JP2006201896A - Network system and mobile communication node - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform distributed coordinated processing of high quality by efficiently constructing a grid service environment on an ad-hoc network. <P>SOLUTION: A client node 20c gets into client operation at the time of requesting the execution of jobs to inquire throughputs and participation intention of other node groups as a client operation, determines nodes capable of accepting the job processing based on response contents therefrom, requests the execution and distribution of jobs thereto, and tabulates job processing results. Server nodes 20s-1 to 20s-n get into server operations at the time of executing the requested jobs to response to the inquiry from the client node, for whether the execution of the requested jobs is accepted or not based on the throughputs and participation intentions thereof, and executes the jobs when accepting the execution, and transmits processing results to the client node 20c. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a network system and a mobile communication node, and more particularly to a network system that performs communication on a wireless network and a mobile communication node that performs wireless communication.

  In recent years, digital wireless communication technologies such as wireless LAN and Bluetooth, which are standardized by IEEE (the Institute of Electrical and Electronic Engineers) 802.11, have been developed. Mobile computing used is rapidly spreading, and ad hoc network technology that takes advantage of both characteristics is drawing attention.

  An ad hoc network does not require an access point such as a wireless LAN (a repeater (wireless station) connected to the LAN) and can be connected wirelessly (for example, a notebook computer, a PDA, a mobile phone, etc.) It is a network composed of Since an ad hoc network does not have a base station or an access point, it can construct a network at a low cost and is regarded as a promising means for constructing a simple network in a limited area.

  In an ad hoc network, mobile nodes communicate wirelessly. For example, when nodes A and B communicate, if the electromagnetic waves used for communication are within a reachable range, nodes A and B perform direct communication. Also, if the nodes A and B are located in a range where the electromagnetic waves cannot reach directly, the node C existing at a position where the electromagnetic waves from the nodes A and B can reach relays the communication contents of the nodes A and B. Realize communication.

  Therefore, the node itself constituting the ad hoc network is a subject (termination point) that performs transmission / reception of communication, and at the same time functions as a subject (router) that relays, that is, routes, communication with other nodes as termination points. To prepare.

  On the other hand, in recent years, a technology called grid computing has been researched and developed. Grid computing is a distributed cooperative processing technology that connects geographically distributed computer resources via a network and uses them as if they were one computer system.

  Grid computing, for example, connects computers that are inferior to supercomputers with a single computing capacity to tens to hundreds of computers via the Internet, and draws computing power that exceeds supercomputers, such as engineering and medicine. It is a large amount of calculation in the field.

  In a grid computing service environment (grid service environment), the user does not need to be aware of which computer is executing the calculation process or from which database the data is transferred, and is a virtual supercomputer. Can be used as

  In the grid service environment, there is a server called a scheduler, which authenticates each node in the grid service environment and manages computing resource information held by each node. Further, the scheduler receives a request from the client and plays a role of distributing the job to each node and scheduling.

As a conventional distributed processing technique, a technique for performing a distributed process by evaluating whether or not to accept a processing request by evaluating its own ability has been proposed (for example, Patent Document 1).
Japanese Patent Laid-Open No. 07-129520 (paragraph numbers [0014] to [0018], FIG. 2)

  In the ad hoc network described above, each node that can move without having a base station performs wireless communication on an equal basis, so that nodes join and leave the ad hoc network as needed.

  Further, due to the movement of the node, a change in the physical positional relationship between the nodes and a change in the network topology due to a change in the propagation state of the electromagnetic wave may occur in a complicated manner. In addition, since the node is a mobile communication device, it is assumed that it is battery-powered. Naturally, each node is conscious of saving power consumption, and the protocol operating on the ad hoc network suppresses unnecessary communication that consumes power. There is a need to.

  In this way, the ad hoc network has completely different characteristics from the network such as Ethernet (registered trademark) based on the existing wired communication technology, so applications and middleware tuned on the existing wired communication technology are not It does not operate efficiently on ad hoc networks. Therefore, even when a grid service environment, which has been researched and developed especially on the premise of wired communication technology, is realized on an ad hoc network, there are the following difficulties, for example.

  On an ad hoc network, the temporary inability to communicate due to changes in the propagation status of electromagnetic waves, the battery running out of nodes, and the detachment from the ad hoc network due to movement can be complicated. When built on an ad hoc network, it is expected to be extremely unreliable.

  On an ad hoc network, since all nodes can freely join and leave the network, it is not possible to install a scheduler as a fixed station that existed in the grid service.

  The nodes that make up the ad hoc network are limited in terms of processing power compared to the workstations and numerical computers that make up the existing grid computing. Further, since the bandwidth of the wireless communication interface is narrower than the bandwidth of the wired communication interface, the performance is inferior to that of the existing grid service environment.

  Because of these obstacles, to implement grid services on an ad hoc network, instead of applying the grid service method based on the existing wired communication technology, the limited processing capability, radio bandwidth, and complexity of mobile terminals It is necessary to realize an optimized grid service in consideration of the characteristics of the ad hoc network such as a network topology that changes rapidly.

  The present invention has been made in view of these points, and an object thereof is to provide a network system that efficiently constructs a grid service environment on an ad hoc network and realizes high-quality distributed cooperative processing.

  Another object of the present invention is to provide a mobile communication node that efficiently constructs a grid service environment on an ad hoc network and realizes high-quality distributed cooperative processing.

  In the present invention, in order to solve the above-mentioned problem, in the network system 1 that performs wireless communication as shown in FIG. The client node 20c that inquires about the ability and intention to participate, determines the node that can accept job processing based on the response content, distributes and requests job execution, and totals the job processing results, and the requested job. When executing, it becomes a server operation, responding to the inquiry from the client node, whether or not to accept the execution of the requested job, depending on the processing capability and willingness to participate. If accepted, the job is executed And server nodes 20s-1 to 20s-n for transmitting the processing results to the client node 20c, As needed, the network system 1 is provided which is characterized in that the distributed cooperative processing on an ad hoc network 1a constructed by a plurality of nodes that can be moved by wireless communication.

  Here, the client node 20c becomes a client operation when requesting execution of a job, inquires about processing capability and intention to participate in other nodes, and can accept job processing based on response contents. Nodes are determined, job execution is distributed and requested, and job processing results are aggregated. When the requested job is executed, the server nodes 20s-1 to 20s-n operate as a server and execute the requested job depending on the processing capability and the intention to participate in response to an inquiry from the client node. Whether or not to accept is returned, and if accepted, the job is executed and the processing result is transmitted to the client node 20c.

