Patent Application
20090216510
The present invention relates to a mobile node simulator and a program for implementing the mobile node simulator.
Along with the recent ubiquitousness of computer system and the popularization of radio technology, a research on a mobile network has been extensively conducted in which mobile communication terminals (hereinafter, mobile terminals) or mobile nodes communicate to one another via a radio LAN or the like. In designing and implementing a mobile network application and protocol, it is inevitable to conduct a network simulation by a computer. However, in conventional network simulators, the movement of a mobile terminal is often limited to a completely random movement and a movement based on a random destination selection. For example, ns−2 (see Reference 1) merely provides an independent program which outputs a trace of a behavior in accordance with random mobility and random waypoint mobility (see Reference 2). A simulator QualNet (see Reference 3) for business purpose provides a behavior only based on group mobility, random way point mobility and trace mobility. Even GloMoSim (see Reference 4) at most supports, in addition thereto, random drunken mobility, ECRV mobility, reference point group mobility and the like. In Wireless Module provided as an expansion module in a simulator OPNET (see Reference 5) for business purpose, the movement of a mobile node on a three-dimensional ground or satellite can be defined by a user.
However, a simulation in a state in which a mobile node changes its behavior due to information obtained from a network application and environmental information around the mobile node itself (visual information and the like) is not considered. In an actual application, the behavior of a person carrying a mobile terminal depends on the information obtained from the application, visual information, mental state or the like. For example, it is assumed that visitors to an event site carry mobile terminals. In this case, a performance evaluation of an application distributing navigation information via a multi-pop ad hoc network between the mobile terminals is considered. If the mobile terminals gather at the event site following the navigation information (e.g., time-limited event information) of the application, there is a possibility that the network topology changes along therewith, resulting in a network with non-uniform density of the mobile terminals. As a result, in an application designed assuming a random movement having the density of mobile terminals being approximately uniform, there is a possibility of a problem occurring that the information arrival rate to a place where the density of mobile terminals is low is decreased, and on the other hand, unnecessary packets appear at a place where the density of mobile terminals is high. However, it is not easy for existing simulators to find such a problem.
Further, the conventional mobile node simulators specialize in a specific movement for a specific mobile node. Thus, in newly developing a simulator having another behavior model, it is necessary to construct the simulator from the beginning for each development, and an enormous amount of cost and time are required.
One of the objectives of the present invention is to provide a mobile node simulator including: a behavior simulator section for simulating behaviors of a plurality of mobile nodes in accordance with a behavior model definable by a user; and a network simulator section for simulating a network topology, including a plurality of mobile nodes, and a plurality of communications on the network, wherein the behavior simulator section and the network simulator section interactively operate with each other; and a program for implementing the mobile node simulator.
[Reference 1] http://www.isi.edu/nsnam/
[Reference 2] Tracy, T., Jeff, B., Vanessa D., “A Survey of Mobility Models for Ad Hoc Network Research”, Wireless Comm. & Mobile Computing (WCMC): Special Issue on Mobile Ad Hoc Networking: Research, Trends, and Applications, vol. 2, no. 5, pp. 483-502 (2002)
[Reference 3] http://www.scalable-networks.com/
[Reference 4] Zeng, X., Bagrodia, R., Gerla, M.: “Glo-MoSim: A Library for the Parallel Simulation of Large-scale Wireless Networks”, Proc. of ACM Parallel and Distributeal Simulation (PADS '98), pp. 154-161 (1998)
[Reference 5] http://www.opnet.com/
[Reference 6] Okada Kimitaka, Wada Tsuyoshi, Takahashi Yukio, “Urban Pedestrian Model and Urban Pedestrian Flow Simulation based on an Individual Behavior”, The Operations Research Society of Japan, Spring Research Meeting (2003)
[Reference 7] Riley, G. F., “The Georgia Tech Network Simulator”, Proc. of the ACM SIGCOMM Workshopon Models, Methods and Tools for Reproducible Network Research, pp. 5-12 (2003)
In order to conduct a more realistic and detailed evaluation of a mobile network application, the present invention conducts the designing of a mobile node simulator capable of describing a state in which each mobile node changes its behavior based on information from the application and the information obtained from the surrounding environment around the mobile node itself and also capable of performing a simulation based on the description. The mobile node simulator according to the present invention is configured by three components: a behavior simulator section for simulating a part relating to a behavior of a mobile node (person carrying a mobile terminal or vehicle) (e.g., positional information of a mobile node, information regarding the appearance and disappearance of a mobile node and user input to the application); a network simulator section, interactively operating with the behavior simulator section, for simulating the network and the application; and an output section to which the behavior simulator section and the network simulator section outputs simulation results. In preferred embodiments, the simulation results are visually displayed by a graphical user interface (GUI). The simulators according to the present invention enable the following states to be reenacted. For example, in the event navigation application described above, a state in which a movement speed drops due to the concentration of people at a popular event site; a state in which people spontaneously change walking direction to the left in an attempt to avoid collision one another on naturally walking on the left side in order to avoid a collision with each other; and a state in which a number of people move to a specific event site due to the information from the application, thus disrupting the distribution of people and the like can be reenacted. These states affect the network topology and the fluidity thereof. As such, it is possible to evaluate various metrics (e.g., robustness of a routing protocol and adequateness of retransmission control) according to an actual application. Based on that result, it is possible to conduct the setting of parameters (e.g., timing of information retransmission in the application and information transmission interval) and the redesigning of the application.
