WO2012092899A2

WO2012092899A2 - Transfer method and system for peer-to-peer overlay network

Info

Publication number: WO2012092899A2
Application number: PCT/CN2012/071027
Authority: WO
Inventors: 范燕平; 祝宁
Original assignee: 华为技术有限公司
Priority date: 2012-02-10
Filing date: 2012-02-10
Publication date: 2012-07-12
Also published as: CN102714632A; WO2012092899A3

Abstract

The present invention provides a transfer method and system for a peer-to-peer overlay network. The transfer method for the peer-to-peer overlay network is used for searching for a relay node between a source host node and a target host node in the peer-to-peer overlay network which comprises multiple host nodes and a bootstrap server. The transfer method comprises the following steps: acquiring coordinates information by the communication between the host node and the bootstrap servers, wherein the coordinates information is used for calculating the time delay between two host nodes based on a distance prediction mechanism; performing host node clustering to make a part of the host nodes as clustering agent nodes and the other host nodes as the host nodes belonging to the clustering agent nodes; selecting, from the clustering agent nodes, the relay node between the source host node and the target host node based on the distance prediction mechanism. With the transfer method and system for the peer-to-peer overlay network of the present invention, effective relay node can be found rapidly in the peer-to-peer overlay network, so that a substitute route with shorter time delay, higher efficiency and stronger stability is established, and high performance communication of peer-to-peer is realized.

Description

End-to-end overlay network transfer method and system

Embodiments of the present invention relate to network access techniques, and in particular to end-to-end overlay network relay methods and systems. Background technique

The current Internet network layer has poor overall routing efficiency due to factors such as routing protocols and network architectures. The default route selected by the IP (internet protocol: Internet Protocol) layer for best delivery principles has been difficult to meet many real-time requirements. High performance requirements for application layer services. The main reasons include the following:

In IP layer routing, when a packet is sent across different domains, the border gateway protocol (BGP: border gateway protocol) only guarantees connectivity, and it does not provide the shortest path. In this case, the transmission delay in the case of two direct connections is often too long to meet the requirements of the service application.

The queuing mechanism of routers forwarding packets in the Internet often causes congestion when the load is large. The network jitter caused by such congestion causes delays and frequent packet loss, which leads to performance degradation of the better quality links.

Because the network provides NAT (network address translation) and firewall settings, it also creates obstacles for network applications while protecting the network. This makes it impossible to directly connect a large number of destination nodes, NAT or Users behind the firewall are completely unable to enjoy application layer services such as Internet telephony.

In order to solve these problems, people began to study the P2P (peer-to-peer: end-to-end) overlay network transfer technology. Because from the current research, there are a lot of TIV (triangle inequality violations) in the Internet network. As shown in Figure 1, about 40%~50% of the paths have alternative paths with smaller or shorter delays. The % path has an alternate path with a lower packet loss rate. Fig. 1 is a schematic diagram showing a TIV phenomenon in an Internet network. The P2P overlay network transit technology is to find a suitable relay node in the P2P overlay network, and the media is forwarded through the relay node, thereby establishing a shorter delay, higher efficiency, and more stability between the two points of the application layer than the default IP route. Strong alternative routing for end-to-end high performance communication.

The P2P overlay network here refers to an application layer node (such as a skype soft terminal) self-organizing into a network according to the P2P mode, which is a logical virtual topology network. Relay node is a special Capable application layer node, which not only provides application services for local users, but also acts as a bridge node between two other application layer nodes to complete media forwarding. The core problem of P2P overlay network transit technology is the discovery algorithm of the relay node, that is, how to find available effective relay nodes with acceptable overhead in a short time. Summary of the invention

The present invention has been made in view of the above objects, and an object of the present invention is to provide an end-to-end overlay network relay method and system for efficiently discovering a relay node.

In one aspect, an end-to-end overlay network relay method is provided, where the end-to-end overlay network includes multiple host nodes and a boot server, and the method is used to find a relay node from a source host node to a destination host node. The method includes: the host node communicating with a boot server to obtain coordinate information, wherein the coordinate information is used to calculate a delay between two host nodes based on a distance prediction mechanism; performing clustering of the host node to a part of the host node acts as a clustering proxy node, and other host nodes act as host nodes belonging to the clustering proxy node; and selects a relay from the source host node to the destination host node from the clustering proxy node based on the distance prediction mechanism node.

In another aspect, an end-to-end overlay network transit system is provided. The end-to-end overlay network includes a plurality of host nodes and a boot server, and the system is configured to search for a relay node from a source host node to a destination host node. The end-to-end overlay network transit system includes: a communication unit, configured to enable the host node to communicate with a boot server to obtain coordinate information, where the coordinate information is used to calculate a time between two host nodes based on a distance prediction mechanism a clustering unit, configured to perform clustering of the host node to use a part of host nodes as clustering proxy nodes, and other host nodes as host nodes belonging to the clustering proxy node; and selecting units, The relay node from the source host node to the destination host node is selected from the clustering proxy nodes based on the distance prediction mechanism.

Through the end-to-end overlay network relay method and system according to the embodiment of the present invention, an effective relay node can be quickly found in the end-to-end overlay network, thereby establishing a shorter delay, higher efficiency, and more stable replacement. Routing, enabling end-to-end high-performance communication. DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings to be used in the embodiments or the description of the prior art will be briefly described below. Obviously, the drawings in the following description are only the present invention. Some embodiments, for those of ordinary skill in the art, do not pay for creation Under the premise of sexual labor, other drawings can also be obtained from these drawings.

Figure 1 is a schematic diagram showing the TIV phenomenon in an Internet network;

2 is a schematic flow chart showing a method of relaying an end-to-end overlay network according to a first embodiment of the present invention;

3 is a flow chart showing a specific example of a node adding process in the end-to-end overlay network relay method according to the first embodiment of the present invention;

4 is a flow chart showing a specific example of an information update process between a host node and a clustering proxy node in the end-to-end overlay network relay method according to the first embodiment of the present invention;

5 is a flowchart showing a specific example of an information update process between a cluster agent node and a BS server in the end-to-end overlay network relay method according to the first embodiment of the present invention;

6 is a flowchart showing a specific example of a Meridian ring construction process in the end-to-end overlay network relay method according to the first embodiment of the present invention;

7 is a flow chart showing a specific example of a Meridian ring maintenance update process in the end-to-end overlay network relay method according to the first embodiment of the present invention;

8 is a flowchart showing a specific example of a relay node selection process in the end-to-end overlay network relay method according to the first embodiment of the present invention;

9 is a schematic flow chart showing a method of relaying an end-to-end overlay network according to a second embodiment of the present invention;

FIG. 10 is a schematic flow chart showing a method for relaying an end-to-end overlay network according to a third embodiment of the present invention; FIG.

11 is a schematic flow chart showing a method for relaying an end-to-end overlay network according to a fourth embodiment of the present invention;

FIG. 12 is a schematic flow chart showing a method for relaying an end-to-end overlay network according to a fifth embodiment of the present invention; FIG.

Figure 13 is a schematic block diagram showing an end-to-end overlay network relay system according to a first embodiment of the present invention;

Figure 14 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a second embodiment of the present invention;

Figure 15 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a third embodiment of the present invention; 16 is a schematic block diagram showing an end-to-end overlay network relay system according to a fourth embodiment of the present invention;

Figure 17 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a first embodiment of the present invention. detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without making creative labor are within the scope of the present invention.

In a network environment according to an embodiment of the present invention, a fixed one or more BS (bootrap server) servers are specifically deployed, which guides network node registration, and the auxiliary host node finds a home clustering proxy node, and manages the storage system. All clustering agent node information, and guiding the new joining clustering node to initially construct their own ring structure. The clustering proxy node is upgraded by an ordinary client terminal that has the public network IP, strong CPU processing capability, large bandwidth, and long online time. The clustering agent node manages the host node information attributed to itself, and maintains the ring structure constructed by other neighboring clustering agent nodes. When the managed host node needs a relay node, the clustering proxy node selects and provides a suitable relay node list from its own ring structure. The clustering proxy node needs to act as a relay node, responsible for NAT traversal of the user plane and media relay. Moreover, the host node refers to a common client terminal, which is attached to a clustering proxy node after joining the network, and relies on the clustering proxy node to complete the searching of the relay node.