  In the network system of the present invention, the client node inquires about the processing capability and intention to participate in other node groups, determines the node that can accept the job processing based on the response content, and distributes the job execution. In response to the inquiry from the client node, the server node responds whether or not to accept execution of the requested job depending on the processing capability and willingness to participate. If accepted, the job is executed and processed. The result is transmitted. As a result, a grid service environment based on wired communication technology can be efficiently constructed on an ad hoc network with many unstable elements, and high-quality distributed cooperative processing can be realized.

  In the mobile communication node of the present invention, when the job management unit requests execution of a job, it becomes a client operation, inquires other nodes for processing capability and intention to participate, and based on the response content. Determine nodes that can accept job processing, distribute and request job execution, aggregate job processing results, and execute the requested job as a server operation, in response to inquiries from clients Depending on the processing capability and willingness to participate, it is configured to respond whether or not to accept execution of the requested job. As a result, a grid service environment based on wired communication technology can be efficiently constructed on an ad hoc network with many unstable elements, and high-quality distributed cooperative processing can be realized.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a principle diagram of a network system. The network system 1 performs distributed cooperative processing (grid service) on a network (ad hoc network 1a) constructed by a plurality of mobile communication nodes (hereinafter simply referred to as nodes) that can move by wireless communication without having a fixed station. System.

  In the network system 1, each node cooperates while performing wireless communication, and realizes processing that cannot be realized by a single node. The grid service realized on the ad hoc network 1a of the network system 1 is also referred to as an ad hoc grid service hereinafter.

  When the client node 20c requests the execution of the job, the client node 20c becomes a client operation, inquires about the processing capability and the intention to participate in the other node group (query: query), and performs job processing based on the response content. The nodes that can be accepted are determined, job execution is distributed and requested, and job processing results are aggregated.

  When the requested job is executed, the server nodes 20s-1 to 20s-n operate as a server and execute the requested job depending on the processing capability and the intention to participate in response to an inquiry from the client node. Whether or not to accept is returned, and if accepted, the job is executed and the processing result is transmitted to the client node 20c.

  FIG. 2 is a diagram showing an example of service operation of the network system 1. In the figure, the node 11 requests a job to become a client node, and the job is distributed to the other nodes 12 to 15 through the grid service environment and processing is performed.

  On the nodes involved in the ad hoc grid service, the grid service environments 11b to 15b are operating, and the client application 21 and the grid job processing units 22-1 to 22-5 perform a cooperative operation via this environment.

  An ad hoc network is composed of a plurality of nodes with wireless communication functions, and there is a node that specializes in relaying traffic, which is communication content (except for gateway nodes that relay to the existing wired Internet). Each node has a function of terminating communication and a function of relaying communication contents of other nodes.

  In addition, since communication between each node is performed wirelessly, there is no visible communication cable, but in FIG. 2, the nodes that can communicate with each other are indicated by arrows in order to clearly indicate the relationship between the nodes. Shown tied.

  Each node is in an equal position, and a node having an ad hoc grid service environment can be a client that requests a job or a server that processes a job in response to a request depending on the situation. For this reason, the relationship between the client and the server is not fixed. When a job is requested, the node assumes the role of client (client role), and the node that receives the job request assumes the role of server (server role). .

Next, before describing the details of the configuration and operation, the overall characteristics of the network system 1 in which the grid service is realized on the ad hoc network will be described in (A1) to (A4).
(A1) On an ad hoc network, the network topology can change complicatedly, so it cannot be assumed that a specific node is always reachable on the network. In particular, it cannot be premised on the existence of a scheduler that performs server node authentication and job distribution among server nodes in an existing grid service environment such as Globus.

  For this reason, in the network system 1, each node constituting the ad hoc network has a scheduler function (one node has functions of both a client and a server).

  (A2) Since the performance of the nodes constituting the ad hoc network and the bandwidth of the network interface are extremely limited, and most of the nodes are battery-powered, unnecessary processing and communication are suppressed as much as possible. There is a need.

  Further, since an ad hoc network is a temporary network composed of portable nodes, it is often inefficient to perform authentication processing between nodes on the network. In addition, authentication processing may be overhead in both processing capacity and network bandwidth.

  For this reason, each node of the network system 1 is configured to perform communication between nodes by omitting authentication processing between nodes in order to save battery power of the own node or reduce overhead.

  In other words, on the ad hoc grid of the network system 1, it is easier to process from input data than an application with confidentiality and confidentiality that requires authentication between both the job processing request source and the request destination nodes. An application is effective in which the results are obtained and the results of processing the measurement results obtained from the sensors of the nodes are totaled utilizing the portability of each node.

  (A3) Since the node is portable and is battery-powered, the server node that requested the job processing may be unable to communicate in an unpredictable situation. For this reason, the network system 1 is configured to execute job processing to other nodes in anticipation of a case where the server node becomes unable to communicate.

  (A4) In an ad hoc grid service, it is difficult to think that an existing application in the HPC field or the like can satisfy performance on an ad hoc network. Therefore, an application suitable for an ad hoc network that does not necessarily require processing capability is more suitable for an application that is requested than a job that cannot be effective unless all the processing results of the requested jobs are obtained. If the processing result is obtained for the part, it is effective.

  Next, the configuration of the node 20 that constructs the network system 1 will be described. FIG. 3 is a diagram showing the configuration of the node 20. The node 20 includes a client application 21, a grid job processing unit 22, a job management unit 23, a resource management unit 24, a node state measurement unit 25, an ad hoc network protocol processing unit 26, and an IP processing / wireless communication interface unit 27.

  The client application 21 is an application that uses an ad hoc grid service, and is deployed on each node that provides the ad hoc grid service. When the ad hoc grid service is in operation, the node that activated the client application 21 plays a role as a client node.

  The grid job processing unit 22 is a part that processes a job received from a client node in an ad hoc grid service environment. The job management unit 23 includes a job management table 23a and a job history table 23b. Upon receiving a job request from the client application 21, the job management unit 23 distributes jobs and totals job processing results.

  The job management table 23a holds and manages information on jobs that are requested to be processed by other nodes and information on jobs that are requested by other nodes. The job history table 23b records and holds a history of job processing in the current node and a history of job requests to other nodes for a certain period for communication contents related to the ad hoc grid service with other nodes.

  The resource management unit 24 includes a resource management table holding unit 24a and a node configuration information holding unit 24b, and grasps resource information necessary for job execution. The resource includes information on other nodes that execute jobs.

  The resource management table holding unit 24a stores a correspondence relationship between a currently requested job and a job request destination node, a correspondence relationship between a requested job and a job request source node, and the like.