A mobile node simulator according to the present invention includes: a behavior simulator section for simulating behaviors of a plurality of mobile nodes in accordance with a behavior model, the behavior model being definable by a user; and a network simulator section for simulating a communication on a network including the plurality of mobile nodes, wherein a network application targeted for an evaluation of a simulation is implemented on the network simulator section, the network simulator section is configured to output an output from the network application to the behavior simulator section at each simulation time t, the behavior simulator section is configured to be capable of changing a behavior of at least one of the plurality of mobile nodes in accordance with the behavior model in response to the output from the network application, and the network simulator section is configured to reflect the change of the behavior of the at least one mobile node on the simulation of the communication of the network at each simulation time t, thereby the objective described above being achieved.
The behavior simulation section may be configured to output an input to the network application and positional information and velocity information of each of the plurality of mobile nodes to the network simulator section at each simulation time t, and the network simulator section may be configured to simulate the communication on the network in response to the input to the network application and the positional information and the velocity information of each of the plurality of mobile nodes.
The behavior model may be described in accordance with Condition Probability Event Model (CPE model).
The behavior model may be defined by a state transition diagram representing a state transition of the mobile node.
The behavior simulator section may simulate the behaviors of the plurality of mobile nodes in accordance with the behavior model and area information definable by the user, the area information defining an area in which the mobile node is movable.
The behavior simulator section may simulate the behaviors of the plurality of mobile nodes in accordance with the behavior model, the area information and a behavior scenario definable by the user, the behavior scenario defining at least a timing of generating a node object.
The network may be an ad hoc network for realizing a communication between mobile nodes.
The mobile node simulator may further include an output section for outputting a result of the simulation by the network simulator section.
The output section may be a display section for displaying the result of the simulation using a graphical user interface.
The mobile node may be a person carrying a mobile terminal.
The mobile node may be a vehicle.
A program according to the present invention is a program for causing a computer to perform a simulation process for simulating a communication on a network using a mobile node simulator, wherein the mobile node simulator includes: a behavior simulator section for simulating behaviors of a plurality of mobile nodes in accordance with a behavior model, the behavior model being definable by a user; and a network simulator section for simulating a communication on a network including the plurality of mobile nodes, wherein a network application targeted for an evaluation of a simulation is implemented on the network simulator section, the simulation process including the steps of: the network simulator section outputting an output from the network application to the behavior simulator section at each simulation time t; the behavior simulator section changing a behavior of at least one of the plurality of mobile nodes in accordance with the behavior model in response to the output from the network application; and the network simulator section reflecting the change of the behavior of the at least one mobile node on the simulation of the communication of the network at each simulation time t, thereby the objective described above being achieved.
According to the present invention, by rendering a behavior model of a mobile node definable by a user, it is possible to provide a simulator which can be easily customized in accordance with a behavior model of a specific mobile node. Further, by rendering information of area, in which a mobile node is movable, definable by a user, it is possible to provide a simulator which can be easily customized in accordance with information of area which a specific mobile node is movable. Furthermore, since a simulation of a behavior of a mobile node and a topology and a plurality of communications on a network are interactively performed, it is possible to simulate a situation closer to the real world.
Hereinafter, embodiments of the present invention will be described with reference to the accompanying figures.
