CN114531388A - Communication method and device - Google Patents
Communication method and device Download PDFInfo
- Publication number
- CN114531388A CN114531388A CN202011193732.9A CN202011193732A CN114531388A CN 114531388 A CN114531388 A CN 114531388A CN 202011193732 A CN202011193732 A CN 202011193732A CN 114531388 A CN114531388 A CN 114531388A
- Authority
- CN
- China
- Prior art keywords
- communication
- communication node
- node
- message data
- 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.)
- Pending
Links
Images
Classifications
-
- 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
- H04L45/02—Topology update or discovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/28—Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
- H04L12/2854—Wide area networks, e.g. public data networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
-
- 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
- H04L45/12—Shortest path evaluation
- H04L45/121—Shortest path evaluation by minimising delays
-
- 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
- H04L45/12—Shortest path evaluation
- H04L45/124—Shortest path evaluation using a combination of metrics
Abstract
The embodiment of the invention discloses a communication method and a device, wherein the method comprises the following steps: the method comprises the steps that a first communication node determines message data, wherein the first communication node is a communication node in a federally learned communication network, and a connection relation is established between any two communication nodes in the federally learned communication network. The first communication node sends message data to a second communication node through the federally learned communication network, wherein the second communication node is a communication node in the federally learned communication network. According to the invention, the connection relation is established between any two communication nodes in the federally learned communication network, so that the first communication node can have a plurality of optional communication paths when sending the message data to the second communication node, thereby avoiding the problem of message data transmission failure caused by the failure of some network nodes and effectively improving the communication stability.
Description
Technical Field
The embodiment of the invention relates to the field of artificial intelligence, in particular to a communication method and a communication device.
Background
In a federal learning scenario, different enterprises and institutions may build their own clusters and form a federation to communicate with each other.
At present, different enterprises and public institutions allow access to specific communication nodes of each other to realize communication between the enterprises and public institutions, and access to other communication nodes and computing nodes is not allowed in the communication process.
However, in the serial connection or tree connection, when a communication node fails, the entire federal system is disabled, resulting in poor communication stability.
Disclosure of Invention
Embodiments of the present invention mainly aim to provide a communication method and apparatus, which aim to improve communication stability.
In order to achieve the above object, the present invention provides a communication method, including:
a first communication node determines message data, wherein the first communication node is a communication node in a federally learned communication network, and a connection relation is established between any two communication nodes in the federally learned communication network;
and the first communication node sends the message data to a second communication node through the federally learned communication network, wherein the second communication node is a communication node in the federally learned communication network.
In one possible implementation manner, the sending, by the first communication node, the message data to the second communication node through the federally-learned communication network includes:
the first communication node acquires communication parameters of a plurality of first communication paths, wherein the starting point of each first communication path is the first communication node, and the end point of each first communication path is the second communication node;
the first communication node determines at least one target communication path according to the communication parameters of the first communication paths;
and the first communication node sends the message data to the second communication node through the at least one target communication path.
In one possible implementation, the communication parameter includes at least one of: packet loss rate, delay time length and retransmission times;
the first communication node determines at least one target communication path according to the communication parameters of each first communication path, and the method comprises the following steps:
the first communication node determines a first communication path meeting a first condition as the target communication path to obtain a plurality of target communication paths;
wherein the first condition comprises at least one of: the packet loss rate is smaller than a first threshold, the delay time length is smaller than a second threshold, and the retransmission times is smaller than a third threshold.
In a possible implementation manner, the sending, by the first communication node, the packet data to the second communication node through the at least one target communication path includes:
the first communication node divides the message data to obtain a plurality of segments of the message data;
and the first communication node sends the plurality of segments of the message data to the second communication node through the plurality of target communication paths respectively.
In one possible implementation, the communication parameter includes at least one of: packet loss rate, delay time length and retransmission times;
the first communication node determines at least one target communication path according to the communication parameters of each first communication path, and the method comprises the following steps:
the first communication node determines a first communication path meeting a second condition as the target communication path to obtain the target communication path;
wherein the second condition comprises at least one of: the packet loss rate is the lowest, the delay time length is the shortest, and the retransmission times are the least.
In a possible implementation manner, the sending, by the first communication node, the packet data to the second communication node through the at least one target communication path includes:
and the first communication node sends the message data to the second communication node through the determined target communication path.
In one possible implementation, the method further includes:
determining a standby node for at least one of the communication nodes;
and if any communication node fails or the load of any communication node is higher than a preset threshold value, switching part of data streams or all data streams on the communication nodes to the standby node for transmission.
The present invention also provides a communication apparatus comprising:
the determining module is used for determining message data by a first communication node, wherein the first communication node is a communication node in a federally-learned communication network, and a connection relation is established between any two communication nodes in the federally-learned communication network;
a sending module, configured to send, by the first communication node, the packet data to a second communication node through the federally-learned communication network, where the second communication node is a communication node in the federally-learned communication network.
The present invention also provides a communication device, comprising: a memory, a processor and a communication program stored on the memory and executable on the processor, the communication program, when executed by the processor, implementing the steps of the communication method according to any of the preceding claims.
The invention also provides a computer readable storage medium having stored thereon a communication program which, when executed by a processor, implements the steps of the communication method as set forth in any one of the preceding claims.