  The node configuration information holding unit 24b stores the static characteristics of the node. The static characteristics of the node include the calculation processing capability of the node, the maximum bandwidth of the wireless communication interface, the maximum capacity of the battery (maximum driving time when fully charged), and the like.

  The node state measurement unit 25 measures the dynamic state of the node and provides the node state to the resource management unit 24. The dynamic state of the node includes a calculation processing load of the node, a bandwidth occupation rate of the wireless communication interface, a remaining battery time, and the like.

  The ad hoc network protocol processing unit 26 is a function responsible for the construction and maintenance of an ad hoc network, and searches for a route according to the necessity of communication (search for a packet transfer destination node based on address information of a destination node), communication Pre-route information accumulation (neighboring node information is collected in advance and the correspondence between the destination node address of the packet and the node to which the packet is to be transferred next) The route information is updated when a movement or failure occurs.

  Note that the routing table that is referred to when controlling the relay of communication contents is dynamically formed by the ad hoc network protocol, so no prior configuration is required (data setting such as an address is not required).

  The IP processing / wireless communication interface unit 27 forms a lower layer of the ad hoc network protocol processing unit 26. That is, interface control for IP communication processing is performed based on wireless communication protocol processing.

  Next, the flow of processing when the node 20 becomes a client node will be described. FIG. 4 is a diagram showing a processing flow of the client node. The flow of processing between functional blocks when the node 20 operates as a client is shown (that is, the operation in the process of requesting processing of a job to another node and receiving the processing result).

[S1] A node having a job to be processed on the ad hoc grid service requests job processing from the client application 21 to the job management unit 23.
[S2] The job management unit 23 estimates the calculation amount of the job, derives the number of servers that request the job, and issues a resource search request to the resource management unit 24.

  [S3] The resource management unit 24 refers to the resource management table holding unit 24a, and extracts server information that is valid as a job request destination (a sufficiently new entry and has attributes necessary for the requested job processing). .

  [S3a] When the number of extracted servers is less than the number of servers requested by the job management unit 23, the resource management unit 24 searches for other nodes that can be servers through the IP processing / wireless communication interface unit 27.

[S4] The resource management unit 24 refers to the resource search result received from the IP processing / wireless communication interface unit 27, and determines whether the node search is successful.
[S5] If the node search is successful, the resource management unit 24 returns the obtained resource search result to the job management unit 23.

[S6] The job management unit 23 determines a job request destination server based on the obtained resource search result, and requests the job via the IP processing / wireless communication interface unit 27.
For example, as a resource search result, based on the job processing execution history executed in the past by the server node, the job processing result of the server node is determined, and a node that can accept the job processing is determined.

  Alternatively, based on the remaining battery time of the server node, a job processing shutdown due to battery exhaustion at the server node is predicted, and a node that can accept the job processing is determined.

[S7] The resource management unit 24 issues a resource monitoring request to the IP processing / wireless communication interface unit 27 until a job processing result is obtained after a job request.
[S8] The IP processing / wireless communication interface unit 27 returns a resource monitoring result for the node currently performing the job processing, and the resource management unit 24 appropriately checks the operation status of the job request destination server.

  [S9] During the job processing request, the topology of the ad hoc network may fluctuate due to the movement of the node, the change of the propagation state of the electromagnetic wave, etc., until the processing result is obtained. When the route information managed by the ad hoc network protocol processing unit 26 changes as a result of such a change on the network, the ad hoc network protocol processing unit 26 notifies the resource management unit 24 of the topology change.

[S10] When a job processing result is returned from the server node, the job management unit 23 updates the job management table 23a based on the processing result and totals the processing results.
[S11] The job management unit 23 registers the job processing result of the job request destination server in the resource management unit 24.

  [S12] When the processing results of all jobs are obtained, the job management unit 23 returns the processing results to the client application 21 and completes the distributed cooperative processing of the ad hoc grid service.

  Next, the flow of processing when the node 20 becomes a server node will be described. FIG. 5 is a diagram showing a processing flow of the server node. A flow of processing between functional blocks when the node 20 operates as a server is shown (that is, an operation in a process of processing a requested job and transmitting a processing result).

[S21] The job management unit 23 becomes a server node by receiving the job request message via the IP processing / wireless communication interface unit 27.
[S22] When the job management unit 23 receives the job request message, the job management unit 23 willingness (participation) becomes an index as to whether or not to accept the job from the job execution history, the node configuration information, and the node state obtained from the node state measurement unit 25. Intention) information is dynamically derived (as a past job execution history, for example, the intention to participate is dynamically determined based on at least one of the calculation processing load of the node, the bandwidth occupancy, and the remaining battery time), and this value Based on the above, it is determined whether or not the job can be accepted, and the inquiry is answered (reply of a job request response message).

[S23] The job management unit 23 receives a job processing request from the client.
[S24] The job management unit 23 determines whether or not the job can be finally received. If the job management unit 23 determines that the job is received, the job management unit 23 returns a response indicating that the job is received to the client node (reply of job reception notification).

  [S25] If the job management unit 23 determines that the requested job is to be processed by its own node, the job management unit 23 records the job content in the job management table 23a and passes the job processing request to the grid job processing unit 22.

[S26] When the job processing is completed, the grid job processing unit 22 returns the processing result to the job management unit 23.
[S27] The job management unit 23 deletes the corresponding entry from the job management table 23a, and records the processing result in the job history table 23b. Thereafter, the job management unit 23 returns the job processing result to the client node via the IP processing / wireless communication interface unit 27.

Next, a detailed operation of the job management unit 23 that is one of the main components of the node 20 will be described. First, the operation during client roll will be described with reference to FIGS.
FIG. 6 is a diagram showing an operation flow of the job management unit 23 during the client roll. FIG. 7 shows the structure of the job management table 23a. The job management table 23a includes a requested job management table 23a-1 and a requested job management table 23a-2. The requested job management table 23a-1 is used during the client roll, and the requested job management table 23a-2 is used during the server roll.

[S30] The client application 21 transmits a job processing request to the job management unit 23.
[S31] Upon receiving a job processing request, the job management unit 23 issues a resource search request to the resource management unit 24. At this time, a condition relating to the processing capability of the server as a resource and a resource search range are designated.

[S 32] The job management unit 23 receives a resource search result from the resource management unit 24.
[S32a (FIG. 7)] The job management unit 23 adds the node group information obtained as a search result to the requested job management table 23a-1. At this time, the column of the original job identifier k1 and the column of the job request source k2 are set to blank.

