KR101079660B1 - An overlay-based resource monitoring scheme for social applications in manet and method thereof - Google Patents

Abstract

The present invention relates to an overlay based resource monitoring system and method on a mannet. That is, according to the present invention, clusters adjacent to each other on a mannet are formed, nodes using the same application program are formed as a group, an overlay network is formed between clusters having the same group members, and resource information between group members. Can be passed through the overlay network to reduce unnecessary overhead.

Description

OVERLAY-BASED RESOURCE MONITORING SCHEME FOR SOCIAL APPLICATIONS IN MANET AND METHOD THEREOF}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a resource monitoring system on a mobile ad-hoc network (MANET), and in particular, nodes adjacent to each other on a mannet form clusters and use the same application program. After forming nodes into groups, an overlay network is formed between clusters having the same group member so that resource information between group members is transmitted through the overlay network. The present invention relates to an overlay-based resource monitoring system and method on a mannet that can reduce overhead.

As mobile devices supporting wireless internet and the opportunities for social interaction between these devices increase, the importance of mobile social software is also increasing. For this social interaction, it is necessary to monitor resource information of each mobile device in real time so that the application can be applied to a dynamic mobile ad-hoc network (MANET) environment.

For this purpose, resource monitoring on the mannet is important. At this time, the resource monitoring system on the existing mannet is largely proactive (a method of delivering resource information before a request is made), a reactive (a method of delivering a request for resource information whenever a request is made), And it can be divided into three types of hybrid method that combines the above two methods.

The proactive method periodically broadcasts resource information to neighboring nodes. Even if a request for resource information for a specific mobile device occurs, the node already has the resource information of the device, and thus there is an advantage in that the information can be used immediately. However, this method has a disadvantage of generating unnecessary traffic overhead if the node does not want the resource information of the neighboring mobile devices.

In the reactive method, when a node requests resource information for a specific mobile device, the corresponding mobile device transmits resource information to the node. This method does not create unnecessary traffic overhead, as only the required resource information is conveyed. However, when most nodes on a mannet periodically or frequently request resource information, they incur much overhead compared to the proactive method.

The hybrid scheme is to solve the problems of the proactive scheme and the reactive scheme. The clusters are generated between adjacent nodes, and the nodes in the cluster report resource information periodically to the cluster head. If a node outside the cluster requests resource information, the resource information is exchanged through communication between cluster heads. Using this method, even if the request for resource information occurs frequently, the overhead is reduced as compared to the reactive method, and the ratio of broadcasting is significantly reduced, which significantly reduces unnecessary traffic overhead compared to the proactive method. .

However, when attempting to apply the group-based resource management system to such a hybrid resource management monitoring system, there are two major problems.

First, since the existing method does not have information about the topology of the group members, to request the resource information of all group members, a query must be sent throughout the network. Therefore, this method has a problem of generating unnecessary query overhead.

Second, when sending a query across the network, the cluster head must send a query to the surrounding clusters, where each cluster head must wait for a response from all the surrounding clusters. In this case, since the group member topology is not included, a query must be sent to a cluster to which the group member does not belong. This causes a delay in waiting for a response in addition to generating unnecessary traffic.

Therefore, in the present invention, clusters adjacent to each other on a mannet are formed, nodes using the same application program are formed into groups, and then overlay networks are formed between clusters having the same group members, thereby providing resource information between group members. It is an object of the present invention to provide an overlay-based resource monitoring system and method on a mannet that can reduce unnecessary overhead by allowing a network to be transmitted through an overlay network.

The present invention described above is an overlay-based resource monitoring system on a mannet, and transmits a group-based overlay participation request message to a head node of an adjacent neighbor cluster for group-based overlay, and responds to the request message from the head node of the neighbor cluster. A head node of the root cluster that receives the message and forms an overlay network, and if there is the overlay participation request message from the head node of the root cluster, checks whether there is a member node for the group-based overlay; And if there is a node, the head node of the neighboring cluster that sends the response message to the head node of the root cluster.

In addition, when the overlay network is formed, the head node of the root cluster transmits a group query for monitoring the overlay-based resource to a neighbor cluster included in the overlay network, and transmits a group query for the group query from the neighbor cluster. And receiving a response message to perform resource monitoring of the member node located in the neighbor cluster.