The ASAP (AS-aware peer-relay protocol) algorithm is an example of the P2P overlay network transit technology. It needs to deploy a boot server, which is responsible for collecting the distribution of the access network AS (Autonomous System). Map, AS and IP prefix mapping tables. User nodes are clustered based on IP prefix information, and each cluster selects a node with stronger capabilities as a clustering proxy node. The boot server needs to store maintenance clustering and clustering proxy node information. Each clustering proxy node obtains the neighboring AS information through the breadth-first search based on the AS topology map, and then constructs the neighboring cluster set by actual sounding. In the actual relay node selection process, the algorithm selects a suitable relay node based on the intersection of the primary and the called neighboring cluster sets.

In the above ASAP algorithm, the AS inter-domain relationship needs to be inferred by acquiring a BGP broadcast message.

The content of BGP broadcasts is limited and difficult to obtain. In addition, the algorithm is based on the detection side of the AS topology map. The method is too large and the actual cost is relatively high.

The Meridian loop algorithm is another example of the P2P overlay network relay technology. In the Meridian algorithm, when each node joins the network, it first obtains some of the existing network node information through the boot server or the existing node. Then, the actual detection of these nodes is performed, and a circular topology structure is constructed based on the obtained RTT (round-trip time) information, and the nodes are placed at different distances from the inside and the outside by themselves. Delay on the ring. In addition, it periodically maintains and updates this ring structure based on the gossip protocol. When the actual relay node is selected, the initiator selects a node from its own ring structure for broadcasting according to the set threshold, and finds a suitable relay node by hop-by-hop iteration.

However, the discovery of the relay node by the gradual approaching multi-hop method makes the setting of the approach parameters difficult, and the multi-hop mode is likely to cause media delay and packet loss, and multiple forwarding nodes cause downlink bandwidth consumption on the access side. . In addition, node clustering is not mentioned in the meridian algorithm to reduce network overhead and improve hit rate.

In order to solve the problems of Internet network failure, NAT traversal, and communication delay too much, the present invention proposes a novel and improved end-to-end overlay network relay method and system.

The basic idea of the embodiment of the present invention is an end-to-end overlay network relay method and system using one or more technologies of node clustering, meridian ring and GNP (Global Network Positioning).

In an embodiment of the invention, nodes in the network system may perform clustering based on delay and partial network topology information (e.g., IP address or ASN (Autonomous System Number)). Within each cluster, a node with strong capabilities is selected to act as a clustering proxy node to manage other nodes within the cluster. In addition, the meridian ring can be constructed only by the clustering agent node, thereby ensuring the geographical dispersion of the nodes on the meridian ring by node clustering.

Moreover, through the multi-layered parsing loop structure of the meridian ring, the clustering proxy nodes in the network system can more understand and carry the node information of the entire network topology on the basis of storing a limited number of nodes.

In addition, when performing the selection of the relay node, the distance prediction mechanism such as the GNP distance prediction mechanism is used to filter and obtain the one-hop candidate relay node from the meridian ring structure of the clustering proxy node to which the source and destination nodes belong, which can greatly Reduce unnecessary detection overhead and improve prediction accuracy.

Next, a first embodiment of the end-to-end overlay network relaying method of the present invention will be described based on FIGS. 2 through 8. 2 is a schematic flow chart showing a method for relaying an end-to-end overlay network according to a first embodiment of the present invention. As shown in FIG. 2, an end-to-end overlay network relay method according to a first embodiment of the present invention is applied to an end-to-end overlay network including a plurality of host nodes and a boot server for searching from a source host node to a destination host. a relay node of the node, the method includes: S1000, the host node communicates with a boot server to obtain coordinate information, where the coordinate information is used to calculate a delay between two host nodes based on a distance prediction mechanism; S1001, execute Clustering of the host node to use a part of the host node as a clustering proxy node, and other host nodes as host nodes belonging to the clustering proxy node; S1002, based on a distance prediction mechanism from the clustering proxy node Select the relay node from the source host node to the destination host node.

Next, the above-described end-to-end overlay network relaying method according to the first embodiment of the present invention will be described in detail based on Figs. 3 to 8 as follows.

In an embodiment of the present invention, the node joining step may include registration, GNP coordinate calculation, and node clustering. In the network system according to the embodiment of the present invention, BS server address information is stored on each node, and when the node joins the network, the node to be joined first initiates a registration request to the BS server. Then, the node to be joined obtains the landmark information after the BS server successfully completes the authentication, and then calculates its own GNP coordinates through the GNP algorithm. After the GNP coordinate calculation is completed, the node performs a clustering operation and initiates a home clustering proxy node information request to the BS server. The BS server assists the newly joined nodes to complete clustering based on topology information such as delay and IP address. 3 is a flow chart showing a specific example of a node joining process in the end-to-end overlay network relay method according to the first embodiment of the present invention.

As shown in FIG. 3, S1100, the host node to join the network initiates a registration request message to the BS server, for example, the message may carry the user name and password of the user; S1101, the BS server authenticates the host node to join the network, For example, authentication is performed by identifying the validity of the user's username and password. If the authentication succeeds, the BS server returns a success response message to the host node to join the network, and carries a plurality of landmark node information, where the landmark node information may include the IP address and GNP coordinates of the landmark node. Etc. (The landmark node is a device deployed in the network in advance, and its information is stored on the BS server.) If the authentication fails, the BS server returns a failure response message; S1102, the host node to join the network obtains the landmark from the success response message. The IP address of the node, and then sends a probe message to the obtained landmark node based on the IP address and receives a response, and calculates a corresponding RTT (the RTT calculation method calculates the time of receiving the response message of the probe message and sends the corresponding probe message. Time between times The difference is obtained, and the subsequent calculation of RTT is similar to the method here. Combining the obtained RTT and the GNP coordinates of multiple landmark nodes, for example, the GNP coordinates of the host node joining the network are calculated by the Simplex Downhill method; S1103, to join the network. The host node sends a probe message to the BS server to obtain the RTT of the BS server, and completes the NAT type detection of the local node (here, a person skilled in the art can understand the specific manner of NAT type detection, for example, refer to RFC3489); S1104, to join The host node of the network sends a clustering proxy node information request to the BS server, and the parameter is the RTT acquired in the previous step; S1105, the BS server selects multiple aggregations from the stored clustering proxy node database according to the topology information such as the RTT value and the IP address. The class proxy node is used as a candidate. For example, the criterion for the BS server to select the candidate node may be that the delay of the clustering proxy node to the BS server is similar to the delay of the host node joining the network to the BS server (ie, the difference between the two meets a certain Range, specific value can be selected according to network characteristics), in addition, selection criteria The IP address of the clustering proxy node may be as long as the same IP prefix of the IP address of the joining node, and so on; S1106, the BS server returns a success response message, and carries the candidate clustering proxy node selected in S1105 in the message. Information, for example, the information may include an IP address of the candidate clustering proxy node, or other identification information of the candidate clustering proxy node, etc.; S1107, the joined host node receives the response message, if the candidate clustering proxy node information in the response message If yes, go to S1117; S1108, if the candidate clustering proxy node information in the response message is not empty, the joined host node sends a probe message to the candidate clustering proxy nodes; S1109, the candidate clustering proxy node receives the probe message. Immediately after returning the response message; S1110, the joined host node calculates the delay RTT of the plurality of candidate clustering proxy nodes; S1111, if there is no delay RTT is less than the set intra-cluster delay maximum value (according to the network feature value In the existing case, the delay RTT between nodes in an AS on the Internet is 60 ms on average; for example, the time within the cluster can be If the maximum value is set to 60ms, then go to S1117; S1112, if the delay RTT is less than the set maximum delay within the cluster, the clustering proxy node corresponding to the minimum delay is taken as the home clustering proxy node; S1113, sending a join request message to the home clustering proxy node selected in S1112; S1114, the home clustering proxy node determines whether the join is successful according to the number of the cluster nodes, for example, if the number of storage nodes of the cluster proxy node is full Then, if it is not full, it succeeds, and returns a response message; S1115, if a success response message is received, the joined host node stores information of the home clustering proxy node, such as IP address information, and simultaneously starts and clusters between the proxy nodes. The periodic information update task, the joining process is completed; S1116, if a failure response is received, the remaining candidate clustering proxy nodes are repeated S1111 and S1112, and if the repetition reaches a certain number of times, then go to S1117; S1117 , joined the host The node applies to the BS server to generate a new cluster, and acts as a proxy node for the cluster. At this time, the BS server generates a new cluster ID for the node, and simultaneously sets the IP address of the node, GNP coordinates, to the BS server. The delay RTT, the node network status (bandwidth and CPU usage), the update time, etc. are stored in the database corresponding to the cluster ID, and the joining system is completed.