[S33] The job management unit 23 compares the obtained number of server nodes with the number (n) of server nodes required by the job.
[S34] If the obtained number of server nodes is smaller than the number of server nodes required by the job, the resource search range and necessary server conditions are relaxed, and a resource search request is issued to the resource management unit 24 again.

  [S35] When the obtained number of server nodes is larger than the number of server nodes required by the job, after the server search is completed, the job management unit 23 passes through the IP processing / wireless communication interface unit 27 to start the job management table 23a. The job processing request is transmitted to the n servers.

  [S35a (FIG. 7)] For the server node that sent the job processing request, the request destination status k3 is set to “requesting” (note that the job management that has received the job processing result through the IP processing / wireless communication interface unit 27) The unit 23 updates the request destination status k3 of the request job management table 23a-1 to complete).

[S36] The resource management unit 24 notifies the job management unit 23 when the server requesting the processing becomes unmonitorable or when the job processing progress is slow.
[S37] Upon receiving the notification, the job management unit 23 sets the request destination status k3 corresponding to the server to offline, and selects a node whose request destination status k3 is neither requested nor offline from the requested job management table 23a-1. , Retry the job request with no chance of recovery to the selected node.

[S38] The job management unit 23 receives the job processing result.
[S39] If reception of all job processing results is not completed, the process returns to step S36, and if completed, the process proceeds to step S39a.

[S39a] The job management unit 23 returns the job processing result to the client application 21, and then adds an entry corresponding to the job to the job history table 23b.
Next, the operation of the job management unit 23 during server roll will be described with reference to FIGS. FIG. 8 is a diagram showing an operation flow of the job management unit 23 during server roll. FIG. 9 shows the structure of the job history table 23b. The job history table 23b includes a job processing history table 23b-1 and a job request history table 23b-2.

[S41] A job server request message from the client is received by the job management unit 23 via the IP processing / wireless communication interface unit 27.
[S42a] The job management unit 23 queries the resource management unit 24 about the node configuration and node status.

[S42b] The job management unit 23 receives the inquired node configuration and node state from the resource management unit 24.
[S 43] The job management unit 23 returns a job server request response message containing the obtained result via the IP processing / wireless communication interface unit 27.

[S44] The job management request message from the client is received by the job management unit 23 via the IP processing / wireless communication interface unit 27.
[S45] The job management unit 23 analyzes the contents of the job processing request message. If the job management request message is to be processed within the own node, the job management unit 23 proceeds to step S46a.

  [S46a] When processing is performed in the own node, the job management unit 23 adds the corresponding job information to the requested job management table 23a-2 in FIG. At this time, the job processing category k4 is updated to owned. After adding to the table, a job processing request is sent to the grid job processing unit 22.

[S46b] The grid job processing unit 22 processes the job and transmits the processing result to the job management unit 23.
[S46c] The job management unit 23 deletes the corresponding entry in the requested job management table 23a-2, and then returns the processing result to the client node via the IP processing / wireless communication interface unit 27.

[S46d (FIG. 9)] The job management unit 23 adds an entry to the job processing history table 23b-1.
[S47a] When delegating the processing to another node, after adding the entry of the corresponding job to the requested job management table 23a-2, the client node operates according to the operation flow of the job management unit 23 at the time of the client node.

  [S47b (FIG. 7)] However, at this time, the job processing classification of the corresponding entry in the requested job management table 23a-2 is updated to “passed”, and a request is sent to the original job identifier k1 in the requested job management table 23a-1. The job identifier of the received job is set (001 is set in the figure), and the address of the request source node is set to the job request source k2 (addr a is set in the figure). Subsequent processing is in accordance with processing at the time of client roll. After the processing is completed, an entry is added to the job processing history table 23b-1 and the job request history table 23b-2.

Next, the detailed operation of the resource management unit 24, which is one of the main components of the node 20, will be described. The operation of the resource management unit 24 includes the following operations (A) to (D).
(A) Normal operation of the resource management unit 24 In the normal operation of the resource management unit 24 of the node, the routing table information in the ad hoc network protocol processing unit 26 is periodically acquired. FIG. 10 is a diagram showing a route table that the ad hoc network protocol processing unit 26 has. The figure shows a route table (tn) at a certain timing tn and a route table (tn-1) at the previous timing (tn-1).

  FIG. 11 is a diagram showing the configuration of the resource management table holding unit 24a. The resource management table holding unit 24a includes a resource management table 24a-1 and a resource management history table 24a-2. The resource management unit 24 takes a difference between the route tables continuous in time series, and updates the resource management table 24a-1 when a change occurs.

  [S50a (FIG. 10)] When the next hop address changes for an arbitrary destination address, an entry for the corresponding address is newly created in the resource management history table 24a-2, and the change time and change contents (neighbor changed) are displayed. Set.

  [S50b (FIG. 10)] Even when the number of hops changes, an entry of the corresponding address is newly created in the resource management history table 24a-2, and the change time and the change content (distance changed) are set. The same applies to the destination address newly appearing on the routing table and the destination address deleted from the routing table.

(B) Operation of Resource Management Unit 24 at Server Search FIG. 12 is a diagram showing an operation flow when the resource management unit 24 searches. The operation when searching for a server when a node operates as a client role is shown.

[S51] The resource manager 24 receives a resource search request from the job manager 23.
[S52] The resource management unit 24 generates a job server request message. At this time, if the job server request message may contain the job processing history of the own node as an option, the resource management unit 24 may generate a resource before the job server request message is generated. The contents of the management table 24a-1 are extracted.

[S53] The resource management unit 24 outputs the generated job server request message via the IP processing / wireless communication interface unit 27.
[S54] The resource management unit 24 receives the job server request response message from the node via the IP processing / wireless communication interface unit 27, and the latest time of the entry corresponding to the address of the corresponding server node in the resource management table 24a-1. The stamp e1 (FIG. 11) is updated.

  [S55] The elapsed time of resource search is measured. If the job server request response message is received within the set time of the timer, the process proceeds to step S56. Otherwise, the server waits for a response from the server.

[S56] The resource search result is returned to the job management unit 23.
(C) Operation of Resource Management Unit 24 at Node Monitoring FIG. 13 is a diagram showing an operation flow of the resource management unit 24 at the time of node monitoring. The operation of the resource management unit 24 at the time of node monitoring when a node requests a job and operates as a client role, or when receiving a job request and operates as a server role is shown.

  [S61] When the node leaves the ad hoc network, joins the node to the ad hoc network, or changes in the topology of the ad hoc network, the ad hoc network protocol processing unit 26 operates to derive a route. At this time, the topology change information is sent from the ad hoc network protocol processing unit 26 to the resource management unit 24.

