parallel line number of passes.
Central processor unit 402 is used for the message in socket buffering areas is asynchronous by network interface card
Send to destination server.
For example, in one embodiment, graphic process unit 401 sets 64 Block in Grid,
32*32 totally 1024 threads, the current CUDA versions thread of correspondence are set in each Block
It is 32 that bundle branch holds each Block concurrent threads number of synchronization.Therefore, it is by 32 circulations
Each port (totally 65536 ports) generation IP messages can be corresponded to, and is stored in socket bufferings
Qu Zhong, then controls network interface card to the message asynchronous transmission in socket buffering areas by central processing unit
To destination server.Because central processing unit is only used for control network interface card to the report in socket buffering areas
Text sends, so as to reduce the requirement to central processing unit.
In one embodiment, it is Intel i3-2120 that experiment porch uses central processing unit, interior
8G is deposited, graphic process unit is Geforce GTX 550Ti, according to prior art, i.e., is only made
Counterfeit message is generated and sent with central processing unit, as shown in fig. 7, being circulated using 16 for
Abscond 65536 port cycle patterns, used time 1138ms, data traffic is about 40.00MB/s.And
Message, central processing unit control network interface card asynchronous transmission message are generated using graphic process unit of the invention
Mode, using following encapsulation, wherein dimBlock (32,32), dim3dimGrid (64,16)
Double-deck parallel schema, i.e. the present embodiment employs two-dimentional Grid structures, common 64*16 Block,
There is 32*32 thread in each Block, parallel Thread Count is 32*64*16 every time, with Fig. 7
Illustrated embodiment is obvious compared to speed lifting, obtains result as shown in figure 8, used time only 25ms,
Data traffic is 1600MB/s, performance boost 46 times, effect is significant.
Method and system for realizing pressure test of the invention, introduces many by CUDA technologies
In thread pressure test, scalability is strong, reduces testing cost, improves test performance,
Having a extensive future in the performance test and security hardening of information technology service system.
One of ordinary skill in the art will appreciate that realizing all or part of step of above-described embodiment
Suddenly can be completed by hardware, it is also possible to instruct the hardware of correlation to complete by program, institute
The program stated can be stored in a kind of computer-readable recording medium, and storage mentioned above is situated between
Matter can be read-only storage, disk or CD etc..
Description of the invention is given for the sake of example and description, and is not exhaustively
Or limit the invention to disclosed form.Common skill of many modifications and variations for this area
It is obvious for art personnel.Selection and description embodiment are to more preferably illustrate principle of the invention
And practical application, and make one of ordinary skill in the art it will be appreciated that the present invention is suitable so as to design
In the various embodiments with various modifications of special-purpose.
Claims (12)
1. a kind of method for realizing pressure test, it is characterised in that including：
First processing units parallel generation message, and the message is put into socket
Buffering area；
Second processing unit takes the message asynchronous transmission in the socket buffering areas to target
Business device.
2. the method according to any one in claim 1, it is characterised in that described
First processing units are the graphic process unit of unified calculation equipment framework CUDA, described second
Processing unit is central processing unit.
3. method according to claim 2, it is characterised in that first processing units are simultaneously
The step of row generation message, includes：
Graphic process unit is set parallel by setting grid Grid structures and block Block structures
Thread Count；
The graphic process unit utilizes the parallel thread, circulation generation message.
4. method according to claim 1, it is characterised in that the message includes report
Literary stem and corresponding verification are with the heading includes the source address forged and the source forged
Port, and meet the related content of the message communication agreement.
5. method according to claim 3, it is characterised in that the central processing unit
By the message in the socket buffering areas by network interface card asynchronous transmission to destination server.
6. method according to claim 5, it is characterised in that
The graphic process unit according to test port number, central processing unit disposal ability and network interface card at
The Grid structure settings are two-dimensional structure or three-dimensional structure by reason ability.
7. a kind of system for realizing pressure test, it is characterised in that including：
First processing units, for parallel generation message, and are put into socket by the message
Socket buffering areas；
Second processing unit, for by the message asynchronous transmission in the socket buffering areas extremely
Destination server.
8. system according to claim 7, it is characterised in that first treatment is single
Unit is the graphic process unit of unified calculation equipment framework CUDA, and the second processing unit is
Central processing unit.
9. system according to claim 8, it is characterised in that the graphic process unit,
Specifically for setting parallel line number of passes by setting grid Grid structures and block Block structures；
Using the parallel thread, circulation generation message.
10. system according to claim 7, it is characterised in that the message includes report
Literary stem and corresponding verification are with the heading includes the source address forged and the source forged
Port, and meet the related content of the message communication agreement.
11. systems according to claim 9, it is characterised in that the central processing unit,
Specifically for the message in the socket buffering areas is taken by network interface card asynchronous transmission to target
Business device.