After completing the process of node joining, the node may need to perform self-organizing management for periodic information update (the information includes, for example, delay and update time), and the end-to-end overlay network transit method in the first embodiment of the present invention The self-organizing management process of the node includes an information update process between the host node and the clustering proxy node and an information update process between the clustering proxy node and the BS server.

4 is a flow chart showing a specific example of an information update process between a host node and a clustering proxy node in the end-to-end overlay network relay method according to the first embodiment of the present invention. In the above information update process, the information is periodically refreshed between the common host node and the clustering agent to ensure the accuracy of the stored data. As shown in FIG. 4, the information update process includes: S1200, the host node periodically sends an information update message to the attached clustering proxy node, and sets a timeout timer; S1201, after the clustering proxy node receives the information update message, Updating the stored host node information, such as the delay RTT and update time of the host node to the BS server; S1202, the clustering proxy node returns a success response message, wherein if the timer expires, the host node still does not receive the cluster proxy node return The success response message, the host node sends a request message to the BS server to obtain new home clustering proxy node information; S1203, the clustering proxy node periodically detects the stored host node database, and the deletion does not periodically send the update message. Node.

Fig. 5 is a flowchart showing a specific example of an information update process between a clustering proxy node and a BS server in the end-to-end overlay network relay method according to the first embodiment of the present invention. During the above information update process, the information is periodically refreshed between the clustering proxy node and the BS server to ensure the accuracy of the stored data. The information includes, for example, the node network status (bandwidth and CPU usage) and the update time. As shown in FIG. 5, the information update process includes: S1300: The clustering proxy node periodically sends an information update request message to the BS server; S1301: The BS server updates the stored clustering proxy node information after receiving the information update request message, For example, clustering the node network status and update time of the proxy node, etc.; S1302, the BS server returns a success response message; S1303, the BS server periodically detects the stored cluster proxy node database, and deletes the information update request message without periodically sending Clustering proxy node.

In the end-to-end overlay network relay method according to the first embodiment of the present invention, self-organization of nodes is performed. Management is performed to periodically update information between the host node and the clustering proxy node, and between the clustering proxy node and the BS server. However, in other embodiments of the present invention, for example, in an environment where the change in the network system is relatively small, the self-organizing management process of the above-mentioned nodes may be omitted, thereby reducing network overhead. Moreover, instead of periodically updating the information, other embodiments of the present invention may also perform information update based on other conditions, for example, updating information in response to a predetermined operation or predetermined instruction of the user or the network administrator, thereby improving network usage. Convenience. Here, the embodiments of the present invention are not intended to impose any limitation.

6 is a flow chart showing a specific example of a Meridian ring construction process in the end-to-end overlay network relay method according to the first embodiment of the present invention. In the first embodiment of the present invention, the construction of the meridian ring is performed by the clustering proxy node, thereby ensuring the geographical dispersion of the nodes on the meridian ring. Moreover, when the clustering proxy node joins the network for the first time, it is necessary to first acquire a batch of neighboring clustering proxy node information through the BS server to construct a meridian ring structure. In order to alleviate the load of the network and the pressure of the BS server, in the first embodiment of the present invention, the BS server returns not the global clustering proxy node information. As shown in FIG. 6, the meridian ring construction process includes: S1400, the clustering proxy node sends a neighboring clustering proxy node request message to the BS server; S1401, the BS server sends the neighboring clustering proxy based on the clustering proxy node to construct the Meridian ring. The node request message calculates the delay of the other clustering proxy nodes stored on the BS server to the clustering proxy node that is to construct the Meridian ring according to the GNP coordinate prediction (here, the parameter used in the calculation is other stored on the BS server) The GNP coordinates of the clustering proxy node and the GNP coordinates of the clustering proxy node, and the calculation method may be the Euclidean distance: d=sqrt(∑(xil-xi ² ) ^A ) where i=l, ² ..n ), Selecting a certain number of clustering proxy nodes as candidates; S1402, the BS server returns a response message, where the response message carries the clustering proxy node information selected in S1401, for example, the information may include the IP address of the selected clustering proxy node, Or other identification information, etc.; S1403, the clustering proxy node obtains the information of the neighboring clustering proxy node from the response message returned by the previous BS server For example, an IP address, and sending a probe message to the acquired nodes; S1404, the proximity clustering node returns a response message; S1405, the clustering proxy node calculates a delay RTT from the sending of the probe message to the reception of the response message, and then according to the RTT These neighboring clustering proxy node information is stored into the corresponding ring structure to construct its own original meridian ring. Here, the specific construction method of the Meridian ring can be as follows. First, the clustering proxy node generates a plurality of concentric rings of different radii with its own center. The inner radius of these rings is ri=as ^A (il), and the outer radius Ri=as ^A (i)(i>0), a is A constant, s is a multiplication factor. The innermost ring r0=0, R0=a. Where the number of specific rings, The number of storage nodes on the ring and the delay settings between the rings need to be analyzed by actual test data to obtain better values. Then, the clustering proxy node finds the corresponding ring interval according to the delay of the other clustering agent nodes, and the other clustering agent node information (IP address, GNP coordinates, delay to the clustering proxy node, update time) Etc.) Stored in the corresponding ring database. Since the ring radius of the meridian ring's ring structure grows exponentially, each node can learn more about and carry the node information of the entire network topology on the basis of storing a limited number of nodes.

In the above-described first embodiment of the present invention, in order to reduce the load of the network and the pressure of the BS server, a certain number of clustering proxy nodes are selected as candidates based on the delay between the clustering proxy nodes obtained by the GNP coordinate prediction. However, those skilled in the art can understand that the BS server can also arbitrarily select a certain number of clustering proxy nodes as candidates, or perform selection of candidate clustering proxy nodes based on other mechanisms, as described in the second embodiment of the present invention. The embodiments of the present invention are not intended to impose any limitation on this.

After completing the construction of the meridian ring, the meridian ring may also need to be maintained and updated. That is to say, after each cluster agent node completes the construction of its own meridian ring, it may need to be dynamically maintained in real time to ensure the real-time performance of the stored node delay information. In addition, since the BS server does not return all the clustering proxy node information in the network during the login process, the ring update maintenance needs to acquire more neighboring clustering proxy node information in the current network in time to ensure that when searching for the transit node , can cover a larger range and increase the probability of finding a transit path. 7 is a flow chart showing a specific example of a Meridian ring maintenance update process in the end-to-end overlay network relay method according to the first embodiment of the present invention. As shown in FIG. 7, the maintenance update process adopts a lightweight gossip mechanism, including: S1500, the clustering proxy node randomly selects some nodes as seed nodes from each ring of the meridian ring; S1501, the seed selected in S1500 The node sends a neighboring node request message; S1502, each seed node randomly selects some nodes in its own ring structure; S1503, the seed node returns a response message, where the response message carries node information of the node selected in S1502, for example, the selected node IP address, GNP coordinates, delay to the local node, update time, etc.; S1504, the clustering proxy node processes the node carried in the response message based on the acquired node information, deletes the duplicate item and repeats with the node on the meridian ring itself Nodes, these nodes are used as backup ring nodes; S1505, the clustering proxy node sends a probe request message to the backup ring node acquired at S1504; S1506, the backup ring node returns a response message; S1507, calculates a send probe request message to receive a response message The delay RTT, then find its own meridian ring structure based on the delay RTT On the corresponding ring interval, then the node information, such as IP address, GNP The coordinates, the delay to the node, the update time, etc. are stored in the database of the corresponding ring. In the first embodiment of the present invention, since the number of nodes stored on the ring is limited, it is necessary to formulate a replacement mechanism based on comprehensive information such as delay, node IP address, and the like, and the number of times the node is selected as the relay node.