(1 Whole Design)
The present specification will describe the designing of a mobile node simulator which is a network simulator capable of evaluating a mobile network application based on which a user can define a behavior of a mobile node based on, for example, a human behavior in the real world. The mobile node simulator according to the present invention is capable of reenacting an affect by the network application on a behavior of the mobile node and the change of the network topology resulting therefrom since a behavior simulator section 140 for performing a simulation relating to mobile nodes and a network simulator section 120 for performing a simulation relating the network interactively operate with each other. As such, the mobile node simulator according to the present invention can perform a more realistic simulation.
An application developer who utilizes the mobile node simulator 100 can specify the behavior of each mobile node (this is referred to as behavior model) for the behavior simulator section 140. In some embodiments, the behavior model is specified by a state transition diagram. As a transition condition for each state transition, a condition relating to information obtained from the network application targeted for the simulation evaluation and information such as obstacles and other mobile nodes existing around the mobile node itself (external information) can be specified. In addition, each transition diagram can hold several internal parameters (e.g., transit place, destination and a degree of interest in a specific spot), and these internal parameters can be also used as a transition condition. Further, with the execution of the transition, the values of those parameters can be altered. Based on this transition diagram, the behavior simulator section 140 calculates information relating to the mobile node at each time (e.g., position and velocity vector of the mobile node). Information of a simulation field including obstacle information (area information), an initial disposition of the mobile node and the like can be specified as a transition condition. These information are passed to the network simulator section 120. Components, such as the network simulator section 120, the behavior simulator section 140 and the output section 160, included in the present simulator can exist in different places from one another and communicate with one another via a wireless link or alternatively, they can be integrated and communicate with one another via a wired link. In some embodiments, since it is ineffective to specify internal parameters for each of a multiplicity of mobile nodes, it is possible to randomly generate the internal parameters based on, for example, a certain distribution and specify the internal parameters to a plurality of mobile nodes collectively (this is referred to as behavior scenario). In addition, similar to a normal network simulator, the application developer can implement an application 180 targeted for the evaluation on the mobile node simulator 100. Further, the application developer who utilizes the mobile node simulator 100 can specify information representing a behavior area of each mobile node (referred to as area information). In some embodiments, the area information is represented by a rectangle representing a geographical area and a set of objects (e.g., buildings) disposed on the geographical area, each of which is represented by a set of line segments. An attribute value can be given to each object, wherein the attribute value indicates whether or not a mobile node can pass through the object or whether or not the radiowave can be propagated over the object.
Here,
An embodiment of the mobile node simulator according to the present invention is, for example, an apparatus including each component shown in
The CPU 210 executes a program. This program is, for example, a program for executing the function of each component shown in
Further, the input interface section 230 functions as an interface for receiving an input from a user. The output interface section 240 functions as an interface for outputting a calculation result. The bus 250 is used for connecting the components 210 to 240 in the computer 200 to each other.
(1.1 Mutual Cooperation Between the Behavior Simulator and the Network Simulator)
Specifically, the information exchange can be performed as in the following. At the time of starting the simulation, information relating to: the disposition of an obstacle; coordinates and a velocity vector of each node at the simulation time 0; and an input to the application 180 are passed from the behavior simulator section 140 to the network simulator section 120. However, information exchanged between the behavior simulator section 140 and the network simulator section 120 is not limited to these information. Thereafter, at each simulation time n*t (n=0, 1, . . . ), the behavior simulator section 140 passes (a) an input to the network application 180 (user input) and (b) coordinates and a velocity vector at time n*t, which have been obtained during the behavior simulation of the immediately previous t simulation time [(n−1)*t, n*t), to the network simulator section 120 for each node (indicated by arrow (a) in
During the simulation time t, the network simulator section 120 calculates, independently from the behavior simulator section 140, the position of each node using the received velocity vector. However, during this time, since there is a possibility that the velocity vector of the node is being changed at the behavior simulator section 140, the network simulator section 120 appropriately receives the accurate current positional coordinates from the behavior simulator section 140 at the information exchange at each time t and corrects the position of the node. The determination of the size of t can be set by a simulator user based on the trade-off between a simulation precision required for the application 180 and a time required for a simulation.