In the invention, message data is determined through a first communication node, wherein the first communication node is a communication node in a federally-learned communication network, and a connection relation is established between any two communication nodes in the federally-learned communication network. The first communication node sends message data to a second communication node through the federally learned communication network, wherein the second communication node is a communication node in the federally learned communication network. The connection relation is established between any two communication nodes in the communication network which is set for federal learning, so that the first communication node can have a plurality of optional communication paths when sending message data to the second communication node, the problem that the message data transmission fails due to faults of some network nodes can be avoided, and the communication stability is effectively improved.
Drawings
FIG. 1 is a diagram illustrating a communication node in a federated learning network provided by an embodiment of the present invention;
fig. 2 is a schematic diagram illustrating a connection mode of a current communication node according to an embodiment of the present invention;
fig. 3 is a flowchart of a communication method according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a connection relationship between network nodes according to an embodiment of the present invention;
fig. 5 is a flowchart of a communication method according to another embodiment of the present invention;
fig. 6 is a schematic diagram illustrating an implementation of segmented transmission of message data according to an embodiment of the present invention;
fig. 7 is a schematic diagram illustrating another implementation of segmented transmission of message data according to an embodiment of the present invention;
fig. 8 is a flowchart of a communication method according to another embodiment of the present invention;
fig. 9 is a schematic diagram illustrating an implementation of whole segment transmission of message data according to an embodiment of the present invention;
fig. 10 is a schematic diagram illustrating an implementation of a standby node according to an embodiment of the present invention;
fig. 11 is a schematic diagram of a standby node when a communication node fails according to an embodiment of the present invention;
fig. 12 is a schematic diagram of an operation of a standby node when a communication node is overloaded according to an embodiment of the present invention;
fig. 13 is a schematic structural diagram of a communication device according to an embodiment of the present invention;
fig. 14 is a schematic structural diagram of a communication device according to an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
In order to better understand the technical scheme of the invention, firstly, the federal study is briefly introduced:
federal learning (Federated bone learning/Federated learning) refers to a machine learning framework, which can effectively help a plurality of nodes (representing individuals or organizations) to jointly train a machine learning or deep learning model under the condition of meeting the requirement of data privacy protection.
As a novel machine learning concept, federal learning ensures that user privacy data is protected to the maximum extent through distributed training and encryption technology so as to promote trust of users on artificial intelligence technology. Under a federal learning mechanism, each participant contributes an encrypted data model to a federation, jointly trains a federal model, and opens the federal model for each participant to use. In the training process of the federal learning, how to improve the interaction efficiency of the participants of the federal learning and the federal learning device is of great significance for improving the model training efficiency of the federal learning.
The federated learning models machine learning by combining different participants (or party, also called data owner or client), and in the process of federated learning, the participants do not need to expose own data to other participants and coordinators (coordinator, also called server), parameter server or aggregation server), so that the federated learning can well protect user privacy and guarantee data security, and can solve the data islanding problem.
The federate can also carry out model training to obtain model parameters under the condition that own data are not required to be given by both federate and federate, the federate can protect the privacy of user data through a parameter exchange mode under an encryption mechanism, the data and the model can not be transmitted, and the data of the other party can not be guessed reversely, so that the possibility of leakage does not exist in the data layer, and the data privacy can be ensured while the integrity of the data is kept at a higher degree.
It can be understood that each party of the federation needs to perform data interaction, and during the data interaction, a person or an organization needing to perform data access only allows access to a specific communication node of the other party, which is described below with reference to fig. 1, where fig. 1 is a schematic diagram of a communication node in a network for federal learning provided in an embodiment of the present invention.
Taking a participant in federal learning as an enterprise and public institution as an example, as shown in fig. 1, in a federal learning scenario, different enterprise and public institutions may build respective clusters to form a federation, for example, an enterprise and public institution 1 and an enterprise and public institution 2 exist in fig. 1, for example, a cluster X and a cluster Y may be built in the enterprise and public institution 1, and a cluster Z may be built in the enterprise and public institution 2, for example, the enterprise and public institution 1 and the enterprise public institution 2 and the built clusters thereof may form the federation.
Communication and communication isolation are realized between different enterprises and public institutions, so that except a specific communication node, the other communication node and the computing node are not allowed to access in federal learning.
Currently, inside a communication node, a software module dedicated to communication may be deployed, and the software module is used for network forwarding, isolating irrelevant network data, and isolating different clusters.
For example, referring to fig. 1, currently, each cluster includes a computing node and an intra-cluster communication node, and assuming that a cluster X needs to interact with a cluster Z, the cluster X may send data that needs to interact to an intermediate communication node in an enterprise 1 through the intra-cluster communication node of the cluster X, and then the intermediate communication node in the enterprise 1 may send the data to an intermediate communication node of an enterprise 2, and the intermediate communication node of the enterprise 2 sends the data to the intra-cluster communication node of the cluster Z, so that data interaction between the cluster X and the cluster Z is implemented.
It can be understood that, different enterprises and public institutions can only access the intermediate communication node of the other party, for example, for the enterprise and public institution 1, only the intermediate communication node of the enterprise and public institution 2 can be accessed, the intra-cluster communication node of the cluster Z built in the enterprise and public institution 2 is not allowed to be accessed, and the computing node in the cluster Z is not allowed to be accessed, so that communication isolation between different enterprises and public institutions can be realized, and data security is effectively ensured.