  [S61a (FIG. 11)] Upon receiving the topology change notification, the resource management unit 24 displays the address, change time, and change contents of the node affected by the topology change in the resource management table 24a-1. -2.

[S62] It is determined whether or not the node has become unreachable in terms of network due to a topology change.
[S63] For a node that has become unreachable, if the corresponding node exists in the resource management table 24a-1, the resource response reception waiting timer is started, and then the corresponding node is set via the IP processing / wireless communication interface unit 27. Send resource monitoring messages to the server.

[S64] When the resource management unit 24 receives the resource response message, the resource management unit 24 updates the latest time stamp of the corresponding address in the resource management table 24a-1.
[S65] It is determined whether the resource response reception waiting timer has expired.

  [S66] If the resource response message cannot be received before the expiration of the resource response reception waiting timer, the corresponding entry in the resource management table 24a-1 is deleted and the job management unit 23 is notified of the node monitoring failure.

(D) Operation of Resource Management Unit 24 During Server Search Response Processing FIG. 14 is a diagram showing an operation flow of the resource management unit 24 during server search response processing. The operation of the resource management unit 24 in response processing to the client node during server search is shown.

[S71] A job server request message is received from the client node via the IP processing / wireless communication interface unit 27.
[S72] Node configuration information is acquired.

[S73] The node state measurement unit 25 is inquired about the node state.
[S74] The measurement result from the node state measurement unit 25 is received.
[S75] It is determined whether or not the job server request message is a response request in a willingness format (a format for requesting participation in whether to process the job). If the response request is in the willingness format, go to step S76, otherwise go to step S77.

[S76] If a response in the willingness format is requested, the willingness is dynamically derived from the collected information.
[S77] A job server request response message is constructed from the collected and derived information.

[S78] A job server request message is returned to the source client node via the IP processing / wireless communication interface unit 27.
Next, job processing is shared and interchanged between the nodes constituting the ad hoc network, and the operation of the network system 1 when the calculation process is performed as if the entire ad hoc network is a single high-performance computer is described in detail. To do. First, details of the message will be described in the following (1), (1-1) to (1-4).

(1) Transmission of Protocol Message FIG. 15 is a diagram showing the configuration of an ad hoc network. The nodes 51 to 58 can move by performing wireless communication, and arrows a1 to a7 in the drawing represent a state in which electromagnetic waves used for wireless communication are reachable and can be directly communicated.

  Ad hoc grid protocol messages for controlling the ad hoc grid service are exchanged on the ad hoc network. Each node operates an ad hoc network protocol, and exchanges protocol messages to generate routing information, which is information necessary for message routing, and holds the corresponding information in the node.

  In the routing information, when a received message is transferred, a routing table for determining whether or not the node should transfer the message from the message transfer source and the message destination, and when a broadcast message is received. Information for determining whether or not the own node should transfer the message.

  As the types of ad hoc grid protocol messages, there are two types of unicast messages and broadcast messages as in the existing network protocol.

(1-1) Transmission of Unicast Message Consider a case in which a unicast message having a node 51 as a transmission source and a node 57 as a destination is transmitted. The transfer process of the unicast message is processed on the existing ad hoc network protocol. The unicast message transmitted from the node 51 is first received by an adjacent node 56 (an electromagnetic wave used for communication is directly reachable) (arrow y1).

  When the node 56 searches the route table for controlling the ad hoc network and recognizes that its own node exists on the route for transferring the message received from the node 51 direction to the destination node 57, the node 56 transfers the message to the node 57 ( Arrow y2).

  When the node 57 confirms that the destination of the message is its own node, the message is processed by the application. At this time, the nodes (nodes 52 and 54) that do not have their own node on the path for transferring the message received from the node 51 to the destination node 57 do not transfer even if the message is received, and particularly notify an error. Discard the message silently.

  Of the ad hoc grid protocol, a job request message, a node monitoring message, a node monitoring response message, and a job processing result message are accommodated in the unicast message.

(1-2) Transmission of Broadcast Message FIG. 16 is a diagram illustrating how a broadcast message is transmitted. A state is shown in which a broadcast message with a node 51 as the transmission source and a message reach range of 2 is transmitted. Broadcast message transfer processing is performed on an existing ad hoc network protocol. The broadcast message transmitted from the node 51 is first received by the adjacent nodes 52, 54, and 56 (arrows b1, b3, and b5).

  When a message is transmitted, a non-zero integer value is accommodated in the packet lifetime field in the message header. The value of the packet lifetime field represents the maximum number of messages that can be transferred (ie, the reach of broadcast messages).

  Each node 52, 54, 56 that has received the message searches a routing table that controls the ad hoc network, and determines whether or not the node is responsible for broadcast forwarding when broadcasting the message received from the node 51 direction. Refer to the information. If the own node is a node that transmits a broadcast message, the message tries to be transferred (arrows b2, b4, b6, b7).

  The node decrements by 1 when the packet lifetime field included in the header of the message is not 0 at the time of transfer. If the packet lifetime field is 0, the corresponding message is not transferred, and the message is silently discarded without any error notification (silently).

  In the case of the figure, since the nodes 53, 55, 57, and 58 receive the broadcast packet whose packet lifetime field is 0, these nodes do not forward the corresponding broadcast packet any more. In addition, a job server request message in the ad hoc grid protocol is accommodated in the broadcast message.

(1-3) Message Format FIG. 17 is a diagram showing a format of protocol messages exchanged on the ad hoc grid service. As described in (1-1) and (1-2), there are two types of protocol messages: unicast message U1 and broadcast message B1. The unicast message U1 includes a unicast message header U1a and a message body m10.

  The broadcast message B1 includes a broadcast message header B1a and a message body m10. The message body m10 used in the ad hoc grid message has a common format and includes an ad hoc grid message header m11 and an ad hoc grid message body m12.

  The ad hoc grid message header m11 includes a message generation node address m11a for specifying the node that generated the message, a message destination node address m11b for specifying the destination node of the message, a protocol version m11c indicating the version of the ad hoc grid protocol used for message exchange, It consists of a message type m11d indicating the type of message accommodated in the ad hoc grid message body, and a message length m11e indicating the size of the message.

  On the other hand, the ad hoc grid message body m12 includes an arbitrary number of concatenations of message fields m12d, which are attributes constituting the message. Each message field includes a field type m12a indicating the type of attribute accommodated, a field length m12b indicating the size of the message field, and a field body m12c indicating the attribute value.