(2 Behavior Simulator Section)
(2.1 Behavior Scenario)
The behavior scenario will be described using a case in which the mobile node corresponds to a pedestrian. When pedestrians are taken into consideration as a whole, a certain urban pedestrian flow (flow of a multiplicity of pedestrians) often occurs in any state. For example, at a ticket gate of a station, before the arrival of a train, an urban pedestrian flow toward the ticket gate occurs since there are many pedestrians walking toward the ticket gate. On the other hand, after the arrival of a train, an urban pedestrian flow away from the ticket gate occurs since there are many pedestrians walking exiting from the ticket gate. Further, in a business space, the number of people going out of and coming into at each store is different. Thus, an urban pedestrian flow in the neighborhood of the store is different from each store. With provision of a parameter for controlling such an urban pedestrian flow as the behavior scenario, it is possible to perform a simulation close to the real life even at a macro point of view. As the behavior scenario in
(2.2 Behavior Model)
Generally, the behavior model is a state transition diagram using an input/output interface for an application and an input interface for visual information as an input/output port and holding a destination list and the like as internal variables. A simulator user can specify several state transition diagrams and a set of nodes which behaves in accordance with thereof. Based on this and the basic values specified in the behavior scenario, an instance provided with initial values of the internal variables can be generated for each node.
(2.3 Area Information)
Area information is defined as an area in which the mobile node can move. The area information is, for example as shown in
A simulation of a mobile node in the case when area information is specified will be considered. In the area targeted for the simulation, objects such as buildings and roads are disposed, and attribute values are given to the objects, wherein the attribute value indicates whether or not the radiowave can be propagated over the objects or whether or not a mobile node can pass through the objects. As an initial state, it is assumed that a mobile node, such as a person carrying a mobile phone, is at Point A on the area targeted for the simulation. The behavior simulator section 140 makes references to the area information to calculate a direction in which the mobile node can propagate radiowave and a direction in which the mobile node cannot propagate based on the objects around Point A and the attribute values of the objects. Further, the behavior simulator section 140 makes reference to the area information to determine a direction in which the mobile node move and a direction in which the mobile node cannot move based on the objects around Point A and the attribute values of the objects and the behavior simulator section 140 also makes reference to the behavior model of the mobile node to calculate a direction in which the mobile node actually moves. The behavior simulator section 140 sends the results of these calculations to the network simulator section 120.
(2.4 Library Group for a Realistic Behavior Model)
The mobile node simulator according to the present invention can include libraries for performing a group behavior and collision avoidance process in order to describe a more realistic behavior easily.
(2.4.1 Group Behavior)
In town, not only does a person walk alone, but also the person walks with his/her friends. Therefore, elements of a group behavior can be introduced and used as a library in order to represent a behavior as a group. In order to realize the group behavior, for example, a reference point in the group behavior presented in Reference 2 can be introduced. The reference point itself moves like an individual node. However, nodes in a group having the reference point behave in accordance with the behavior of the reference point, so that the behavior of the group can be unified. The library for this group behavior can be used, for example, in determining a speed velocity in states “normal move” and “stop-by” shown in
(2.4.2 Collision Avoidance)
In real life, it is natural to walk such that people do not collide with each other. In order to realize such a behavior by the behavior simulator, a collision avoidance process presented in Reference 6 can be introduced. The target to be avoided for collision corresponds to a neighbor in
(3 Network Simulator Section)
The network simulator section 120 according to the present invention changes the topology of the network including mobile nodes in accordance with the behavior of the mobile nodes by the behavior simulation section 140. Also, the network simulator section 120 simulates a network communication between mobile terminals thereunder. Hereinafter, the network simulator section 120 to be used will be outlined and the cooperation between the network simulator section 120 and the behavior simulator section 140 will be described in detail.
(3.1 GTNetS)
In the mobile node simulator according to the present invention, GTNetS (see Reference 7) developed by Georgia Institute of Technology, the United State, can be used as the network simulator section 120, for example. However, a network simulator to be utilized in the present invention is not limited to GTNets, but can be other network simulators having an arbitrary network simulation function. GTNetS is designed under the concept that it overcomes the problem of the scalability in the existing network simulators and performs a simulation of a large-sized network at higher speed. GTNetS has features that it can be implemented by C++, perform operations in parallel and the like. GTNetS implements, by C++ language, each component making up the network simulation and provides the component as a library. By utilizing these libraries so as to describe a simulation scenario, it is possible to utilize a network simulation environment provided by GTNetS. In addition, by creating classes as necessary, it is possible to easily extend the simulator.