Currently, in the related art, serial connection or tree connection is adopted between each communication node, and a network between each enterprise and public institution is connected through a wide area network, for example, see fig. 2, and fig. 2 is a schematic diagram of a connection mode of a current communication node provided by an embodiment of the present invention.
As shown in fig. 2, it is assumed that there are currently 6 communication nodes, which are communication node a, communication node B, communication node C, communication node D, communication node E, and communication node F, where a serial connection relationship is established among communication node a, communication node B, and communication node C, a serial connection relationship is established among communication node D, communication node E, and communication node F, and a connection relationship is also established between communication node B and communication node E.
However, in the above connection relationship, if a communication node fails, an abnormality may occur in communication, for example, the current communication node a needs to send data to the communication node C, and if the communication node B fails, data may not be transmitted.
It can be understood that, the above describes a simpler serial connection mode with fewer communication nodes, and as the federal establishment increases, the number of the communication nodes connected in series or in a tree form also increases, and as more and more clusters access a communication node, the resource of the communication node may be close, and the communication node may malfunction.
When a communication node in the federal learning network fails, the whole federal may be paralyzed, so that the connection mode of serial connection or tree connection of the communication nodes in the prior art can cause poor communication stability.
Meanwhile, because the networks between the enterprises and public institutions are usually connected by the wide area network, the problems of low network speed and long time exist, and then network jitter, packet loss and other phenomena can be caused, so that the performance of federal learning is restricted, and in the prior art, when the packet loss problem occurs, the overtime retransmission is usually carried out through a software module which is specially used for federal communication in a communication node, so that the original network performance is slower in the speed of the wide area network, and the communication speed is low.
In view of the above-mentioned problems in the prior art, the present invention proposes the following technical concepts: the connection relation is established among all communication nodes in the federal learned network, so that a full-connection network is formed, the number of optional communication paths can be ensured to be large, when one or more communication nodes break down, the communication paths without the broken communication nodes can be selected for data transmission, data transmission failure caused by the communication nodes breaking down can be effectively avoided, and the stability of communication is improved.
The communication method provided by the present invention is described below with reference to specific embodiments, fig. 3 is a flowchart of the communication method provided by the embodiment of the present invention, and fig. 4 is a schematic diagram of a connection relationship between network nodes provided by the embodiment of the present invention.
As shown in fig. 3, the method includes:
s301, the first communication node determines message data, wherein the first communication node is a communication node in a federally learned communication network, and a connection relation is established between any two communication nodes in the federally learned communication network.
In this embodiment, a connection relationship is established between any two communication networks in the federally learned communication network, and in a possible implementation, the connection manner of each communication node in the federally learned communication network may be, for example, as shown in fig. 4, which is shown in fig. 4:
assuming that there are 6 communication nodes, namely, a communication node a, a communication node B, a communication node C, a communication node D, a communication node E, and a communication node F, referring to fig. 6, a connection relationship is established between any two communication nodes, and in a possible implementation manner, after a connection relationship is established between any two communication nodes, a peer-to-peer (P2P) full-connection network may be formed.
It is understood that the connection between two communication nodes is logical, not physical, and on the current Internet, the connection between the communication nodes may refer to network connectivity between Transmission Control protocols/Internet protocols (TCP/IP).
In this embodiment, a plurality of interactions may be performed between communication nodes in the federally learned communication network, where the first communication node is a communication node in the federally learned communication network, and in a possible implementation manner, the first communication node may be any one communication node that needs to perform data transmission.
For example, if the current communication node a needs to perform data transmission, the communication node a may serve as a first communication node to determine message data that needs to be transmitted; or, if the current communication node B needs to perform data transmission, the communication node B may serve as a first communication node, and determine the packet data that needs to be transmitted.
The message data transmitted by the first communication node may be, for example, the first transmitted message data, or may also be the message data retransmitted after the packet loss occurs, which is not particularly limited in this embodiment.
S302, the first communication node sends message data to a second communication node through the federally learned communication network, wherein the second communication node is the communication node in the federally learned communication network.
The first communication node may send the message data to the second communication node, where the second communication node may also be a communication node in a federally-learned communication network, and it can be understood that the second communication node is a destination node of the first communication node for data transmission, and this embodiment also does not specifically limit which node is a specific node of the second communication node.
In this embodiment, because a connection relationship is established between any two nodes in the federally-learned communication network, when the first communication node sends the message data to the second communication node through the federally-learned communication network, many communication paths may be available for selection.
For example, referring to fig. 4, assuming that the current communication node a needs to transmit data to the communication node B, the communication node a may serve as a first communication node and the communication node B may serve as a second communication node, where the communication node a may transmit the message data through the path a → B, or may transmit the message data through the path a → C → B; alternatively, the message data may be transmitted in a → D → E → B manner, wherein the communication node a may select a plurality of communication paths.
Therefore, based on the technical scheme of the embodiment, even if some network nodes fail, the first communication node may still select the remaining communication paths, and the selected communication paths may not include the failed network nodes, so that the influence of the failure of some network nodes on the transmission of the message data may be avoided.