(1-4) Message Body Format FIGS. 18 and 19 are diagrams showing details of the message body m10. The contents of the message body m10 of the protocol message exchanged on the ad hoc grid service are shown.

  The message type is represented by the message type m11d of the ad hoc grid message header m11. The message includes a job server request message 101, a job server request response message 102, a job processing request message 103, a job processing result message 104, a node monitoring message 105, There are types such as the node monitoring response message 106.

  The job server request message 101 includes a job ID 101a given by the job requester for identifying the job, a job requester ID 101b for identifying the job requester, a job field 101c schematically indicating the job type, a job A job type 101d indicating the type of job in detail, a job unit 101e indicating the load of job processing, a required device 101f indicating a device necessary for executing the job, such as a sensor or a secondary storage buffer, and the requested job on another node Delegation / distribution availability 101g indicating whether or not delegation / distribution is possible, allowable response time 101h indicating the maximum turnaround time from when a job processing request is transmitted until the result is received, the job that the node has processed so far It consists of processing history 101i (option) and the like.

  The job server request response message 102 includes the job ID 102a in the corresponding job server request message, the server calculation capability 102b indicating the processing capability of the processing server, the possessed device 102c indicating the sensor held by the node, and the own node wants to take over the processing. A process willingness 102d indicating the desired degree, a guaranteed response time 102e indicating a length of time that can be guaranteed as a process delay from receiving the process request to transmitting the process result, network mobility of the own node (the own node from the ad hoc network) Mobility 102f indicating that the value is high when the frequency is high, and the job processing history 102g (option) processed by the node so far.

  The job processing request message 103 includes an original job ID 103a assigned by the original job requester when the job server request message 101 is transmitted, a job unit ID 103b assigned to each job subdivided at the time of request, and a job required for individual job processing. It consists of parameters 103c, job data 103d, and the like.

  The job processing result message 104 includes a job ID 104a and a job unit ID 104b in the corresponding job processing request message 103, a job processing history 104c for storing the history when job processing is delegated or distributed to another node, and a job processing result 104d. Etc.

  The node monitoring message 105 includes a job ID 105a and a job unit ID 105b in the corresponding job processing request message 103, a monitoring message ID 105c assigned by the transmission source of the corresponding node monitoring message to identify the message, and the like.

  The node monitoring response message 106 includes a job ID 106a, a job unit ID 106b, a monitoring message ID 106c in the corresponding node monitoring message, a job progress 106d indicating the progress status of the job, and an error message for reporting an error or failure occurring during job processing. 106e (only when an error has occurred).

Next, details of the job server request message 101 will be described in the following (2), (2-1), and (2-2).
(2) Job server request message 101
FIG. 20 is a diagram illustrating how the job server request message 101 is transmitted. A state of collecting server node information for starting an ad hoc grid service is shown. The client node 51 transmits a job server request message 101 as a broadcast message in order to recruit nodes that can be used as servers (arrows c1 to c3).

  The broadcast job server request message 101 is received by all adjacent nodes (which can reach the electromagnetic wave) that can communicate directly. When the node receiving the message confirms that the corresponding message is a broadcast message and the value of the packet lifetime field set in the message is non-zero, the node decrements the packet lifetime in the message by 1 and then adjoins. Transfer to node (arrows c4 to c7). In the case of the figure, since the initial value of the packet packet life is 2, the transfer is performed only once, and is not transferred to a node farther than that (arrow c8).

  FIG. 21 is a diagram showing how the job server request response message 102 is transmitted and received. The job server request response message is transmitted from each server node toward the client message (arrows d1 to d7).

(2-1) Transmission and Response of Job Server Request Message 101 FIG. 22 is a diagram showing how job requests and responses are transmitted and received between a client node and a server node.

[S81] The client node transmits a job server request message 101 by broadcast to the server node group.
[S82] Each server node that receives the job server request message 101 transmitted from the client node measures the node state and obtains node configuration information.

  [S83] The server node accommodates and responds to the job server request response message 102 with its own node configuration, its own node status, its own network mobility, and its own job processing history information.

  [S84] The client node that has received the job server request response message 102 collates with the configuration and status required by the ad hoc grid application, and determines whether or not the client node is the job request target.

[S85] It is determined whether the network mobility condition is met.
[S86] It is determined whether the job processing history condition is met.
[S87] If the conditions in steps S84 to S86 are met, the node is added to the resource management table 24a-1.

  [S88] The job server request response message waiting time is measured, and if the message waiting time has expired, the job server request processing is terminated. If it has not expired, the process returns to step S83 to enter the job server request response message standby state again.

(2-2) Alternative means of the procedure of (2-1) FIG. 23 is a diagram showing how job requests and responses are transmitted and received between a client node and a server node. 23 shows an alternative process of the process shown in FIG.

  [S91] The client node transmits a job server request message 101 by broadcast to the server node group. The job server request message 101 includes processing capacity necessary for job execution, functions necessary for job execution, network mobility of the client node, job processing history of the client node, job ID, necessary node configuration, and the like.

  [S92] Upon receiving the job server request message 101, the server node collects the configuration, status, function, network mobility of the client node, and job execution history of the client node.

[S93] The server node derives willingness representing the intention of executing the job.
[S94] The willingness is accommodated in the job server request response message 102, and a response is made to the job client node.

  [S95] The client node that has received the job server request response message 102 determines whether or not the willingness in the message is greater than a minimum threshold value indicating whether or not the job request is to be made. If so, go to step S96, otherwise go to step S97.

[S96] The node is added to the resource management table 24a-1.
[S97] A job server request response message waiting time is derived, and when the message waiting time expires, the job server request processing is terminated. If it has not expired, the process returns to step S94 to enter the job server request response message standby state again.

(3) Job Request from Client FIG. 24 is a diagram showing a job request from the client. The job request message is transmitted from the job client to the server nodes 52 and 56 by unicast (arrows f1 and f2). When the server node that has received the job request message is a node that relays the broadcast message in accordance with the ad hoc network control protocol (node 56), the processing capability of its own adjacent node according to the instruction in the job server request message, other A job server request response message can be constructed based on the information.

  The server that has received the job request message from the client can request all or part of the job to the adjacent nodes 57 and 58 other than the adjacent node that has received the message (arrows f3 and f4). There can be a plurality of adjacent nodes to be requested for a job.

(4) Server Progress Monitoring FIG. 25 is a diagram showing the state of monitoring processing during job execution by the server. After the job request from the client node to the server node, the server is executing the job until the job processing result is returned from the server to the client.