In the above embodiment, the clustering proxy node may periodically send a probe request to the node on the ring structure to determine whether the node is online, and adjust the position of the node in the ring structure according to the returned RTT.

As in the case of the self-organizing management of the above-described nodes, in other embodiments of the present invention, for example, in a relatively small change in the network system, the maintenance update process of the meridian ring described above can be omitted, thereby reducing network overhead. Moreover, the maintenance update process of the Meridian ring may be performed periodically, or may be performed based on other conditions, for example, performing maintenance and update of the meridian ring in response to a predetermined operation or a predetermined instruction of the user or the network administrator, thereby improving the network. The convenience of use. Here, the embodiments of the present invention are not intended to impose any limitation.

FIG. 8 is a flowchart showing a specific example of a relay node selection process in the end-to-end overlay network relay method according to the first embodiment of the present invention. In the first embodiment of the present invention, the selection of the relay node is based on the meridian ring information of the clustering proxy node to which the source node and the destination node belong, and combines the GNP distance prediction mechanism, thereby reducing unnecessary detection overhead. As shown in FIG. 8, the relay node selection process includes: S1600, the source node searches for the IP address of the destination node through the clustering proxy node that belongs to itself, and the source node and the destination node perform message interaction, and the two obtain the GNP of the other party respectively. Coordinates, the IP address of the clustering proxy node to which the attribute belongs, and the NAT type of the network. S1601: The source node and the destination node calculate the delay RTT and the packet loss rate of the default path by using multiple probe packet interactions. S1602: The source node sends a relay node selection request message to the clustering proxy node that belongs to the home node, where the message carries the GNP coordinate information of the source node and the destination node respectively; S1603, the source node sends the cluster node to the destination node to which the destination node belongs. Following the node selection request message, the message carries the GNP coordinate information of the source node and the destination node respectively; S1604, the clustering proxy node to which the source node belongs calculates the stored data by GNP distance prediction according to the destination node coordinates provided in the above message. The delay of other clustering proxy nodes in the ring structure to the destination node (the calculation parameter is the GNP coordinates of other clustering proxy nodes in the ring structure stored by itself and the GNP coordinates of the destination node, and the calculation method is the Euclidean distance: d=sqrt (∑(xil-xi2) ^A ) where i=l,2..n) is added to the actual probe delay of the other clustering proxy nodes stored in the ring structure to the local node to obtain the total delay of each relay path. Then, deleting the node whose total delay is greater than the default path delay, obtaining a candidate node set, and then selecting a certain number of nodes as candidates Following the node; S1605, the clustering proxy node to which the source node belongs returns a response message, where the message carries the candidate relay node information selected in S1604; S1606, the clustering proxy node to which the destination node belongs is based on the source provided in the relay node selection request message Node coordinates, the GNP distance prediction is used to calculate the delay of other clustering proxy nodes in the ring structure stored by themselves to the source node (the calculation parameters are the GNP coordinates and source nodes of other clustering proxy nodes in the ring structure stored by itself) The delay value of the GNP coordinate is calculated as the Euclidean distance: d=sqrt( X(xil-xi2) ^A ) where i=l,2..n, and the calculated result is also the delay value) and in the ring structure The stored other clustering agent nodes are added to the actual probe delay of the node to obtain the total delay of each relay path, and then the nodes whose total delay is greater than the default path delay are deleted, and a candidate node set is obtained, and then Selecting a certain number of nodes as candidate relay nodes; S1607, the clustering proxy node to which the destination node belongs returns a response message, and the message carries the candidate relay selected in S1606 Node information; S1608, the source node obtains the candidate relay node information from the response message, deletes the duplicate item, obtains a candidate node set, and then initiates delay and packet loss rate detection for the relay path corresponding to the relay node (specific detection The method is as follows: The source node sends a probe message to the relay node, and the relay node forwards the message to the destination node, and then returns the response in reverse, and the delay and the packet loss rate of the relay path can be calculated through multiple message interactions. Obtaining the delay and packet loss rate of each relay path; S1609, the source node selects three nodes with better default path quality from the candidate relay nodes of S 1608 as one master and two standbys of the current session. The relay node, the judgment criterion is the delay and the packet loss rate, and which condition can be selected according to the network condition and the service requirement. If the currently obtained relay node cannot find the appropriate three nodes, then S1602 is repeated; S1610, if the number of repetitions of S1602 reaches the set threshold, and one of the three primary and secondary relay nodes that meet the demand is still not found, Then, the node with the best path performance is selected from the previously detected nodes to complete the primary and secondary backup mechanisms; S1611, the primary and secondary backup three relay node information is notified to the destination node, and then the three relay paths are cycled. Sexually detect, obtain delay and packet loss rate, and complete the switch between paths according to network quality changes and set switching parameters.

In addition, in the foregoing relay node selection process, when selecting a candidate relay node by GNP distance prediction, another node may be selected as a candidate relay node according to a random principle to avoid a large-area failure in the network. The selected candidate relay nodes are all invalid. This is because when candidate relay nodes are selected using GNP distance prediction or other distance prediction mechanisms, usually these candidate relay nodes will be geographically relatively close, so when large-area failures occur in the network, it is likely that these candidates The relay nodes will all fail, making it impossible to select a relay node. By randomly selecting some candidate relay nodes, it is guaranteed that the relay node can be selected. Thereby, the adaptability of the end-to-end overlay network relay method according to the embodiment of the present invention is improved.

The end-to-end overlay network transit method according to the first embodiment of the present invention can solve the problems of poor routing efficiency, frequent congestion, routing failure, NAT traversal, etc. in the Internet network, and provide the problem with a small detection overhead. The hit rate of the higher transit node. By storing neighboring node information in the form of a meridian ring structure, each node can learn more about and carry the node information of the entire network topology on the basis of storing a limited number of nodes. Moreover, the nodes are clustered by combining the RTT value and the IP address factor, and only the clustering proxy nodes participate in the construction of the meridian ring, which can ensure the geographical dispersion of the nodes on the meridian ring. In addition, when selecting a relay node, the GNP distance prediction mechanism selects and obtains a one-hop candidate relay node from the meridian ring structure of the calling party, which can greatly reduce unnecessary detection overhead and improve prediction accuracy.

In the first embodiment of the present invention described above, the clustering of nodes is combined with the RTT value and the IP address factor as an example. However, in the end-to-end network, there are other node clustering methods, and these nodes are also present. The clustering method can effectively support end-to-end network routing, application layer multicasting, resource efficient placement, and perceptual topology construction, thereby greatly improving the performance of the end-to-end network. For example, the node clustering mechanism can also include the following scenarios:

(1) By clustering according to the BGP routing table prefix, each node joins the network, and the clustering is completed by IP address prefix matching.

(2) Based on router clustering: Using the router in the transport network (Router) is more stable than the end node located at the edge of the network, the matching relationship between the overlay network and the physical network of the P2P is used between the end node and the router. Match the relationship to represent. When each node joins the network, it first sends a traceroute request to the pre-selected landmark node of the network. By analyzing the delay information recorded by the traceroute, it can find the routers located between the areas, and then implement the aggregation according to these routers. class.