In addition, when the head node of the neighboring cluster receives the group query, the head node searches for a resource of a member node corresponding to the group query and transmits it as a response message to the group query, and includes the group query in the overlay topology. It is characterized by transmitting to another neighboring cluster head node.

The present invention also provides an overlay-based resource monitoring method on a mannet, the method comprising: transmitting an overlay participation request message from a head node of a root cluster to a head node of a neighbor cluster, and a head of the neighbor cluster receiving the overlay participation request message; Checking whether there is a member node to participate in the overlay, if the member node exists, the head node of the neighboring cluster sends an overlay participation response message to the root cluster, and the head node of the root cluster. Forming an overlay network using the overlay participation response message.

The present invention also provides an overlay-based resource monitoring method on a mannet, the method comprising: requesting the overlay-based resource monitoring from a group-based coordinator of a root cluster; and when the resource monitoring request is received, a head node of the root cluster is included in an overlay network. Transmitting a group query for monitoring the resource to the neighboring cluster, and the head node of the neighboring cluster receiving the group query searches for resources of the member node corresponding to the group query and sends a response message to the group query. And transmitting, at the head node of the root cluster, the response message for the group query from the head node of the neighboring cluster to perform the overlay-based resource monitoring.

The transmitting of the response message to the group query may include: searching a resource of a member node corresponding to the group query by a head node of a neighboring cluster that receives the group query; And sending the group query to other neighboring cluster head nodes in the periphery of the overlay network.

In the present invention, clusters adjacent to each other on a mannet are formed, and nodes using the same application are formed as a group, and then an overlay network is formed between clusters having the same group members, so that resource information between group members is overlayed. It is advantageous to reduce unnecessary overhead by allowing it to pass through the network.

1 illustrates an overlay network on a mannet according to an embodiment of the present invention;
2 is a flowchart illustrating an overlay-based resource monitoring operation control on a mannet according to an embodiment of the present invention;
3 is a diagram illustrating a simulation environment parameter table of overlay-based resource monitoring;
4 is an exemplary view illustrating a graph of experiment results when nodes are distributed according to an embodiment of the present invention;
5 is an exemplary view illustrating a graph of experiment results when the moving speeds of nodes according to an embodiment of the present invention are varied;
6 is an exemplary view illustrating a graph of experimental results when several groups do not overlap according to an embodiment of the present invention;
7 is a graph illustrating an experimental result when several groups overlap each other according to an exemplary embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings will be described in detail the operating principle of the present invention. In the following description of the present invention, when it is determined that a detailed description of a known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. Terms to be described later are terms defined in consideration of functions in the present invention, and may be changed according to intentions or customs of users or operators. Therefore, the definition should be made based on the contents throughout the specification.

Node clustering

In the present invention, several nodes form one cluster, and each cluster is composed of one head node and other claim nodes. The claim node informs the head node of its resource information, and the head node collects resource information of all nodes in the cluster.

To form a cluster, the head node of the cluster must be selected. Once a node enters the network, it becomes a head node. If there is no head node in the cluster, the node continues to maintain the head node of the cluster. If there are multiple cluster head nodes around, the node with the oldest generation time becomes the head node of the cluster, and the remaining nodes become claims nodes. If the head node of the cluster disappears in a specific cluster due to network interference or cluster movement, the node with the oldest generation time becomes the head node of the cluster.

Overlay formation

Group-based overlays are formed by two messages: an overlay join request (JREQ) and a join reply (JREP). At this time, the cluster with group coordinator among the clusters on the mannet becomes the root cluster. When the root cluster sends a JREQ message to the head node of the neighboring cluster, the head node of the neighboring cluster checks whether the member is in its own cluster. Check it.

The head node of the neighboring cluster notifies this by sending a JREP to the cluster that sent the JREQ if the member is in its cluster, and becomes a child cluster of the cluster that received the JREQ. Next, the head node of the neighbor cluster forwards the JREQ back to the head node of the neighbor cluster.

The above process is repeated to find out which cluster the group members belong to and form an overlay network.

1 illustrates a concept of forming an overlay network for overlay-based resource monitoring on a mannet according to an embodiment of the present invention.