  The client node checks the time series change of the route table of the ad hoc network grasped by the ad hoc network protocol processing unit 26 in order to grasp the state of the job being executed. When a change occurs in the routing table, it is determined whether the server node that is executing the job is included in the change, and if included, a monitoring request message is transmitted to the corresponding server node.

  In the figure, a client node 51 that has detected a server node 52 that is no longer adjacent to the client node as a result of movement during job processing transmits a monitoring request message to the corresponding server node. The monitoring request message transmission is retried several times. In the figure, a new route p1 is formed by another node 50a that has moved, and a monitoring request message is transferred on the corresponding route.

  The server node that has received the monitoring request message responds with a monitoring response message containing its own job processing status. The job client node that has received the monitoring response message checks the job processing status in the message. If the job processing status is significantly less than the expected job processing amount expected from the product of the processing capacity requested by the server and the elapsed time from the request, or if the monitoring response message could not be received by the time out, job recovery Process.

(5) Job recovery processing In the server progress monitoring process of (4), if the server is confirmed to leave the ad hoc grid service environment, or if the progress of job processing is significantly behind expectations, or the network environment If the requested job cannot be executed due to a change, the job is distributed to other server nodes as an exceptional situation, and recovery processing is performed.

  FIG. 26 is a diagram showing the state of the recovery process by the server. In the job recovery process, the job client node searches the resource management table 24a-1 created at the job server request stage, and transmits a job request message to an empty server that has not requested a job (arrow g1). ).

  If there is no free server, and the entry in the resource management table 24a-1 is invalid due to elapse of time, the job server request process is executed again. At this time, a new server node search is performed by taking measures such as expanding the broadcast range.

  In the figure, since the job server request message is transmitted by increasing the number of times that the message can be transferred, the job server request message arrives farther than the message reachable range (arrow c8) in FIG. 20 (arrow g2).

(6) Job processing result buffering and aggregation processing When the job processing server completes the job processing and returns the processing result to the client node, the network connectivity between the client node and the job processing server is lost. In this case, the job processing server buffers the job processing result for a certain time.

  When the network connectivity is restored (for example, when the loss of network connectivity is due to accidental non-arrival of electromagnetic waves), or the ad hoc network protocol processing unit 26 successfully searches for an alternative route with the job client. If so, the buffered job processing result is transmitted to the job client.

  The server node that has received the job request message from the server node instead of the client returns the job processing result to the client node that is the transmission source of the job request message after the job processing is completed. The client node that has received the job processing result totals the job processing results received from other server nodes and the job processing results of its own node, and responds to the job request source.

  FIG. 27 is a diagram illustrating a state where the job processing server completes the job processing and returns the processing result to the client node. When the job ends, the server nodes 57 and 58 transmit a job processing result message to the server node 56 that has received the job request (arrows h1 and h2). Upon receiving the job processing result message, the server node 56 integrates the job processing result message and transmits it to the job client node (arrow h3).

  As described above, the network system 1 does not require a scheduler and assembles nodes that can request a job in an ad hoc manner according to a situation that requires the use of an ad hoc grid service, and requests the job. did.

  This makes it possible to efficiently build a grid service environment on the ad hoc network from the viewpoint of processing capacity, power consumption, network bandwidth, etc., and share a large number of computing resources existing on the ad hoc network to create virtually powerful calculations. Ability will be available.

  In addition, by utilizing the characteristics of the ad hoc network of the network system 1, it can be used as a wide variety of data sources reflecting sensors covering a wide area and the users and locations of nodes.

  Furthermore, by realizing an efficient grid service on the ad hoc network, measurement information is collected on the sensor network configured by the ad hoc network, and at the same time, the measurement results are distributed and processed on the ad hoc network configured in the event of a disaster. This makes it possible to execute unprecedented applications such as evacuation route simulation and marketing research using mobile devices in urban areas.

(Supplementary note 1) In a network system that performs communication on a wireless network,
When requesting job execution, it becomes a client operation, inquires about the processing capability and intention to participate in other node groups, determines the node that can accept job processing based on the response content, A client node that distributes and requests execution and aggregates job processing results;
When the requested job is executed, it becomes a server operation, and responds to the inquiry from the client node by accepting whether or not to accept the requested job depending on the processing capability and willingness to participate. In this case, a server node that executes a job and transmits a processing result to the client node;
A network system characterized in that distributed cooperative processing is performed on an ad hoc network constructed by a plurality of nodes that can move by wireless communication without using a fixed station.

  (Additional remark 2) The said server node hold | maintains the execution history of the job processing performed in the past, responds with the said execution history with respect to the inquiry from the said client node, and the said client node is based on the said execution history. The network system according to claim 1, wherein a job processing result of the server node is determined to determine a node that can accept the job processing.

  (Additional remark 3) The said server node measures the battery remaining time of a self-node, responds to the said battery remaining time with respect to the inquiry from the said client node, The said client node is based on the said battery remaining time, The network system according to claim 1, wherein a node that can accept job processing is determined by predicting shutdown of job processing due to battery exhaustion at the server node.

  (Appendix 4) The server node dynamically obtains an intention to participate in job processing based on the past job execution history, and responds to the job execution request from the client node whether it can accept job processing. The network system according to appendix 1, wherein:

  (Additional remark 5) The said client node and the said server node detect the change of the network topology which arises by joining / separation of the said ad hoc network of another node, derivation | leading-out a path | route, and confirming the network connection condition of a node The network system according to supplementary note 1, which is characterized.

  (Supplementary Note 6) When the server node receives the requested job from the client node and cannot execute the requested job, all or a part of the requested job is transferred to another The network system according to appendix 1, wherein distribution and transfer to nodes are performed.

  (Supplementary note 7) When network connectivity between the client node and the server node is lost, job processing results from a plurality of nodes are aggregated after being received by a node other than the client node of the job request source. The network system according to appendix 1, wherein buffering is performed for a certain period of time, and processing results are transmitted to the original client node when network connectivity is restored within a certain period of time.

  (Supplementary Note 8) The client node and the server node perform communication between nodes by omitting an authentication process between the nodes in order to save battery power of the own node and reduce overhead. The network system according to attachment 1.

  (Supplementary note 9) The network system according to supplementary note 1, wherein the client node has a sensor function, requests the server node to process measurement information measured by the sensor, and totals the processing results.