[3.2 Realization of the Network Simulator Section Utilizing GTNetS]
Hereinafter, the network simulator section will be described using the case when GTNetS is utilized as an example. In the mobile node simulator according to the present invention, an interaction portion with the behavior simulator section 140 is implemented as a unique class and incorporated into GTNetS, so that the cooperation between the behavior simulator section 140 and the network simulator section 120 can be realized. Also, in a similar manner, a class handling a movement of a node or the like is implemented and incorporated into GTNetS, so that a position of a node can be updated based on the velocity information of the node received from the behavior simulator section 140. The interaction portion can perform a data exchange with the behavior simulator section and perform a process on input data. An input to the interaction portion from the behavior simulator section 140 can be divided into two types: an input relating to a node (e.g., appearance, disappearance and move of a node) and an input to an application. The input to the application is an input to the implemented network application 180. Hereinafter, an implementation of the interaction portion and an input process will be described.
(3.2.1 Interaction with the Behavior Simulator Section)
GTNetS is implemented as a network simulator of a discrete-event type. Simulation events, such as a transmission and receipt of a packet, are inserted into an event queue together with the executed time thereof. Thereafter, the events which have been inserted into the event queue are retrieved and processed in the order of simulation time. A simulation is constituted by repeating the insertion and process of the events into the event queue. In order for the mobile node simulator according to the present invention to perform a data exchange with the behavior simulator section 140 and perform a process on the node information and the application information passed from the behavior simulator section 140, the mobile node simulator according to the present invention creates an interaction event and processes the created interaction event at each simulation time t. First, before the start of the simulation, the interaction event is scheduled at time t. The simulation of [0, t] is performed according to the scheduling, and thereafter, the interaction event is processed at time t, and the exchange of the data with the behavior simulator section 140 and the process on the received data are performed. At the end of the interaction event, the next interaction event is scheduled at time 2t and a preparation is made for next data exchange with the behavior simulator section 140. Such a series of processes is repeated until information of the completion of the simulation is passed by the behavior simulator section 140.
(3.2.2 Extension for Appearance and Disappearance of Node)
GTNetS is targeted for a wired network which is poor in the change of the network topology. Thus, it is preferable to generate all the nodes for performing a network communication before the start of the simulation and not add any new node or delete any existing node during the simulation. On the other hand, in a wireless network, not only do nodes move, but also the appearance and disappearance of the nodes frequently happen. Therefore, it is necessary to appropriately deal with such change of the nodes by GTNeS. For example, a sufficient number of nodes is generated in advance, and a node management class for mapping a node of the behavior simulator section 140 and a node of the network simulator section 120 is introduced, so that it is possible to address the appearance and disappearance of the node by adding and deleting the nodes generated in advance to and from the network topology for node addition and deletion information from the behavior simulator section 140.
(3.2.3 Extension for the Move of a Node)
In GTNetS, a mobility class is prepared, which calculates and manages a node move. The move of each node depends on an instance of the mobility class which is allocated to the node. In the mobile node simulator according to the present invention, in order for the move information (e.g., velocity vector) of the node received from the behavior simulator section 140 to be reflected on the network simulator section 120, a mobility class capable of accumulating sets of velocity and changed time is created and incorporated in GTNetS. At the time when the velocity vector of the node is received from the behavior simulator section 140, the position of the node is not updated, but the accumulated information is updated. Only when the current position of the node is required for a radio calculation or the like, then the positional calculation is performed based on the position and velocity of the node and the accumulated information. However, in such a method which receives only a velocity vector and which cooperates with the behavior simulator, there is a possibility that error may occur due to the cause of a vector precision and the positional calculation. Therefore, positional information of the node is periodically received to correct the error. When the positional information is received, the accumulated information is initialized since there is no need for the previous velocity vector.
(4 Output Section)
The mobile node simulator according to the present invention includes the output section 160. As the output section 160, GUI is preferably used. However, it can be other arbitrary components capable of outputting a simulation result.