The communication method provided by the embodiment of the invention comprises the following steps: the first communication node determines message data, wherein the first communication node is a communication node in a federally learned communication network, and a connection relation is established between any two communication nodes in the federally learned communication network. The first communication node sends message data to a second communication node through a federally-learned communication network, wherein the second communication node is a communication node in the federally-learned communication network. The connection relation is established between any two communication nodes in the communication network which is set for federal learning, so that the first communication node can have a plurality of optional communication paths when sending message data to the second communication node, the problem that the message data transmission fails due to faults of some network nodes can be avoided, and the communication stability is effectively improved.
On the basis of the above embodiment, because the connection network performance between different communication nodes in the wide area network is different, the communication quality between some federates is poor, the packet loss rate is high, and the communication delay is high.
Meanwhile, in the process of transmitting the message data, the first communication node may divide the message data into a plurality of segments, and then send the message data of the plurality of segments to the second communication node through the selected plurality of communication paths, respectively; or, the first communication node may also directly send the message data to the second communication node through the selected communication path.
The following describes possible implementations of packet data transmission from the first communication node to the second communication node, which are introduced herein, with reference to specific embodiments.
First, an implementation manner that a first communication node divides message data into a plurality of segments to send is described with reference to fig. 5 and fig. 6, fig. 5 is a flowchart of a communication method according to another embodiment of the present invention, fig. 6 is an implementation schematic diagram of segmented transmission of message data according to an embodiment of the present invention, and fig. 7 is another implementation schematic diagram of segmented transmission of message data according to an embodiment of the present invention.
As shown in fig. 5, the method includes:
s501, a first communication node acquires communication parameters of a plurality of first communication paths, wherein the starting point of each first communication path is a first communication node, and the end point of each first communication path is a second communication node.
In this embodiment, the first communication path is a link between the first communication node and the second communication node, and the embodiment does not limit the communication nodes specifically included in the first communication path as long as the starting point of the first communication path is the first communication node and the end point of the first communication path is the second communication node.
It is to be understood that, in the present embodiment, a connection relationship is established between each communication node of the federally-learned communication network, and therefore, a plurality of first communication paths may exist between the first communication node and the second communication node.
The first communication node may obtain communication parameters of a plurality of first communication paths, where the communication parameters include at least one of: in the actual implementation process, the specific implementation of the communication parameters may also be selected and expanded according to actual requirements, and any parameter used for indicating the communication quality may be used as the communication parameter in this embodiment.
In a possible implementation manner, during the data transmission process of each communication node, each communication node may record, for example, communication parameters of each communication path during the data transmission, and store the recorded communication parameters, so that when a communication path needs to be selected according to the communication parameters, the communication parameters of the communication path may be directly acquired.
The first communication node may acquire the communication parameters of all the first communication paths, or the first communication node may acquire only part of the communication parameters of the first communication paths.
S502, the first communication node determines the first communication path meeting the first condition as a target communication path to obtain a plurality of target communication paths.
Wherein the first condition comprises at least one of: the packet loss rate is smaller than a first threshold, the delay time length is smaller than a second threshold, and the retransmission times is smaller than a third threshold.
In this embodiment, after determining the communication parameters of each first communication path, the first communication node may select a target path for packet data transmission based on the communication parameters of each first communication path.
In this embodiment, because the first communication node needs to send the packet data in segments, in order to improve the transmission rate and security, the first communication node may determine, for example, multiple target communication paths for sending the segmented packet data respectively.
The first communication node may determine, for example, a first communication path satisfying a first condition as a target communication path, where setting of the first condition may select a plurality of target communication paths.
In a possible implementation manner, the first communication node may determine, for example, a first communication path with a packet loss rate smaller than a first threshold as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with a delay duration smaller than a second threshold as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with a retransmission number smaller than a third threshold as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path whose packet loss rate is smaller than a first threshold and whose delay duration is smaller than a second threshold, as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path in which the packet loss rate is smaller than a first threshold and the number of retransmissions is smaller than a third threshold, as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path whose delay duration is less than a second threshold and whose retransmission times are less than a third threshold, as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with a packet loss rate smaller than a first threshold, a delay duration smaller than a second threshold, and a retransmission number smaller than a third threshold, as the target communication path.
The specific implementation of the first threshold, the second threshold, and the third threshold may be selected according to actual requirements, which is not particularly limited in this embodiment.
It can be understood that, what has been described above is an implementation manner in which the communication parameters include a packet loss rate, a delay time length, and a retransmission number, and when the communication parameters include other parameters, the specific implementation of the first condition may also be selected according to actual requirements, as long as the setting of the first condition is to select a communication path with better communication quality from the plurality of first communication paths.
S503, the first communication node divides the message data to obtain a plurality of segments of the message data.
In a possible implementation manner, the first communication node may, for example, divide the message data into a preset number of segments, so as to obtain a plurality of segments of the message data; or, the first communication node may also divide the packet data into a plurality of segments according to the preset number of bytes, and the specific implementation manner of dividing the packet data by the first communication node is not limited in this embodiment, as long as the plurality of segments of the packet data can be obtained after division, where the number of the divided segments and the length of each segment may be selected according to actual requirements.
For example, referring to fig. 6, it is assumed that after the message data is divided, 3 equal-length segments of the message data, namely segment 1, segment 2, and segment 3, can be obtained.
S504, the first communication node sends a plurality of fragments of message data to the second communication node through a plurality of target communication paths respectively.
In this embodiment, the first communication node divides the packet data into a plurality of segments, and determines a plurality of target communication paths at the same time, so that the first communication node may send the plurality of segments of the packet data to the second communication node through the plurality of target communication paths, respectively.