Referring to FIG. 1, a black node in a cluster means a group member using the same application. Since there is a group coordinator 150 in the cluster 1 (100), cluster 1 (100) is a root cluster (root cluster).

If node A, the head node of cluster 1 (100), sends JREQ to node B and node D, which are the head nodes of neighboring clusters (102, 106), because cluster 2 (102) has its group members, Node B, the head node of cluster 2 (102), sends JREP to node A, the head node of cluster 1 (100).

On the other hand, since there is no corresponding group member in cluster 4 106, node D, which is the head node of cluster 4 106, does not send JREP to node A, which is the head node of cluster 1 (100).

At this time, cluster 2 (102) becomes a child cluster of cluster 1 (100), and node B, which is the head node of cluster 2 (102), forwards JREQ to node C, which is the head node of the neighboring cluster. Node C then sends a JREP back to Node B, since there are members of that group in its cluster, and cluster 3 104 becomes a child cluster of cluster 2 102.

As described above, an overlay network is formed through transmission / reception of JREQ and JREP with neighboring clusters.

2 illustrates an operation control flow for forming an overlay network for overlay-based resource monitoring on a mannet according to an embodiment of the present invention. Hereinafter, embodiments of the present invention will be described in detail with reference to FIGS. 1 and 2.

First, cluster 1 (100) in which the group coordinator 150 exists becomes a root cluster and transmits JREQ to cluster 2 (102) and cluster 4 (106) which are neighboring neighbor clusters (S200).

Then, node B and node D, which are the head nodes of cluster 2 (102) and cluster 4 (106) that have received the JREQ from the cluster 1 (100), which is the root cluster, search whether the group member exists in their cluster (S202). ).

Subsequently, node B and node D, which are the head nodes of cluster 2 102 and cluster 4 106, send JREP to the head node of cluster 1 100 in response to JREQ when the group member is found to exist. And a child cluster of cluster 1 (100).

That is, since the group member exists in cluster 2 (102), node B, the head node of cluster 2 (102), sends JREP to node A, the head node of cluster 1 (100), and cluster 4 (106). Since there is no corresponding group member, node D, which is the head node of cluster 4 (106), does not send JREP to node A, which is the head node of cluster 1 (100).

Subsequently, Node B, which is the head node of Cluster 2 (102), transmits JREQ to Cluster 3 (104), which is the neighboring cluster (S204). At this time, node C, which is the head node of cluster 3 104, has a member of the group in its cluster, and thus sends JREP to node B again. Cluster 3 104 is a child cluster of cluster 2 102. Becomes

As above, the overlay network is formed through the transmission and reception of JREG and JREP between neighboring clusters.

Subsequently, when a resource monitoring request for the same group members in the cluster is generated from the group coordinator 150 in the state where the overlay network is formed as described above, the neighbor cluster forming the overlay network from the head node of the cluster 1 (100) which is the root cluster. Resource monitoring requests are sent to cluster 2 (102) and cluster 3 (104). Node B and node C, which are the head nodes of cluster 2 (102) and cluster 3 (104), are resources of group members in their cluster. The information is transmitted to cluster 1 (100), which is the root cluster, so that resource monitoring is performed.

As described above, the present invention assumes a situation in which a user using a mobile device such as a net book, a mobile phone, or a UMPC is connected to an ad hoc network environment.

For example, users with similar interests use a photo sharing program to create groups to share their photos. One user in these groups becomes the coordinator. The coordinator monitors the resource status of group members and adjusts resources or changes the environment accordingly. At this time, for example, in the photo sharing group, the coordinator checks the resource information of the group member so that all members in the group can share the photo, and then determines the appropriate picture quality. Similarly, in the music sharing program, the coordinator can determine the quality of music, and several application groups can be simultaneously created on one mannet according to the number of applications.

Overlay Management

In the present invention, four methods of receiver-initiated join, accuracy triggered on-demand recreation, fast disconnection detection, and redundant message aggregation for overlapping groups are used to manage the overlay network generated as described above.

When the head node of the cluster detects that the group overlay is disconnected, it reconnects with the group member using a receiver-initiated join. The receiver-initiated join finds the head nodes of the cluster that have the same group members at close range. If there is no cluster with group members at close range, the hop count is increased by one. This process is repeated until the maximum value of the predetermined hop count is reached.