Moreover, in the first embodiment of the present invention described above, the method for screening and obtaining a one-hop candidate relay node from the meridian ring structure of the called party is described by using a GNP distance prediction mechanism, but may also be based on other distance predictions. The scheme is to select a relay node from the Meridian ring structure. In fact, the main task of network distance prediction is to estimate the delay value between any two nodes in the actual network based on the limited measurement information, and then provide relevant information to the user. For example, the distance prediction mechanism can also include the following scenarios:

IDMaps: IDMaps is the earliest proposed large-scale network distance prediction mechanism. The medium service provides distance prediction services for network nodes. The mechanism includes two core contents: an address prefix block (the address prefix, called the AP) and a Tracer node. Each AP represents a set of nodes with the same IP address prefix, and the Tracer node is a measurement node deployed in the network, and ensures that each address block has at least one Tracer close to it. Different Tracer nodes periodically measure each other's distances, while each Tracer node periodically measures the distance to its neighboring APs. All of this measurement information is collected into a centralized server. Thus, the distance between any two hosts is equal to the sum of the distance between the APs from the two hosts to their respective Tracer nodes and the distance between the two associated Tracer nodes.

Lighthouse: This method uses coordinate transformation to calculate network coordinates. Firstly, a global coordinate base is set in the system. The newly added node first obtains the corresponding distance information by randomly detecting the K+1 (K is the coordinate dimension) landmark, and calculates its global coordinate base through the Gram-Schmidt process, and then passes the local sum. The transformation matrix of the global coordinate base to calculate the global coordinates of the new node. The distance between any two nodes in the system is calculated by global coordinates.

Virtual landmarks: Use Principal Component Analysis (PCA) to extract the topology information of the network to reduce the dimension of the embedded space, and obtain the coordinates of the node in the embedded space by matrix multiplication. The algorithm needs to first deploy N reference nodes to form a set of reference nodes, and the reference nodes measure each other to reach each other to form an N*N distance matrix. Then use PCA technology to eliminate similar vectors, reduce dimensions, and calculate transformation matrices. When a node joins the system, the distances of all the reference nodes are first measured to form a distance vector, and then the vector is mapped to the new space according to the transformation matrix, and the mapped vector value is the new coordinate.

Vivaldi: A heuristic coordinate adjustment process that simulates the spatial embedding problem as a spring elastic potential minimization problem. Each node in the system has a network coordinate X and a local error e. When the node joins the system, it randomly measures the distance to a group of nodes in the system, determines its own initial coordinates based on the measured values, and then makes the network coordinate X and the local error e continuously close to the ideal value by multiple updates. In each update, the node first measures the actual delay between it and a neighbor node, and then adjusts the current network coordinate X and local error e by the measured delay.

9 is a schematic flow chart showing a method for relaying an end-to-end overlay network according to a second embodiment of the present invention. As shown in FIG. 9, the end-to-end overlay network transit method according to the second embodiment of the present invention includes the following steps: S2000, adding a node to a network and performing node clustering to select a part of nodes as a clustering proxy node, and other nodes As a host node belonging to the clustering proxy node; S2001, for each clustering proxy node, constructing its own Meridian ring node based on other clustering proxy nodes And S2002, selecting a relay node from the source node to the destination node from the nodes on the Meridian ring structure. That is, in the end-to-end overlay network relay method according to the second embodiment of the present invention, the node clustering and the feature of constructing the meridian ring are combined.

In the end-to-end overlay network relay method according to the second embodiment of the present invention shown in FIG. 9, the steps of node clustering and the step of constructing a Meridian ring structure by clustering proxy nodes may be the same as the above according to the present invention. The corresponding steps in the end-to-end overlay network relay method of the embodiment are the same. Moreover, similar to the above-described end-to-end overlay network transfer method according to the first embodiment of the present invention, in the end-to-end overlay network transfer method according to the second embodiment of the present invention, node self-organizing management and Meridian ring maintenance can also be performed. The update is the same as the above, and will not be described here for the sake of clarity. In addition, those skilled in the art can understand that the above-mentioned end-to-end overlay network transfer method according to the embodiment of the first embodiment of the present invention is not intended to be arbitrarily limited.

As described above, by clustering when nodes join the network, and only participating in the construction of the Meridian ring by the clustering proxy node, the geographical dispersion of the nodes on the Meridian ring can be ensured, thereby avoiding the candidate nodes in the case of large-area failures in the network. All failures, and improve the selection efficiency of candidate nodes. Moreover, by storing neighboring node information in the form of a meridian ring structure, each node can learn more about and carry the node information of the entire network topology on the basis of storing a limited number of nodes.

Compared with the above-described first embodiment of the present invention, in the end-to-end overlay network relay method according to the second embodiment of the present invention, a specific method of selecting a relay node from a node on a Meridian ring structure is not limited, which makes The appropriate relay nodes can be selected based on the existing Meridian algorithm, thereby reducing system changes and facilitating compatibility with existing systems.

FIG. 10 is a schematic flow chart showing a method for relaying an end-to-end overlay network according to a third embodiment of the present invention. As shown in FIG. 10, the end-to-end overlay network transit method according to the third embodiment of the present invention includes the following steps: S3000, adding a node to a network and performing node clustering to select a part of nodes as a clustering proxy node, and other nodes As a host node belonging to the clustering proxy node, wherein the node calculates information required by the distance prediction mechanism when joining the network; S3001, selecting a slave source node from the clustering proxy node based on a distance prediction mechanism The relay node to the destination node. That is, in the end-to-end overlay network relay method according to the third embodiment of the present invention, the features of node clustering and distance prediction are combined. In the end-to-end overlay network relay method according to the third embodiment of the present invention shown in FIG. 10, the steps of node clustering and the step of selecting a relay node based on the distance prediction mechanism may be the same as the above-described first embodiment according to the present invention. The corresponding steps in the end-to-end overlay network relay method are the same. Moreover, similar to the above-described end-to-end overlay network transfer method according to the first embodiment of the present invention, in the end-to-end overlay network transfer method according to the third embodiment of the present invention, node self-organizing management may also be performed, and The process is the same as above, and will not be repeated here for reasons of clarity. In addition, those skilled in the art can understand that other aspects of the above-described end-to-end overlay network transfer method according to the first embodiment of the present invention can also be equally applied to the end-to-end overlay network transfer method according to the third embodiment of the present invention. The embodiments of the present invention are not intended to be any limitation.

As described above, since the clustering proxy node to which the source and destination nodes belong stores information of other clustering proxy nodes, the neighboring clustering proxy node stored by the clustering proxy node to which the source and destination nodes belong is adopted by using the distance prediction mechanism. Filtering data to obtain a candidate relay node can greatly reduce unnecessary detection overhead and improve prediction accuracy.

Moreover, in the end-to-end overlay network relay method according to the third embodiment of the present invention, the Meridian ring structure is not limited to be constructed by the clustering proxy node, and therefore, according to the first embodiment of the present invention, The end-to-end overlay network relay method of the third embodiment of the present invention can also be applied to other algorithms than the Meridian algorithm, for example, the ASAP algorithm, thereby improving the adaptability of the end-to-end overlay network relay method according to the embodiment of the present invention.

In the above-described first to third embodiments according to the present invention, each includes a process of clustering to join a network to generate a clustering proxy node, and then selecting a relay node from the clustering proxy node. Here, those skilled in the art can understand that when a node joins an end-to-end overlay network, for example, as in the existing Meridian algorithm, node clustering may not be performed, but an appropriate relay node may be directly selected.

Figure 11 is a schematic flow chart showing a method of relaying an end-to-end overlay network according to a fourth embodiment of the present invention. As shown in FIG. 11, the method for relaying the end-to-end overlay network according to the fourth embodiment of the present invention includes the following steps: S4000, a node joins a network to construct a Meridian ring structure, where the node calculates a distance prediction mechanism when joining the network. Required information; S4001, selecting a relay node from the source node to the destination node from the nodes of the Meridian ring structure based on a distance prediction mechanism. That is to say, in the end-to-end overlay network relay method of the fourth embodiment of the present invention, the feature of constructing the meridian loop and the distance prediction is combined.