In a possible implementation manner, for example, referring to fig. 6, the assumption of the above embodiment is also followed, assuming that there are 6 communication nodes currently, namely, communication node a, communication node B, communication node C, communication node D, communication node E, and communication node F, and the connection relationship between the respective communication nodes may refer to the description in the above embodiment, and assuming that the current communication node a transmits data to the communication node B.
In one possible implementation, assuming that the communication node a selects a 3-entry-labeled communication path satisfying the first condition among the communication paths between the communication node a and the communication node B, which are a → B, A → C → B, A → D → E → B, respectively, shown in fig. 6, the 3 pieces can be transmitted through the three target communication paths, respectively.
For example, the segment 1 may be transmitted through the communication path a → B, the segment 2 may be transmitted through the communication path a → C → B, and the segment 3 may be transmitted through the communication path a → D → E → B.
Fig. 6 illustrates that the number of the determined target communication paths is the same as the number of the divided segments of the message data, and in another possible implementation manner, the number of the target communication paths may be different from the number of the divided segments of the message data.
For example, see fig. 7, the above assumption is also followed, and it is assumed that the current communication node a transmits data to the communication node B.
In one possible implementation, assuming that the communication node a selects 2-entry-label communication paths satisfying the first condition among the communication paths between the communication node a and the communication node B, which are respectively a → B, A → C → B shown in fig. 7, 3 pieces can be respectively transmitted through the 2-entry-label communication paths.
For example, the segment 1 and the segment 2 may be transmitted through the communication path a → B, and the segment 3 may be transmitted through the communication path a → C → B.
In the actual implementation process, the specific division of the message data, the specific selection of the target communication path, and which communication path is specifically adopted to send which segment may all be selected according to the actual requirement, which is not particularly limited in this embodiment.
Meanwhile, it can be understood that the second communication node can obtain the complete message data sent by the first communication node after receiving the plurality of message data fragments.
In this embodiment, the message data is divided into a plurality of segments, and the plurality of segments are sent through the determined plurality of target communication paths, so that the data transmission speed can be effectively increased, and meanwhile, the message data is sent in segments, so that only part of the message data is transmitted on each communication path, and the security of message data transmission can be further ensured.
Meanwhile, the target communication path is selected according to the first condition in the embodiment, so that the communication quality during data transmission according to the target communication path can be effectively ensured, and the stability of data transmission is effectively improved.
In another possible implementation manner of the present invention, the first communication node may also directly transmit the packet data without segmenting the packet data, so that only an optimal communication path needs to be selected, and the current implementation manner is described below with reference to fig. 8 and 9.
Fig. 8 is a flowchart of a communication method according to another embodiment of the present invention, and fig. 9 is a schematic diagram of implementing whole segment transmission of message data according to the embodiment of the present invention.
As shown in fig. 8, the method includes:
s801, a first communication node acquires communication parameters of a plurality of first communication paths, where a start point of each first communication path is a first communication node, and an end point of each first communication path is a second communication node.
The implementation manner of S801 is similar to that of S501, and is not described herein again.
S802, the first communication node determines the first communication path meeting the second condition as a target communication path to obtain a target communication path.
Wherein the second condition comprises at least one of: the packet loss rate is lowest, the delay time is shortest and the retransmission times are least.
In this embodiment, because the first communication node does not need to send the packet data in segments, but directly sends the entire segment of packet data, in order to improve the stability of transmission, the first communication node directly selects the communication path with the best communication quality as the target communication path.
The first communication node may determine, for example, a first communication path satisfying a second condition as a target communication path, where the setting of the second condition may select a target communication path with the best communication quality.
In a possible implementation manner, the first communication node may determine, for example, a first communication path with a lowest packet loss rate as a target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with the shortest delay duration as a target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with the least number of retransmissions as a target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with the lowest packet loss rate and the shortest delay duration as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with the lowest packet loss rate and the smallest retransmission times as a target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with the shortest delay duration and the smallest retransmission times as the target communication path;
in another possible implementation manner, the first communication node may determine, for example, a first communication path with the lowest packet loss rate, the shortest delay time, and the smallest retransmission times as the target communication path.
It can be understood that, what has been described above is an implementation manner in which the communication parameters include a packet loss rate, a delay time, and a retransmission number, and when the communication parameters include other parameters, the specific implementation of the second condition may also be selected according to actual requirements, as long as the second condition is set to select a communication path with the best communication quality from the plurality of first communication paths.
And S803, the first communication node sends the message data to the second communication node through the determined target communication path.
In this embodiment, the first communication node does not need to send the packet data in segments, but directly sends the entire segment of packet data, so that the first communication node can directly send the packet data to the second communication node through the determined target communication path.
In one possible implementation, see for example fig. 9, the above-mentioned assumption is also followed, and it is assumed that the current communication node a transmits data to the communication node B.
In a possible implementation manner, assuming that the communication node a selects a 1-entry-label communication path satisfying the second condition, that is, a → C → B shown in fig. 9, from among communication paths between the communication node a and the communication node B, the message data may be directly transmitted through the entry-label communication path.
In an actual implementation process, the specific selection of the target communication path may be selected according to actual requirements, as long as the target communication path is the communication path with the best communication quality selected according to the second condition.