(Supplementary Note 10) In a mobile communication node that performs communication on a wireless network,
A grid job processing unit for processing a job requested by a client;
When requesting job execution, it becomes a client operation, inquires about the processing capability and intention to participate in other node groups, determines the node that can accept job processing based on the response content, When the requested job is executed by distributing and executing the execution and the requested job is executed, it becomes a server operation, and in response to the inquiry from the client, the requested job is processed according to the processing capability and the intention to participate. A job management unit that responds whether or not to accept execution;
A resource management unit that manages node resources necessary for job execution and dynamically requests participation in job processing based on past job execution history;
A mobile communication node characterized in that distributed cooperative processing is performed on an ad hoc network constructed by a plurality of nodes that can be moved by wireless communication without using a fixed station.

  (Additional remark 11) The said job management part hold | maintains the execution log | history of the job process performed in the past at the time of a server operation | movement, respond | corresponds with an execution log | history with respect to the inquiry from a client, and the said execution log | history at the time of a client operation | movement 11. The mobile communication node according to supplementary note 10, wherein a node that can accept the job processing is determined based on the determination of the job processing performance of the server.

  (Additional remark 12) The said job management part measures the battery remaining time of a self-node at the time of a server operation | movement, responds to the said battery remaining time with respect to the inquiry from a client. 11. The mobile communication node according to supplementary note 10, wherein a node that can accept job processing is determined by predicting shutdown of job processing due to running out of battery in the server.

  (Additional remark 13) It further has the ad hoc network protocol processing part which detects the change of the network topology which arises by joining / separation of the said ad hoc network of another node, derives | leads-out a path | route, and confirms the network connection condition of a node The mobile communication node according to appendix 10.

  (Supplementary Note 14) When the job management unit receives a requested job from a client during server operation and cannot execute the requested job, the job management unit performs all or part of the requested job. The mobile communication node according to appendix 10, wherein the mobile communication node is distributed / transferred to another node.

  (Supplementary Note 15) When the network connectivity between the nodes is lost, the job management unit aggregates job processing results from a plurality of nodes after being received by a node other than the client that is the job request source, for a certain period of time. 11. The mobile communication node according to appendix 10, wherein the mobile communication node performs control to transmit a processing result to an original client when buffering is performed and network connectivity is restored within a predetermined time.

  (Supplementary Note 16) The supplementary note 10 is characterized in that the job management unit performs communication between nodes in which authentication processing between the nodes is omitted in order to save battery power of the node and reduce overhead. The mobile communication node described.

  (Supplementary note 17) The mobile communication node according to supplementary note 10, further comprising a sensor function, wherein the job management unit requests the server node to process measurement information measured by the sensor, and totals the processing results. .

  (Supplementary Note 18) Based on the information generated and held on the node by the ad hoc network protocol on the client, the change of the connection status to the network of the own node is judged, and to which node the client requests a job Mobile communication control method to be determined.

  (Supplementary note 19) Mobile communication control for determining a change in the connection status of a client to a network based on information generated and held by a routing protocol on the node and determining whether the node can accept a job from the client Method.

1 is a principle diagram of a network system. It is a figure which shows the service operation example of a network system. It is a figure which shows the structure of a node. It is a figure which shows the flow of a process of a client node. It is a figure which shows the flow of a process of a server node. It is a figure which shows the operation | movement flow of the job management part at the time of a client roll. It is a figure which shows the structure of a job management table. It is a figure which shows the operation | movement flow of the job management part at the time of a server roll. It is a figure which shows the structure of a job history table. It is a figure which shows the routing table which an ad hoc network protocol process part has. It is a figure which shows the structure of a resource management table holding part. It is a figure which shows the operation | movement flow at the time of the search of a resource management part. It is a figure which shows the operation | movement flow at the time of the node monitoring of a resource management part. It is a figure which shows the operation | movement flow at the time of the server search response process of a resource management part. It is a figure which shows the structure of an ad hoc network. It is a figure which shows the mode of transmission of a broadcast message. It is a figure which shows the format of the protocol message exchanged on an ad hoc grid service. It is a figure which shows the detail of a message body. It is a figure which shows the detail of a message body. It is a figure which shows the mode of transmission of the job server request message. It is a figure which shows the mode of transmission / reception of a job server request response message. It is a figure which shows the mode of transmission / reception of the job request / response between a client node and a server node. It is a figure which shows the mode of transmission / reception of the job request / response between a client node and a server node. It is a figure which shows the mode of the job request from a client. It is a figure which shows the mode of the monitoring process during the job execution by a server. It is a figure which shows the mode of the recovery process by a server. It is a figure which shows a mode that a job processing server completes job processing and returns a processing result to a client node.

Explanation of symbols

1 network system 1a ad hoc network 20c client node 20s-1 to 20s-n server node

Claims (5)

In a network system that performs communication on a wireless network,
When requesting job execution, it becomes a client operation, inquires about the processing capability and intention to participate in other node groups, determines the node that can accept job processing based on the response content, A client node that distributes and requests execution and aggregates job processing results;
When the requested job is executed, it becomes a server operation. In response to the inquiry from the client node, whether or not the execution of the requested job is accepted depending on the processing capability and the intention to participate is accepted. In this case, a server node that executes a job and transmits a processing result to the client node;
A network system characterized in that distributed cooperative processing is performed on an ad hoc network constructed by a plurality of nodes that can move by wireless communication without using a fixed station.   The server node holds an execution history of job processing executed in the past, responds to the execution history in response to an inquiry from the client node, and the client node is configured to return the server node based on the execution history. 2. The network system according to claim 1, wherein a node that can accept the job processing is determined by judging a result of the job processing.   The server node dynamically obtains an intention to participate in job processing based on a past job execution history, and responds to the job execution request from the client node whether the job processing can be accepted. The network system according to claim 1.   When network connectivity between the client node and the server node is lost, job processing results from a plurality of nodes are aggregated after being received by a node other than the client node of the job request source and buffered for a certain period of time. The network system according to claim 1, wherein when the network connectivity is recovered within a predetermined time, the processing result is transmitted to the original client node. In a mobile communication node that performs communication on a wireless network,
A grid job processing unit for processing a job requested by a client;
When requesting job execution, it becomes a client operation, inquires about the processing capability and intention to participate in other node groups, determines the node that can accept job processing based on the response content, When the requested job is executed by distributing and executing the execution and the requested job is executed, it becomes a server operation, and in response to the inquiry from the client, the requested job depends on the processing capability and the intention to participate. A job management unit that responds whether or not to accept execution;
A resource management unit that manages node resources necessary for job execution and dynamically requests participation in job processing based on past job execution history;
A mobile communication node characterized in that distributed cooperative processing is performed on an ad hoc network constructed by a plurality of nodes that can be moved by wireless communication without using a fixed station.

Images