If no head node is found for a cluster that has the same group member, the node is isolated and terminates the join process. At this time, if the hop count maximum value is too small, many nodes will be isolated, and conversely, if the hop count maximum value is too large, a lot of traffic overhead will be generated. Accordingly, the present invention will be described with an example in which the maximum hop count number is set to two.

On the other hand, as the number of isolated clusters grows, the accuracy of group queries becomes poor. Accuracy is calculated as the ratio of the number of responding members to the total number of members. If the accuracy drops below a certain number, accuracy triggered on-demand recreation is performed, recreating the overlay. In the present invention, 0.5 is recommended as the limit value of accuracy.

In addition, due to node movement or network problems, communication between overlay networks may be broken. In other words, if a membership changes due to a particular node terminating an application, the overlays of the parent and child clusters connected to this node will be incorrectly connected. In order to be able to identify and fix such erroneous or broken connections quickly, the present invention periodically uses a survival confirmation message.

In addition, several group overlays may exist simultaneously. In this case, since overlapping groups are generated, duplicate control messages are generated. To prevent this, the present invention aggregates and sends duplicate control messages. The collected control messages have a list of groups to be delivered, which are then sent to that group.

Group queries

If the group coordinator wants to collect resource information of the same group member node, the group coordinator sends a group query request (GREQ) message to the head node of the cluster to which the group coordinator belongs.

Then, the head node of the cluster delivers GREQ to the head node of the child cluster, and the head node of the child cluster repeats this process until there are no more head nodes of the cluster to deliver. If there is no more head node in the cluster to send, the head node of that cluster forwards the group query reply (GREP) message to the parent cluster's head node, and the parent node's head node receives the GREP message from the child cluster. Merges its own GREP message and passes it back to the parent cluster. Finally, the group cluster knows about GREP messages that contain information about all group members.

That is, when gathering group information, as shown in FIG. 1, the group coordinator 150 sends a GREQ message to node A, which is a head node of the cluster 1 100. Node A then sends GREQ back to node B, which is the head node of child cluster 2 (102), and node B sends back to node C, which is the head node of child cluster 3 (104). Node C then gathers the resource information of the group members and sends GREP to node B, the head node of the parent cluster, and node B combines the information received from node C with the resource information of the group members of cluster 2 (102) and sends GREP to node A. Send to Node A finally delivers a message containing the resource information of the group member to the group coordinator 150.

evaluation

The resource monitoring system of the present invention was evaluated in comparison with the methods proposed by Tuduce and Gross.

There are three criteria for evaluation: accuracy, response time, and traffic overhead. Accuracy is the number of members answered divided by the total number of members in the group. The response time is from one hour of querying to the time of receiving a reply. Traffic overhead is calculated by adding (1) queries between neighboring nodes and (2) queries to maintain group queries and overlays.

In the case of one group, three groups existed but did not overlap, and three groups overlapped each other. 3 shows a simulation environment.

One group

One group was evaluated in two separate cases.

(1) Position of various members: This case was evaluated to see how the distance between members would affect the evaluation. (2) Various moving speeds: This case was evaluated to see how the moving speed of the nodes affects the evaluation. In the former case, the moving speed is zero.

Figure 4 shows the results of the experiment for the position of (1) the various members in the case of one group. Referring to FIG. 4, the accuracy is not significantly different but the response time is about 90%. Traffic overhead is cut by half. Tuduce and Gross's methods are heavily influenced by the formation of faulty overlays. Spreading members widespreadly affects traffic overhead or response time because the probability of a message being delivered to the wrong overlay is high. Since the present invention knows in advance about the topology of the members, this adverse effect can be avoided.

Figure 5 shows the experimental results for (2) the various moving speed of the case of one group. Referring to Figure 5, in this experiment, the members are up to three hops apart, and the number of group members is ten. In this case, both show similar accuracy.

However, the present invention shows good results with both response speed and traffic overhead. Increasing the node's moving speed increases the message drop probability, which reduces the accuracy in both cases. However, Tuduce and Gross's method has a greater probability of dropping messages, which in turn affects the accuracy. In particular, as the number of message drops increases, so does the waiting time for a query, which increases the response time.