(4.1 Function of GUI)
The GUI of the mobile node simulator according to the present invention operating on Windows (Registered Trademark) employs, for example, DirectX9 for a processing speed and a drawing capability. As shown in
The GUI can implement the function of color-coding the circles representing the people such that differences between the states of the nodes can be recognized at a glance. For example, focusing on specific information being broadcast at the network application 180, when a color-coding is performed to check whether or not that information has been transmitted, it is possible to visually comprehend the state of the information being transmitted. Further, when people who have changed their behavior due to the transmission of the information are color-coded, then it is possible to see how the transmission of the information affects the behavior of the people. Moreover, the mobile node simulation according to the present invention can be targeted for a large-sized simulation at a level of thousands or ten thousands of nodes in the future. Abstraction of the representation on the GUI can make the relatively high number of nodes easily recognized and comprehended even if the nodes are simultaneously displayed on the GUI. In addition to these, it is possible by the GUI to assist an input from a user, for example, an input of map information and scenario and an input of an initial disposition of a node.
As described above, the present invention is exemplified by the use of its preferred embodiment(s). However, the present invention should not be interpreted solely based on the present embodiment(s). It is understood that the scope of the present invention should be interpreted solely based on the claims. It is also understood that those skilled in the art can implement equivalent scope of technology, based on the description of the present invention and common knowledge from the description of the detailed preferred embodiment(s) of the present invention. Furthermore, it is understood that any patent, any patent application and any references cited in the present specification should be incorporated by reference in the present specification in the same manner as the contents are specifically described therein.
(5 Exemplary Implementation of the Mobile Node Simulator 100, MobiREAL)
Hereinafter, a description will be given regarding Condition Probability Event Model (hereinafter, referred to as “CPE model”) which is capable of describing a dynamic behavior in which a node changes its behavior in accordance with the surrounding environment around the node itself and the behavior of the network application. Also, a description will be given regarding the designing and implementing of the network simulator MobiREAL capable of simulating the behavior of the node which is based on the CPE model and also capable of simulating the network application which is based on the behavior of the node. The MobiREAL simulator is one exemplary implementation of the mobile node simulator 100 shown in
The CPE model is a model proposed by the inventors as a preferred model for describing a realistic behavior of a node. With the simulation of the realistic behavior of the node based on the CPE model, the influence on the network application and other nodes by the behavior of the node is reenacted, so that it is possible to perform a more realistic system performance evaluation.
(5.1 Configuration of MobiREAL Simulator)
The MobiREAL simulator includes: a behavior simulator for simulating a behavior of a node; and a network simulator for simulating a communication on the network. Herein, the behavior simulator is one exemplary implementation of the behavior simulator section 140 shown in
In the MobiREAL simulator, the behavior simulator and the network simulator are realized as two programs independent from each other. As such, when the behavior simulator and the network simulator are realized as two programs independent from each other, it is possible to utilize the behavior simulator of the MobiREAL simulator in another network simulator. For example, by implementing an interface portion with the behavior simulator in open source simulators, such as ns-2 (see Reference 1) and GloMoSim (see Reference 4), it is possible to easily simulate a realistic behavior of a mobile node in those simulators. Further, when two programs are realized on different computers, it is possible to distribute the load.
Colored components in
The behavior simulator and the network simulator communicate with each other via a TCP connection and at the same time they are executed in parallel. At the start of the execution of the simulation, the simulation area information is passed from the behavior simulator to the network simulator, and they independently hold the area information. After the start of the simulation, information of a node object generated at the behavior simulator is notified to the network simulator whenever necessary, and a corresponding node object is generated on the network simulator side. Further, by providing updated information of a position and velocity speed of the node from the behavior simulator to the network simulator whenever necessary, the position of the node is updated on the network simulator side. The network simulator calculates electronic wave propagation between nodes based on a current node position and area information and it simulates a communication (e.g., packet routing) on the network. The behavior that the node utilizes the network application (data input to the network application) is provided from the behavior simulator to the network simulator, and a data output from the network application to the node is passed from the network simulator to the behavior simulator. Thus, it is possible to conduct a feedback to the behavior of the node due to the network application.
(5.2 Modeling of Behavior Simulator and Node)
Hereinafter, a description will be given regarding a method of modeling of the environment of the real world and the behavior of the node. Also, a description will be given regarding the designing and implementing of the behavior simulator based thereon.
(5.2.1 Modeling of Simulation Area)
An unpassable closed space of such as buildings and squares is specified by closed polygons (e.g., colored areas in
Such a modeling of the simulation area can be effectively performed by using a GUI input assist tool. For example, it is possible to provide a GUI input assist tool having a function of defining a simulation area, including the passable closed space and the entrance to the area described above, by selecting a point on the map using a mouse and outputting a program (e.g., program described by C++) representing the defined simulation area.