In this embodiment, a target communication path with the best communication quality is selected from the plurality of first communication paths according to the second condition, and the message data is directly transmitted through the target communication path, so that the communication quality during data transmission according to the target communication path can be effectively ensured, and the stability of data transmission is effectively improved.
On the basis of the foregoing embodiment, in another possible implementation manner, the communication nodes in the federal learning network have a data caching function in addition to forwarding the packet, so that a backup node may be added in advance, or a backup node may be added when a certain communication node fails, and when a certain communication node fails or a load of a certain communication node is high, for example, higher than a preset threshold, a part of data streams or all data streams on the current communication node may be switched to the backup node for transmission.
Fig. 10 is a schematic diagram illustrating an implementation of a standby node according to an embodiment of the present invention, fig. 11 is a schematic diagram illustrating a working of the standby node when a communication node according to the embodiment of the present invention fails, and fig. 12 is a schematic diagram illustrating a working of the standby node when a load of the communication node according to the embodiment of the present invention is too high.
As shown in fig. 10, it is assumed that there are 6 communication nodes, which are a communication node a, a communication node B, a communication node C, a communication node D, a communication node E, and a communication node F, respectively, and a connection relationship is established between any two communication nodes, for example, a standby node D 'may be set for the communication node D currently, where the standby node D' may be set in advance, for example, or may be set when the communication node D fails, which is not limited in this embodiment.
Referring to fig. 10, it can be determined that the standby node D' also establishes a connection relationship with each communication node, so as to subsequently replace or assist the communication node D in data transmission.
In an actual implementation process, specifically, which communication node is set with the standby node, and the number of the standby nodes set for each communication node may be selected according to actual requirements, for example, two standby nodes may also be set for the communication node D currently, which is not limited in this embodiment.
In one possible implementation manner, the standby node may, for example, replace the failed communication node for data transmission when the corresponding communication node fails, that is, all data streams on the failed communication node are put on the standby node for transmission.
For example, referring to fig. 11, assume that the first communication node a currently needs to send packet data to the first communication node B, and that the first communication node a determines that the target communication path includes a → D → B.
If the communication node D fails during the data transmission process, the first communication node a may switch all the data streams originally transmitted via the communication node D to the standby node D' for transmission, so as to avoid the influence of the failure of the communication node D on the data transmission, and effectively ensure the stability of the communication.
In another possible implementation manner, the standby node may further assist the communication node with the overload to transmit data when the load of the corresponding communication node is too high, that is, part of the data stream of the communication node with the overload is diverted to the standby node for transmission.
For example, referring to fig. 12, assume that the first communication node a currently needs to send packet data to the first communication node B, and that the first communication node a determines that the target communication path includes a → D → B.
If the load of the communication node D is too high during data transmission, for example, the load is higher than a preset threshold, at this time, the first communication node a may split a part of the data stream on the communication node D to the standby node D' for transmission, so as to avoid that the data transmission speed is slow due to the too high load of the communication node D, and effectively improve the data transmission speed.
In this embodiment, by determining a standby node of at least one communication node, if any one of the communication nodes fails or a load of any one of the communication nodes is higher than a preset threshold, a part of or all of data streams on the communication node are switched to the standby node for transmission, and through cooperative work of the standby node, when a problem occurs in the work of the communication node, smooth switching without interruption can be achieved, so that stability and speed of communication are ensured.
In summary, the communication method provided by the present invention has the following beneficial effects:
because the communication nodes for external communication for federal learning are usually deployed on servers, and the communication in federal learning mainly employs a wide area network, the performance bottleneck is mainly that the network transmission performance of the wide area network is low, and the computation and storage capacity of the servers themselves is generally free. Therefore, in the invention, the connection relation is established between any two communication nodes in the federally learned communication network, so that the idle computing power or storage space can be effectively utilized to forward and cache the message of each communication node in the distributed network, and the network transmission performance is improved.
And, since the service is distributed among the respective communication nodes, even if some of the nodes or the network is destroyed, the influence on other parts is small. And based on the full-connection network provided by the invention, the whole topology can be automatically adjusted when part of communication nodes fail, the connectivity of other communication nodes is kept, and the distributed architecture allows the communication nodes to freely join and leave, so that the full-connection network has good expandability.
And the federation learns the communication among different organizations, and the ciphertext is transmitted, but the risk of cracking and attack still exists. In the embodiment, the message data is divided into the plurality of segments, and the message segments are transmitted through the plurality of communication paths, and the transmission of the message data is performed dispersedly among the communication nodes without going through a certain centralized link, so that the possibility that the private information of the user is intercepted and leaked is greatly reduced, and better privacy protection is provided for the user.
Fig. 13 is a schematic structural diagram of a communication device according to an embodiment of the present invention. As shown in fig. 13, the communication apparatus may include:
a determining module 1301, configured to determine packet data by a first communication node, where the first communication node is a communication node in a federally-learned communication network, and a connection relationship is established between any two communication nodes in the federally-learned communication network;
a sending module 1302, configured to send, by the first communication node, the packet data to a second communication node through the federally learned communication network, where the second communication node is a communication node in the federally learned communication network.
In a possible implementation manner, the sending module 1302 is specifically configured to:
the first communication node acquires communication parameters of a plurality of first communication paths, wherein the starting point of each first communication path is the first communication node, and the end point of each first communication path is the second communication node;
the first communication node determines at least one target communication path according to the communication parameters of the first communication paths;
and the first communication node sends the message data to the second communication node through the at least one target communication path.