Multiple groups do not overlap

6 shows experimental results when several groups do not overlap. Referring to FIG. 6, both cases show that the accuracy and the reaction rate are constant regardless of the number of groups. However, the difference is up to 60% in proportion to the number of groups of traffic overhead. However, the present invention can reduce the traffic overhead even more by using the overhead to reduce unnecessary message delivery.

When multiple groups overlap

7 shows experimental results when several groups overlap. In this experiment, some nodes run two applications at the same time, belonging to both groups. Therefore, duplicate overlay management traffic may occur. Both accuracy and reaction rate were increased compared to the case where the groups did not overlap. This is because the number of nodes sharing a particular application has increased.

At this time, the traffic overhead was divided into two cases of merging the traffic overhead of the intermediate node and the case of not merging. In the former case, the overhead of the present invention is increased by about twice as compared with the case where there is no overlap.

From these results, we found that the effect of merging on overhead increases as the number of overlapping groups increases. Without merging, there was an increase of only 9% overhead compared to the non-overlapping, even though seven groups were created.

As described above, in the present invention, clusters adjacent to each other on the mannet are formed, and nodes using the same application program are formed into groups, and then an overlay network is formed between clusters having the same group members to form group members. Resource information between the nodes is transmitted through the overlay network, thereby reducing unnecessary overhead.

Meanwhile, in the above description of the present invention, specific embodiments have been described, but various modifications may be made without departing from the scope of the present invention. Accordingly, the scope of the invention should not be limited by the described embodiments but should be defined by the appended claims.

100: cluster 1 102: cluster 2
104: cluster 3 106: cluster 4
150: coordinator

Claims (6)

As overlay based resource monitoring system on Mannet,
A head of a root cluster that forms an overlay network by transmitting a group-based overlay participation request message to a head node of an adjacent neighbor cluster for group-based overlay, and receiving a response message for the overlay participation request message from the head node of the neighboring cluster. Nodes,
Upon receiving the overlay participation request message from the head node of the root cluster, checks whether there is a member node for the group-based overlay, and if there is the member node, sends the response message to the head node of the root cluster. Head node of the neighboring cluster
Overlay-based resource monitoring system on the mannet including.
The method of claim 1,
The head node of the root cluster,
When the overlay network is formed, a member located in the neighbor cluster by transmitting a group query for overlay-based resource monitoring to a neighbor cluster included in the overlay network and receiving a response message for the group query from the neighbor cluster. An overlay-based resource monitoring system on a mannet, characterized in that the resource monitoring of the node.
The method of claim 2,
The head node of the neighbor cluster,
When the group query is received, the resource of the member node corresponding to the group query is searched and transmitted as a response message to the group query, and the group query is transmitted to other neighboring cluster head nodes included in the overlay topology. Overlay-based resource monitoring system on a mannet characterized in that.
As an overlay-based resource monitoring method on Mannets,
Transmitting an overlay join request message from the head node of the root cluster to the head node of the neighboring cluster;
Checking whether there is a member node to participate in an overlay at a head node of the neighboring cluster that receives the overlay participation request message;
If the member node exists, the head node of the neighboring cluster transmits an overlay join response message to the root cluster;
Forming an overlay network using the overlay participation response message in the head node of the root cluster;
Overlay-based resource monitoring method on the mannet including.
As an overlay-based resource monitoring method on Mannets,
Requesting overlay-based resource monitoring from the group-based coordinator of the root cluster;
Transmitting a group query for resource monitoring to a neighboring cluster included in an overlay network by a head node of the root cluster when the overlay-based resource monitoring request is received;
Transmitting a response message to the group query by searching for a resource of a member node corresponding to the group query by a head node of a neighbor cluster receiving the group query;
Receiving the response message for the group query from the head node of the neighboring cluster at the head node of the root cluster to perform the overlay-based resource monitoring
Overlay-based resource monitoring method on the mannet including.
The method of claim 5, wherein
Sending a response message to the group query,
Searching for a resource of a member node corresponding to the group query by a head node of a neighbor cluster receiving the group query;
The head node of the neighboring cluster sends the group query to the head node of another neighboring cluster included in the overlay network
Overlay-based resource monitoring method on the mannet including.

Images