(5.2.2 Modeling of the Behavior of the Node)
Generally, there are cases in which a node (e.g., user carrying a mobile terminal) dynamically changes the destination and the like based on information around the node (e.g., obstacles and behavior of adjacent node) or an output of a network application running on the mobile terminal. For example, behavioral changes, such as visiting a crowded destination afterwards or inputting information of a product desired for purchase on a navigation system which searches a neighboring store and moving toward the store presented by the system, are considered. In addition, a node often moves after scheduling the destination, the route, the estimated arrival time and the length of the stay to some degree, and on the other hand, the behavior is often probabilistic.
From the view point described above, the present inventors propose Condition Probability Event Model (CPE model) describing a dynamic behavioral change rule of a node by using a variable. Further, they propose a method of modeling the behavior of the node using the CPE model. In the method, the modeling of the behavior of the node is performed by the following procedure. First, a dynamic behavioral change rule and the like is described by using the CPE model for each behavior type of a node (e.g., commuter, shopper and the like) in an all-purpose format which uses a variable. Next, at the time of executing the simulation, a node object specified with a specific variable value (e.g. destination and the like) is generated in accordance with the simulation scenario.
The CPE model is defined by a list of “rules”, a set of internal variables and a set of external variables. Each rule is configured by a set of “condition”, “probability” and “behavior”. When the “condition” is satisfied, a corresponding rule will describe how a node “behaves” in accordance with the “probability”. The internal variables are variables which can only be updated and made reference to from the inside of the CPE model. The external variables are variables which can be updated and made reference to from the inside and outside of the CPE model.
The external variables include, for example, a simulation time T, a surrounding information (e.g., information of adjacent node and obstacle) E, an input to the network application AI, an output from the network application AO, the current position P of a node and a velocity vector V of the node.
The “condition” of each rule is specified with a logical formula using external variables or internal variables. The “probability” of each rule is specified with a constant real number between 0 and 1 or probabilistic function.
The behavior model describing the behavior of a node based on the CPE model is incorporated into the behavior simulator in a format which can be executed by the behavior simulator (e.g., program described by C++). The behavior simulator searches the list of the rules from the top and performs the behavior specified by the first rule, which satisfies the specified condition, with the specified probability. These steps are repeatedly executed.
Hereinafter, a case in which the network application targeted for the evaluation of the simulation is a network application on a mobile ad hoc network (MANET) will be considered. In this case, a pedestrian carrying an information terminal equipped with a short-range inter-terminal communication device (e.g., IEEE802.11) visits several stores for shopping and distributes the obtained information about the stores (e.g., sale information) to other pedestrians met by the pedestrian on the move, thus sharing useful information. Such a network application is implemented on the network simulator.
For example, in rule E3, in the case of being too late to arrive at the destination within the scheduled time even if walking fast, the estimated arrival time dst. t is set late by 10 minutes with the probability of 0.2 and a value when walking fast is calculated and substituted into the velocity vector V. Generally, a pedestrian temporarily changes the moving direction in an attempt to avoid an approaching pedestrian or a small obstacle. The calculation of the velocity vector V is performed in consideration of such collision avoidance.
In rules E7 and E8, a behavior when staying beyond the scheduled length of the stay at the destination is specified using the same condition. Rule E7 is of a higher priority than rule E8. Thus, when the condition for those rules are true, the behavior of rule E7 is performed with the probability of 0.8, and if the behavior of rule E7 is not performed, then the behavior of rule 8 is always performed.
In rule E5, in the case of obtaining store information new_dst from another pedestrian via the network, it describes the behavior of adding the store to a destination list with the probability of 0.5. Further, in rule E9, in the case of leaving the destination (store) for which the stay was extended, it describes the behavior of distributing the destination information (store information) to other pedestrians via the network.
Rule E1 does not show the probability of the behavior to be performed with a constant. Instead, it is represented by a probabilistic function according to the normal distribution having an average of six minutes and a standard deviation of two minutes. It is possible to use an arbitrary function as the probabilistic function, and it is also possible to specify a deviation of the time until the node makes a move after a certain condition is satisfied and a frequency of the behavior.
The conditions and behaviors which are often used can be provided as libraries. Thus, it is possible to describe the CPE model easily.
(5.2.3 Simulation Scenario)
In the simulation scenario, a parameter such as an execution time of a simulation is set, and also a timing of generating a node object corresponding to each CPE model, an initial value of a variable and the like are specified.