In one possible implementation, the communication parameter includes at least one of: packet loss rate, delay time length and retransmission times;
in a possible implementation manner, the determining module 1301 is specifically configured to:
the first communication node determines a first communication path meeting a first condition as the target communication path to obtain a plurality of target communication paths;
wherein the first condition comprises at least one of: the packet loss rate is smaller than a first threshold, the delay time length is smaller than a second threshold, and the retransmission times is smaller than a third threshold.
In a possible implementation manner, the sending module 1302 is specifically configured to:
the first communication node divides the message data to obtain a plurality of segments of the message data;
and the first communication node sends the plurality of segments of the message data to the second communication node through the plurality of target communication paths respectively.
In one possible implementation, the communication parameter includes at least one of: packet loss rate, delay time length and retransmission times;
the determining module 1301 is specifically configured to:
the first communication node determines a first communication path meeting a second condition as the target communication path to obtain the target communication path;
wherein the second condition comprises at least one of: the packet loss rate is the lowest, the delay time length is the shortest, and the retransmission times are the least.
In a possible implementation manner, the sending module 1302 is specifically configured to:
and the first communication node sends the message data to the second communication node through the determined target communication path.
In a possible implementation manner, the determining module 1301 is further configured to:
determining a standby node for at least one of the communication nodes;
and if any communication node fails or the load of any communication node is higher than a preset threshold value, switching part of data streams or all data streams on the communication nodes to the standby node for transmission.
Fig. 14 is a schematic structural diagram of a communication device according to an embodiment of the present invention. As shown in fig. 14, the apparatus may include: a memory 1401, a processor 1402 and a communication program stored on the memory 1401 and executable on the processor 1402, which communication program when executed by the processor 1402 implements the steps of the communication method according to any of the preceding embodiments.
Alternatively, the memory 1401 may be separate or integrated with the processor 1402.
For the implementation principle and the technical effect of the device provided by this embodiment, reference may be made to the foregoing embodiments, and details are not described here.
An embodiment of the present invention further provides a computer-readable storage medium, where a communication program is stored, and when executed by a processor, the communication program implements the steps of the communication method according to any one of the foregoing embodiments.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described device embodiments are merely illustrative, and for example, the division of the modules is only one logical division, and other divisions may be realized in practice, for example, a plurality of modules may be combined or integrated into another system, or some features may be omitted, or not executed.
The integrated module implemented in the form of a software functional module may be stored in a computer-readable storage medium. The software functional module is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) or a processor to execute some steps of the methods according to the embodiments of the present invention.
It should be understood that the Processor may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the present invention may be embodied directly in a hardware processor, or in a combination of the hardware and software modules within the processor.
The memory may comprise a high-speed RAM memory, and may further comprise a non-volatile storage NVM, such as at least one disk memory, and may also be a usb disk, a removable hard disk, a read-only memory, a magnetic or optical disk, etc.
The storage medium may be implemented by any type or combination of volatile and non-volatile memory devices, such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.
An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. Of course, the storage medium may also be integral to the processor. The processor and the storage medium may reside in an Application Specific Integrated Circuits (ASIC). Of course, the processor and the storage medium may reside as discrete components in an electronic device or host device.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A method of communication, comprising:
a first communication node determines message data, wherein the first communication node is a communication node in a federally learned communication network, and a connection relation is established between any two communication nodes in the federally learned communication network;
and the first communication node sends the message data to a second communication node through the federally learned communication network, wherein the second communication node is a communication node in the federally learned communication network.
2. The method of claim 1, wherein the first communication node sending the message data to a second communication node over the federally learned communication network comprises:
the first communication node acquires communication parameters of a plurality of first communication paths, wherein the starting point of each first communication path is the first communication node, and the end point of each first communication path is the second communication node;
the first communication node determines at least one target communication path according to the communication parameters of the first communication paths;
and the first communication node sends the message data to the second communication node through the at least one target communication path.
3. The method of claim 2, wherein the communication parameter comprises at least one of: packet loss rate, delay time length and retransmission times;
the first communication node determines at least one target communication path according to the communication parameters of each first communication path, and the method comprises the following steps:
the first communication node determines a first communication path meeting a first condition as the target communication path to obtain a plurality of target communication paths;
wherein the first condition comprises at least one of: the packet loss rate is smaller than a first threshold, the delay time length is smaller than a second threshold, and the retransmission times is smaller than a third threshold.
4. The method of claim 3, wherein the first communication node sending the message data to the second communication node via the at least one target communication path comprises:
the first communication node divides the message data to obtain a plurality of segments of the message data;
and the first communication node sends the plurality of segments of the message data to the second communication node through the plurality of target communication paths respectively.
5. The method of claim 2, wherein the communication parameter comprises at least one of: packet loss rate, delay time length and retransmission times;
the first communication node determines at least one target communication path according to the communication parameters of each first communication path, and the method comprises the following steps:
the first communication node determines a first communication path meeting a second condition as the target communication path to obtain the target communication path;
wherein the second condition comprises at least one of: the packet loss rate is the lowest, the delay time length is the shortest, and the retransmission times are the least.