12. systems according to claim 10, it is characterised in that the graphics process
Device, specifically for according to test port number, central processing unit disposal ability and network interface card disposal ability
It is two-dimensional structure or three-dimensional structure by the Grid structure settings.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510968168.6A CN106911526A (en) | 2015-12-22 | 2015-12-22 | Method and system for realizing pressure test |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510968168.6A CN106911526A (en) | 2015-12-22 | 2015-12-22 | Method and system for realizing pressure test |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106911526A true CN106911526A (en) | 2017-06-30 |
Family
ID=59199400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510968168.6A Pending CN106911526A (en) | 2015-12-22 | 2015-12-22 | Method and system for realizing pressure test |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106911526A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113590440A (en) * | 2021-06-28 | 2021-11-02 | 浪潮商用机器有限公司 | Pressure testing method, device, equipment and medium for bank core system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7738396B1 (en) * | 2002-08-26 | 2010-06-15 | Juniper Networks, Inc. | Network device having accounting service card |
| CN101753315A (en) * | 2008-11-27 | 2010-06-23 | 百度在线网络技术（北京）有限公司 | Method, device and system for testing DDOS (distributed denial of service) attacks |
| CN101778015A (en) * | 2010-01-07 | 2010-07-14 | 中兴通讯股份有限公司 | Micro-code test method and device |
| WO2013062956A1 (en) * | 2011-10-26 | 2013-05-02 | Google Inc. | Automatically testing a program executable on a graphics card |
| CN103309889A (en) * | 2012-03-15 | 2013-09-18 | 华北计算机系统工程研究所 | Method for realizing of real-time data parallel compression by utilizing GPU (Graphic processing unit) cooperative computing |
| CN103684890A (en) * | 2012-08-30 | 2014-03-26 | 中国银联股份有限公司 | Server stress testing method and system |
-
2015
- 2015-12-22 CN CN201510968168.6A patent/CN106911526A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7738396B1 (en) * | 2002-08-26 | 2010-06-15 | Juniper Networks, Inc. | Network device having accounting service card |
| CN101753315A (en) * | 2008-11-27 | 2010-06-23 | 百度在线网络技术（北京）有限公司 | Method, device and system for testing DDOS (distributed denial of service) attacks |
| CN101778015A (en) * | 2010-01-07 | 2010-07-14 | 中兴通讯股份有限公司 | Micro-code test method and device |
| WO2013062956A1 (en) * | 2011-10-26 | 2013-05-02 | Google Inc. | Automatically testing a program executable on a graphics card |
| CN103309889A (en) * | 2012-03-15 | 2013-09-18 | 华北计算机系统工程研究所 | Method for realizing of real-time data parallel compression by utilizing GPU (Graphic processing unit) cooperative computing |
| CN103684890A (en) * | 2012-08-30 | 2014-03-26 | 中国银联股份有限公司 | Server stress testing method and system |
Non-Patent Citations (2)
| Title |
|---|
| IDDIALOG: "异构计算(CPU+GPU)编程简介", 《HTTP://BLOG.CSDN.NET.IDDIALOG/ARTICLE/DETAILS/8078747》 * |
| 王旭等: "CPU/GPU异构计算应用于核电模拟机的可行性", 《计算机应用》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113590440A (en) * | 2021-06-28 | 2021-11-02 | 浪潮商用机器有限公司 | Pressure testing method, device, equipment and medium for bank core system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102650950A (en) | Platform architecture supporting multi-GPU (Graphics Processing Unit) virtualization and work method of platform architecture | |
| CN102929718B (en) | Distributed GPU (graphics processing unit) computer system based on task scheduling | |
| CN107819848A (en) | A kind of internet of things equipment autonomy interconnected method based on block chain | |
| CN104216768B (en) | Data processing method and device | |
| CN108460457A (en) | A kind of more asynchronous training methods of card hybrid parallel of multimachine towards convolutional neural networks | |
| CN104579781A (en) | Smart power grid aggregation method and system for differential privacy security and fault tolerance | |
| CN106951926A (en) | The deep learning systems approach and device of a kind of mixed architecture | |
| CN108898219A (en) | A kind of neural network training method based on block chain, device and medium | |
| Qu et al. | ChainFL: A simulation platform for joint federated learning and blockchain in edge/cloud computing environments | |
| CN111726421B (en) | Method and device for realizing interconnection of network target range and industrial control equipment | |
| CN107085743A (en) | A kind of deep learning algorithm implementation method and platform based on domestic many-core processor | |
| CN109597965A (en) | Data processing method, system, terminal and medium based on deep neural network | |
| CN107239352B (en) | The communication optimization method and its system of a kind of dynamics Monte Carlo Parallel Simulation | |
| CN104023062A (en) | Heterogeneous computing-oriented hardware architecture of distributed big data system | |
| Hsieh et al. | A high-throughput DPI engine on GPU via algorithm/implementation co-optimization | |
| CN108366047A (en) | Active power distribution network data safety high efficiency of transmission optimization method and device based on game theory | |
| CN106371893A (en) | Cloud computing scheduling system and method | |
| CN110941421A (en) | Development machine learning device and using method thereof | |
| CN106911526A (en) | Method and system for realizing pressure test | |
| Zhu et al. | Adfl: Defending backdoor attacks in federated learning via adversarial distillation | |
| CN106990913B (en) | A kind of distributed approach of extensive streaming collective data | |
| CN115277692B (en) | Automatic operation and maintenance method, device and system for edge network computing terminal equipment | |
| Bhat et al. | Experiments with in-transit processing for data intensive grid workflows | |
| Ding et al. | Efficient fitness function computation of genetic algorithm in virtual machine placement for greener data centers | |
| CN111147603A (en) | Method and device for networking reasoning service |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170630 |
|
| RJ01 | Rejection of invention patent application after publication |