Here, the end-to-end overlay network relay method according to the fourth embodiment of the present invention shown in FIG. The steps of selecting a relay node from the nodes of the Meiridan ring structure based on the distance prediction mechanism may be the same as the corresponding steps in the above-described end-to-end overlay network relay method according to the first embodiment of the present invention. Moreover, similar to the above-described end-to-end overlay network transfer method according to the first embodiment of the present invention, in the end-to-end overlay network transfer method according to the fourth embodiment of the present invention, the maintenance update of the Meridian ring can also be performed, and The specific process is the same as above, and will not be repeated here for reasons of clarity. In addition, those skilled in the art can understand that the above-mentioned end-to-end coverage peer-end overlay network relay method according to the first embodiment of the present invention is not intended to be any limitation.

As described above, by using the distance prediction mechanism to select a candidate relay node from the nodes on the Meridian ring structure, unnecessary detection overhead can be greatly reduced, thereby improving prediction accuracy.

In addition, by directly applying the distance prediction mechanism in the existing Meridian algorithm, the compatibility of the end-to-end overlay network relay method according to the embodiment of the present invention with the network using the existing Meridian algorithm can be improved.

As described above, in the embodiment of the present invention, the three techniques of node clustering, meridian loop, and distance prediction are used in combination with each other, thereby improving the accuracy of relay node selection. Among them, clustering is performed when the nodes join the network. Only the clustering proxy nodes participate in the selection of the relay nodes, which can ensure the geographical dispersion of the candidate relay nodes, thereby preventing all candidate nodes from being large-area faults in the network. In the case of failure, and improving the selection efficiency of the candidate nodes; and screening and obtaining the one-hop candidate relay node from the candidate relay nodes by the distance prediction mechanism, the unnecessary detection overhead can be greatly reduced, thereby improving the prediction accuracy.

Figure 12 is a schematic flow chart showing a method of relaying an end-to-end overlay network according to a fifth embodiment of the present invention. As shown in FIG. 12, the method for relaying an end-to-end overlay network according to a fifth embodiment of the present invention includes the following steps: S5000: A node joins a network, where the node calculates information required by a distance prediction mechanism when joining a network; S5001, A relay node from the source node to the destination node is selected based on the distance prediction mechanism. That is, in the end-to-end overlay network relay method according to the fifth embodiment of the present invention, only the characteristics of the distance prediction are used to select the relay node.

Here, the process of joining a node to a network and the process of selecting a relay node based on a distance prediction mechanism may be the same as the corresponding steps in the above-described end-to-end overlay network relay method according to the first embodiment of the present invention. Moreover, similar to the above-described end-to-end overlay network transfer method according to the first embodiment of the present invention, in the end-to-end overlay network transfer method according to the fifth embodiment of the present invention, node self-organization can also be performed. The management needs to update the information required by the distance prediction mechanism, and the specific process is the same as above, and will not be repeated here for the sake of clarity. In addition, those skilled in the art can understand that other aspects of the above-described end-to-end overlay network transfer method according to the first embodiment of the present invention can also be equally applied to the end-to-end overlay network transfer method according to the fifth embodiment of the present invention. The embodiments of the present invention are not intended to be any limitation.

By selecting a relay node by using a distance prediction mechanism in the end-to-end overlay network, the TIV phenomenon as in the prior art can be avoided, so that the direct path from the source node to the destination node can be replaced with an alternative path with a smaller delay. Establish alternative routes with shorter latency, higher efficiency and greater stability for end-to-end high-performance communication.

According to another aspect of an embodiment of the present invention, an end-to-end overlay network relay system is also provided. Next, an end-to-end overlay network relay system according to the first to fifth embodiments of the present invention will be described with reference to Figs. 13 to 17 .

Figure 13 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a first embodiment of the present invention. The end-to-end overlay network according to the first embodiment of the present invention is applied to an end-to-end overlay network including a plurality of host nodes and a server, and is used to find a relay node from a source host node to a destination host node. As shown in FIG. 13, the end-to-end overlay network relay system 100 according to the first embodiment of the present invention includes: a communication unit 101 for causing a host node to communicate with a boot server to obtain coordinate information for distance prediction based a mechanism to calculate a delay between two host nodes; a clustering unit 102, configured to perform clustering of the host node, to use a part of the host node as a clustering proxy node, and other host nodes as belonging to the clustering proxy node a host node; and a selecting unit 103, configured to select, from the clustering proxy node, a relay node from the source host node to the destination host node based on the distance prediction mechanism.

In the above-mentioned end-to-end overlay network relay system 100, it may further include: a construction unit (not shown) connected between the clustering unit 102 and the selection unit 103, for each clustering node, for other aggregation The class proxy node builds its own Meridian ring structure based on it; and the selecting unit 103 is specifically configured to select a relay node from the source host node to the destination host node from the clustering proxy nodes on the Meridian ring structure based on the distance prediction mechanism.

In the above-mentioned end-to-end overlay network transit system 100, the method further includes: a first update unit (not shown) for updating at least one of the following information between the host node and the clustering proxy node: a network of the host node Status, delay from the host node to the boot server, and update time of the host node; and a second update unit (not shown) for more between the cluster agent node and the BS server At least one of the following new information: the network status of the clustering agent node, the delay from the clustering agent node to the boot server, and the update time of the clustering agent node.

In the above-mentioned end-to-end overlay network transit system 100, the method further includes: a maintenance update unit (not shown) for causing each cluster proxy node to maintain and update its own Meridian ring structure, wherein each proxy node maintains And updating at least one of the following information: IP address of other clustering proxy nodes on the Meridian ring structure, coordinate information, delay to the clustering proxy node, and update time.

In the above-mentioned end-to-end overlay network transit system 100, the selecting unit 103 is specifically configured to: based on the distance prediction mechanism, the clustering proxy node on the Meridian ring structure of the clustering proxy node to which the source host node belongs and the cluster to which the destination host node belongs The clustering proxy node on the Meridian ring structure of the class proxy node selects the relay node from the source host node to the destination host node.

In the above-mentioned end-to-end overlay network transit system 100, the selecting unit 103 is specifically configured to: cause the clustering proxy node to which the source host node belongs to calculate each aggregate of its own Meridian ring structure by distance prediction based on the coordinates of the destination host node. The delay from the class proxy node to the destination host node is respectively added to the stored delay of each clustering proxy node on the Meridian ring structure to the clustering proxy node to which the source host node belongs, and each clustering proxy node is obtained. The total delay of the associated relay paths; selecting a clustering proxy node associated with the relay path whose total delay is less than a predetermined threshold as a candidate relay node; delaying the relay path associated with the candidate relay node And the packet loss rate detection, and selecting the candidate relay node related to the relay path with the best path quality as the relay node from the source host node to the destination host node based on the delay and packet loss rate detection results.

In the above-mentioned end-to-end overlay network transit system 100, the selecting unit 103 is further configured to: in addition to the clustering proxy node related to the relay path whose total delay is less than a predetermined threshold, additionally select some clustering proxy nodes as candidates Relay node.

In the above-mentioned end-to-end overlay network transit system 100, the clustering unit 102 is specifically configured to: cluster the host nodes based on the round-trip delay of the host node to the boot server and the IP address of the host node.

In the above-mentioned end-to-end overlay network transit system 100, the distance prediction mechanism is GNP distance prediction, and the coordinate information used to calculate the delay between two host nodes based on the distance prediction mechanism is GNP coordinates, and the end-to-end overlay network transit The system 100 further includes: a registration requesting unit (not shown) for causing the host node to transmit a registration request to the bootstrap server; an authentication unit (not shown), connected to the registration requesting unit and the communication unit 101, for causing the boot server And authenticating the host node, and sending a success response message to the host node in the case that the authentication succeeds, in the success response message The information includes a plurality of landmark nodes, wherein the landmark node is a device whose information is previously stored in the network and stored on the boot server, and the information of the landmark node includes at least an IP address or a GNP coordinate of the landmark node.

In the above-mentioned end-to-end overlay network transit system 100, the constructing unit is specifically configured to: enable the boot server to calculate the delay between each cluster proxy node and other cluster proxy nodes by using GNP distance prediction, and based on the delay Selecting a certain number of candidate clustering proxy nodes for each clustering proxy node; causing each clustering proxy node to store other clustering proxy node information to its own Meridian ring based on round-trip delays with other clustering proxy nodes In the structure, thus constructing its own Meridian ring structure.