6. The method of claim 5, wherein the first communication node sending the message data to the second communication node via the at least one target communication path comprises:
and the first communication node sends the message data to the second communication node through the determined target communication path.
7. The method according to any one of claims 1-6, further comprising:
determining a standby node for at least one of the communication nodes;
and if any communication node fails or the load of any communication node is higher than a preset threshold value, switching part of data streams or all data streams on the communication nodes to the standby node for transmission.
8. A communications apparatus, comprising:
the determining module is used for determining message data by a first communication node, wherein the first communication node is a communication node in a federally-learned communication network, and a connection relation is established between any two communication nodes in the federally-learned communication network;
a sending module, configured to send, by the first communication node, the packet data to a second communication node through the federally-learned communication network, where the second communication node is a communication node in the federally-learned communication network.
9. A communication device, characterized in that the communication device comprises: memory, processor and a communication program stored on the memory and executable on the processor, the communication program, when executed by the processor, implementing the steps of the communication method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a communication program which, when executed by a processor, implements the steps of the communication method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011193732.9A CN114531388A (en) | 2020-10-30 | 2020-10-30 | Communication method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011193732.9A CN114531388A (en) | 2020-10-30 | 2020-10-30 | Communication method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114531388A true CN114531388A (en) | 2022-05-24 |
Family
ID=81619754
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011193732.9A Pending CN114531388A (en) | 2020-10-30 | 2020-10-30 | Communication method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114531388A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160379128A1 (en) * | 2015-06-26 | 2016-12-29 | Xerox Corporation | Distributed and privacy-preserving prediction method |
| CN110929880A (en) * | 2019-11-12 | 2020-03-27 | 深圳前海微众银行股份有限公司 | Method and device for federated learning and computer readable storage medium |
| CN111144718A (en) * | 2019-12-12 | 2020-05-12 | 支付宝(杭州)信息技术有限公司 | Risk decision method, device, system and equipment based on private data protection |
| CN111343093A (en) * | 2020-02-28 | 2020-06-26 | 腾讯科技（深圳）有限公司 | Service data transmission method and device |
| CN111600707A (en) * | 2020-05-15 | 2020-08-28 | 华南师范大学 | Decentralized federal machine learning method under privacy protection |
-
2020
- 2020-10-30 CN CN202011193732.9A patent/CN114531388A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160379128A1 (en) * | 2015-06-26 | 2016-12-29 | Xerox Corporation | Distributed and privacy-preserving prediction method |
| CN110929880A (en) * | 2019-11-12 | 2020-03-27 | 深圳前海微众银行股份有限公司 | Method and device for federated learning and computer readable storage medium |
| CN111144718A (en) * | 2019-12-12 | 2020-05-12 | 支付宝(杭州)信息技术有限公司 | Risk decision method, device, system and equipment based on private data protection |
| CN111343093A (en) * | 2020-02-28 | 2020-06-26 | 腾讯科技（深圳）有限公司 | Service data transmission method and device |
| CN111600707A (en) * | 2020-05-15 | 2020-08-28 | 华南师范大学 | Decentralized federal machine learning method under privacy protection |
Non-Patent Citations (1)
| Title |
|---|
| 王亚珅;: "面向数据共享交换的联邦学习技术发展综述", 无人系统技术, no. 06 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3242437B1 (en) | Light-weight key update mechanism with blacklisting based on secret sharing algorithm in wireless sensor networks | |
| US8856317B2 (en) | Secure data transfer in a virtual environment | |
| Syverson et al. | Tor: The secondgeneration onion router | |
| US20180278588A1 (en) | Hardware-accelerated secure communication management | |
| CN110915188B (en) | Method and system for efficiently propagated probabilistic relays in blockchain networks | |
| US11115391B2 (en) | Securing end-to-end virtual machine traffic | |
| US8510562B2 (en) | Content distribution with mutual anonymity | |
| US20210124734A1 (en) | Trusted storage method and system based on directed acyclic graph structure | |
| CN100574249C (en) | virtual router redundancy protocol message transmission method and device | |
| CN101599968B (en) | Reliable anonymous transmission method and system thereof | |
| US9743367B2 (en) | Link layer discovery protocol (LLDP) on multiple nodes of a distributed fabric | |
| CN101277248A (en) | Method and system for distributing network data | |
| CN103118148A (en) | Method and device for updating ARP (address resolution protocol) cache | |
| WO2018068588A1 (en) | Method and software-defined networking (sdn) controller for providing multicast service | |
| WO2017036163A1 (en) | Message sending method and system | |
| EP2126719B1 (en) | Providing distributed convergence nodes in a communication network environment | |
| US10212141B2 (en) | Autonomous key update mechanism with blacklisting of compromised nodes for mesh networks | |
| US20210264051A1 (en) | Blockchain system, blockchain management apparatus, network control apparatus, method and program | |
| CN108200221A (en) | Rule synchronization method and device are converted in a kind of network address translation environment | |
| US9998807B2 (en) | Method and apparatus for establishing trail network | |
| CN114531388A (en) | Communication method and device | |
| CN112491935A (en) | Water wave type broadcasting method and system for block chain | |
| US20160054949A1 (en) | Method for storing data in a computer system performing data deduplication | |
| US7729289B2 (en) | Method, system and computer program product for routing information across firewalls | |
| Nikitinskiy et al. | A stateless transport protocol in software defined networks |
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 |