Such a simulation scenario can be generated based on, for example, the Urban Pedestrian Flow. The urban pedestrian flow shows the frequency of the appearance (person/second) of pedestrians (nodes) who take each route defined in a simulation area. The generation of the urban pedestrian flow is effectively performed by using the GUI input assist tool. For example, it is possible to provide the GUI input assist tool having a function of selecting points on a map using a mouse so as to input the density of the pedestrians (nodes) at several points in the simulation area, inputting candidates for a destination (a point where e node appears and disappears) so as to automatically generate the urban pedestrian flow, and outputting a program representing a simulation scenario described in accordance with the urban pedestrian flow (e.g., program described by C++).
For example, it is possible to generate an urban pedestrian flow closer to the actual urban pedestrian flow by inputting the actual measured density of the pedestrians in the real world as the density of the pedestrians (nodes) described above and inputting points at an edge of an actual map, an entrance to a building, an entrance to an underground shopping area and the like as candidates for a destination described above (a point where a node appears and disappears).
(5.3 Network Simulator)
In addition to the basic function of simulating a communication on the network, the network simulator of the MobiREAL simulator requires in periodical cooperation with the behavior simulator: a function of reflecting information relating to the generation and deletion of the nodes passed from the behavior simulator, information relating to the position and the velocity of the node and the like on the simulation of the network; and a function of exchanging the input to the network application implemented on the network simulator and the output from the network application with the behavior simulator.
The present inventors developed the network simulator of the MobiREAL simulator by implementing the functions described above in GTNetS (see Reference 7).
Generally, many network simulators including GTNetS are implemented without assuming the addition of any new node or the deletion of any existing node during the execution of a simulation. In contrast, the present inventors modified a node management process for GTNetS and realized the dynamic generation and the dynamic deletion of a node. Also, in order to improve the reality of the simulation, a physical layer simulation module was extended so as to calculate radio wave propagation in consideration of the affect by obstacles.
The behavior simulator and the network simulator of MobiREAL simulator are two simulation programs which independently hold simulation area information, node objects and the positions and velocities of the nodes. The behavior simulator only performs a simulation of the node behavior. The network simulator simply moves the node based on the position and velocity of the node grasped by the network simulator itself, and performs a simulation of communication on the network independently from the behavior simulator. As such, the node's positional information updated by the behavior simulator and the data output from the network application calculated by the network simulator are periodically exchanged at t simulation time interval, so that the consistency between the two simulators is kept. When one of the simulators finishes the simulation up to the data exchange earlier than the other, the one of the simulator waits until the other simulator finishes the simulation and completes the data exchange.
If t is set as a value as small as possible, it is possible to perform an accurate simulation which nearly completely reflects the position and the velocity of the node determined by the behavior simulator on the network simulator side. However, the frequency of the communication between simulators increases by that much, thus taking more time for executing the simulation. The determination of the value of t can be set based on the trade-off between a precision required for the simulation and an execution time. t is normally set as 1.0 seconds.
The cycle of each CPE model scan in the behavior simulator is set as 0.2 seconds. In the network simulator (GTNetS), the simulation precision depends on the implementation of each layer, and the simulation time is represented by double-precision floating-point number-type variable (unit is a second).
(5.4 Animator)
The visualization of the simulation result is extremely important for the network simulator. Especially, the visualization of the behavior of a node is inevitable in order to confirm the influence by the network system on the behavior of the node. Thus, the MobiREAL simulator provides an animator for visualizing the simulation result. The animator is one exemplary implementation of the output section 160 shown in
The animator can operate on Windows (Registered Trademark) and can represent the state of a node, the state of a link, the state of a radius extended by radiowave, the state of a packet propagation and the like using animations. Each item of these can be selected for display or non-display. Each type of the packets (data packet, control packet and the like) can be displayed in different color from the others. In addition, the color display of the node can be freely specified in the network application implementation on the network simulator. For example, it is possible to display a node which has received specific information from the network application in red and the other nodes in blue.
The present invention can be easily customized in accordance with a behavior model of a specific mobile node, and the present invention is useful as a simulator or the like capable of simulating a situation closer to the real world.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2004-253457 | Aug 2004 | JP | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/JP05/15774 | 8/30/2005 | WO | 00 | 10/10/2008 |