In the above-mentioned end-to-end overlay network transit system 100, the selecting unit 103 is specifically configured to: enable the source host node to send a relay node selection request message to the clustering proxy node to which the clustering proxy node and the destination host node to which it belongs, The relay node selection request message includes GNP coordinates of the source host node and the destination host node; the cluster proxy node to which the source host node belongs is based on the GNP coordinates of the destination host node, and calculates its own Meridian ring structure by GNP distance prediction. The delay of each clustering agent node to the destination node, and adding the delay of each clustering proxy node to the source host node on its own Meridian ring structure to obtain the total delay of each relay path; selecting the total delay a clustering proxy node corresponding to a relay path less than or equal to a predetermined threshold as a first group of candidate relay nodes; causing a clustering proxy node to which the destination host node belongs to calculate its own GNP distance based on GNP coordinates of the source host node Delay of each clustering proxy node on the Meridian ring structure to the source host node And adding the delay of each clustering proxy node to the destination host node on the Meridian ring structure to obtain the total delay of each relay path; selecting the relay path corresponding to the total delay of less than or equal to the predetermined threshold The clustering proxy node is used as the second group of candidate relay nodes; delaying and packet loss detection are performed on the relay path corresponding to the first group of candidate relay nodes and the second group of candidate relay nodes, and based on delay and packet loss The detection result selects the relay node corresponding to the relay path with the best path quality as the relay node from the source node to the destination node.

Figure 14 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a second embodiment of the present invention. As shown in FIG. 14, the end-to-end overlay network transit system 200 according to the second embodiment of the present invention includes: a node joining and clustering unit 201 configured to join a node to a network and perform node clustering to set a part of nodes to Clustering the proxy nodes, and setting other nodes as host nodes belonging to the clustering proxy nodes; Meridian loop construction unit 202, configured to build its own based on other clustering proxy nodes for each clustering proxy node Meridian ring structure; and relay node The selecting unit 203 is configured to select a relay node from the source node to the destination node from the nodes on the Meridian ring structure.

Figure 15 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a third embodiment of the present invention. As shown in FIG. 15, an end-to-end overlay network transit system 300 according to a third embodiment of the present invention includes: a node joining and clustering unit 301 configured to join a node to a network and perform node clustering to set a part of nodes to Clustering the proxy node, and setting other nodes as the host node belonging to the clustering proxy node, wherein the node joining and clustering unit 301 calculates the information required by the distance prediction mechanism when joining the node to the network; And the relay node selection unit 302 is configured to select a relay node from the source node to the destination node from the clustering proxy node based on a distance prediction mechanism.

Figure 16 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a fourth embodiment of the present invention. As shown in FIG. 16, the end-to-end overlay network relay system 400 according to the fourth embodiment of the present invention includes: a Meridian ring construction unit 401 configured to join a node to a network to construct a Meridian ring structure, wherein the Meridian ring construction unit 401 Calculating information required by the distance prediction mechanism when joining the node to the network; and the relay node selection unit 402 is configured to select a relay node from the source node to the destination node from the nodes of the Meridian ring structure based on the distance prediction mechanism .

Figure 17 is a schematic block diagram showing an end-to-end overlay network relay system in accordance with a fifth embodiment of the present invention. As shown in FIG. 17, the end-to-end overlay network transit system 500 according to the fifth embodiment of the present invention includes: a node joining unit 501 configured to join a node to a network, wherein the node joining unit 501 calculates when a node joins a network. The information required by the distance prediction mechanism; and the relay node selection unit 502 is configured to select a relay node from the source node to the destination node based on the distance prediction mechanism.

In summary, the end-to-end overlay network transfer method and system according to the embodiment of the present invention can quickly find an effective relay node in the end-to-end overlay network, thereby establishing a shorter delay, higher efficiency, and stability. More powerful alternative routing for end-to-end high-performance communication.

Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software or a combination of both, in order to clearly illustrate hardware and software. Interchangeability, the composition and steps of the various examples have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the solution. A person skilled in the art can use different methods for implementing the described functions for each particular application, but such implementation should not be considered to be beyond the scope of the present invention. A person skilled in the art can clearly understand that, for the convenience and the cleaning of the description, the specific working processes of the system, the device and the unit described above can refer to the corresponding processes in the foregoing method embodiments, and details are not described herein again.

In the several embodiments provided herein, it should be understood that the disclosed systems, devices, and methods may be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be electrical, mechanical or otherwise.

The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software function unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may contribute to the prior art or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium. A number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a U disk, a removable hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and the like, which can store program codes. .

The above is only the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention. It should be covered by the scope of the present invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims

Rights request

An end-to-end overlay network transit method, the end-to-end overlay network includes a plurality of host nodes and a boot server, and the method is used to find a relay node from a source host node to a destination host node, Methods include:

The host node communicates with a boot server to obtain coordinate information, and the coordinate information is used to calculate a delay between two host nodes based on a distance prediction mechanism;

Performing clustering of the host node to use a part of the host node as a clustering proxy node, and other host nodes as host nodes belonging to the clustering proxy node;

A relay node from the source host node to the destination host node is selected from the clustering proxy nodes based on a distance prediction mechanism.

2. The method of claim 1, after the step of performing clustering of the host node and the distance prediction mechanism selecting a source host node to a destination host node from the clustering proxy node The steps of the relay node further include:

For each clustering proxy node, build your own based on other clustering proxy nodes.

Meridian ring structure;

And the step of selecting the relay node from the source host node to the destination host node from the clustering proxy node based on the distance prediction mechanism is specifically:

A relay node from the source host node to the destination host node is selected from the clustering proxy nodes on the Meridian ring structure based on the distance prediction mechanism.

3. The method of claim 2, further comprising

Updating at least one of the following information between the host node and the clustering proxy node: network status of the host node, delay of the host node to the boot server, and update time of the host node; and clustering proxy node and BS server At least one of the following information is updated: the network status of the clustering node, the delay from the clustering node to the boot server, and the update time of the clustering node.

4. The method of claim 2, further comprising:

Each cluster node maintains and updates its own Meridian ring structure, where each agent node maintains and updates at least one of the following: IP address, coordinate information, and other cluster agent nodes on the Meridian ring structure The delay and update time to the clustering agent node.

5. The method of claim 2, wherein the distance prediction mechanism is from Meridian The step of selecting a relay node from the source host node to the destination host node in the clustering proxy node on the ring structure specifically includes:

Selecting from the source host node based on the distance prediction mechanism from the clustering proxy node on the Meridian ring structure of the clustering proxy node to which the source host node belongs and the clustering proxy node on the Meridian ring structure of the clustering proxy node to which the destination host node belongs The relay node to the destination host node.

6. The method according to claim 5, wherein the distance prediction mechanism is based on a clustering proxy node on a Meridian ring structure of a clustering proxy node to which a source host node belongs and a clustering proxy node to which the destination node belongs. The step of selecting a relay node from the source host node to the destination host node in the clustering proxy node on the ring structure specifically includes:

The clustering proxy node to which the source host node belongs calculates the delay of each clustering proxy node to the destination host node on its own Meridian ring structure by distance prediction based on the coordinates of the destination host node, and respectively stores the stored delay The delays of each clustering proxy node on the Meridian ring structure to the clustering proxy node to which the source host node belongs are added to obtain the total delay of each relay path associated with each clustering proxy node;

Selecting, as a candidate relay node, a clustering proxy node associated with a relay path whose total delay is less than a predetermined threshold;

Performing delay and packet loss detection on the relay path associated with the candidate relay node, and selecting the candidate relay node related to the relay path with the best path quality as the slave source host based on the delay and packet loss rate detection result The relay node from the node to the destination host node.

7. The method according to claim 6, wherein the step of selecting a clustering proxy node associated with a relay path whose total delay is less than a predetermined threshold as the candidate relay node further comprises: In addition to the clustering proxy nodes associated with the relay path of the threshold, some clustering proxy nodes are randomly selected as candidate relay nodes.

The method according to any one of claims 2 to 7, wherein the step of performing clustering of the host node specifically includes:

The host node is clustered based on a round trip delay of the host node to the boot server and an IP address of the host node.

The method according to any one of claims 2 to 7, wherein the distance prediction mechanism is GNP distance prediction, and the coordinates for calculating a delay between two host nodes based on a distance prediction mechanism Information is GNP coordinates,

The host node communicates with a boot server to calculate two for calculating based on a distance prediction mechanism The steps of the coordinate information of the delay between the host nodes further include:

Sending, by the host node, a registration request to the boot server;

The boot server authenticates the host node, and sends a success response message to the host node if the authentication succeeds, where the success response message includes information of multiple landmark nodes, where The landmark node is a device that is previously deployed in the network and whose information is stored on the boot server, and the information of the landmark node includes at least the IP address or GNP coordinates of the landmark node.

10. The method of claim 9, wherein the step of constructing its own Meridian ring structure based on other clustering proxy nodes for each clustering proxy node comprises:

The boot server calculates a delay between each cluster proxy node and other cluster proxy nodes by using GNP distance prediction, and selects a certain number of candidate cluster proxy nodes for each cluster proxy node based on the delay;

Each clustering proxy node stores its own clustering node information into its own Meridian ring structure based on the round-trip delay with other clustering proxy nodes to construct its own Meridian ring structure.

The method of claim 9, wherein the step of selecting a relay node from the source host node to the destination host node from the clustering proxy nodes on the Meridian ring structure based on the distance prediction mechanism comprises:

The source host node sends a relay node selection request message to the clustering proxy node to which the owning cluster proxy node and the destination host node belong, and the relay node selection request message includes the source host node and the destination host node. GNP coordinates;

The clustering proxy node to which the source host node belongs calculates the delay of each clustering proxy node on its own Meridian ring structure to the destination host node by GNP distance prediction based on the GNP coordinates of the destination host node, and Adding the delay of each clustering proxy node on the Meridian ring structure to the source host node to obtain the total delay of each relay path;

Selecting, as a first group of candidate relay nodes, a clustering proxy node corresponding to the relay path whose total delay is less than or equal to a predetermined threshold;

The clustering proxy node to which the destination host node belongs is calculated based on the GNP coordinates of the source host node, and the delay of each clustering proxy node on the own Meridian ring structure to the source host node is calculated by GNP distance prediction, and is itself The delay of each clustering proxy node on the Meridian ring structure to the destination host node is added to obtain the total delay of each relay path; Selecting, as a second group of candidate relay nodes, a clustering proxy node corresponding to the relay path whose total delay is less than or equal to a predetermined threshold;

Performing delay and packet loss detection on the relay path corresponding to the first group of candidate relay nodes and the second group of candidate relay nodes, and selecting the relay path with the best path quality based on the delay and the packet loss detection result The relay node acts as a relay node from the source host node to the destination host node.

12. An end-to-end overlay network transit system, the end-to-end overlay network includes a plurality of host nodes and a boot server, and the system is configured to search for a relay node from a source host node to a destination host node, The end-to-end overlay network transit system includes:

a communication unit, configured to enable the host node to communicate with a boot server to obtain coordinate information, where the coordinate information is used to calculate a delay between two host nodes based on a distance prediction mechanism;

a clustering unit, configured to perform clustering of the host node, to use a part of the host node as a clustering proxy node, and other host nodes as a host node belonging to the clustering proxy node; The prediction mechanism selects a relay node from the source host node to the destination host node from the clustering proxy node.

13. The end-to-end overlay network relay system of claim 12, further comprising: a building unit coupled between the clustering unit and the selecting unit for each clustering proxy node, The clustering proxy node builds its own Meridian ring structure based on it;

And the selecting unit is specifically configured to select, according to the distance prediction mechanism, a relay node from the source host node to the destination host node from the clustering proxy nodes on the Meridian ring structure.

14. The end-to-end overlay network relay system of claim 13, further comprising a first update unit, configured to update at least one of the following information between the host node and the clustering proxy node: network status of the host node , the delay from the host node to the boot server and the update time of the host node;

a second updating unit, configured to update at least one of the following information between the clustering proxy node and the BS server: a network status of the clustering proxy node, a delay of the clustering proxy node to the boot server, and a clustering proxy node Update time.

15. The end-to-end overlay network relay system of claim 13, further comprising: a maintenance update unit for causing each cluster agent node to maintain and update its own Meridian ring structure, wherein each agent node maintains And updating at least one of the following information: IP address of other clustering proxy nodes on the Meridian ring structure, coordinate information, delay to the clustering proxy node, and update time.

The end-to-end overlay network transit system according to claim 13, wherein the selection unit is specifically configured to:

The end-to-end overlay network transit system according to claim 16, wherein the selection unit is specifically configured to:

The clustering proxy node to which the source host node belongs is calculated based on the coordinates of the destination host node, and the delay of each clustering proxy node on the own Meridian ring structure to the destination host node is calculated by distance prediction, and respectively stored and stored The delays of each clustering proxy node on the Meridian ring structure to the clustering proxy node to which the source host node belongs are added to obtain a total delay of each relay path associated with each clustering proxy node;

18. The end-to-end overlay network transit system of claim 17, wherein the selection unit is further configured to:

In addition to clustering proxy nodes associated with relay paths whose total delay is less than a predetermined threshold, some clustering proxy nodes are randomly selected as candidate relay nodes.

The end-to-end overlay network relay system according to any one of claims 13 to 18, wherein the clustering unit is specifically configured to:

The end-to-end overlay network transit system according to any one of claims 13 to 18, wherein the distance prediction mechanism is GNP distance prediction, and the method is used to calculate two host nodes based on a distance prediction mechanism. The coordinate information of the time delay is the GNP coordinate.

The end-to-end overlay network transit system further includes:

a registration requesting unit, configured to cause the host node to send a registration request to the boot server; An authentication unit, configured to connect to the registration request unit and the communication unit, to enable the boot server to authenticate the host node, and send the successfully sent to the host node if the authentication succeeds a response message, the success response message includes information of a plurality of landmark nodes, wherein the landmark node is a device that is previously deployed in the network and whose information is stored on the boot server, and the information of the landmark node includes at least The IP address or GNP coordinates of the landmark node.

The end-to-end overlay network transit system according to claim 20, wherein the constructing unit is specifically configured to:

Having the boot server calculate a delay between each cluster proxy node and other cluster proxy nodes by using GNP distance prediction, and select a certain number of candidate cluster proxy nodes for each cluster proxy node based on the delay;

Each clustering proxy node is configured to store other clustering proxy node information into its own Meridian ring structure based on the round-trip delay with other clustering proxy nodes, thereby constructing its own Meridian ring structure.

22. The end-to-end overlay network transit system of claim 20, wherein the selection unit is specifically configured to:

And causing the source host node to send a relay node selection request message to the clustering proxy node to which the owning cluster proxy node and the destination host node belong, where the relay node selection request message includes the source host node and the destination host The GNP coordinates of the node;

The clustering proxy node to which the source host node belongs is calculated based on the GNP coordinates of the destination host node, and the delay of each clustering proxy node on the own Meridian ring structure to the destination host node is calculated by GNP distance prediction. And adding the delay of each clustering proxy node on the Meridian ring structure to the source host node to obtain the total delay of each relay path;

The clustering proxy node to which the destination host node belongs is calculated based on the GNP coordinates of the source host node, and the delay of each clustering proxy node on the own Meridian ring structure to the source host node is calculated by GNP distance prediction, and The delay of each clustering proxy node on the own Meridian ring structure to the destination host node is added to obtain the total delay of each relay path; and the aggregation corresponding to the relay path whose total delay is less than or equal to a predetermined threshold is selected. a class proxy node as a second group of candidate relay nodes; Performing delay and packet loss detection on the relay path corresponding to the first group of candidate relay nodes and the second group of candidate relay nodes, and selecting the relay path with the best path quality based on the delay and the packet loss detection result The relay node acts as a relay node from the source host node to the destination host node.