Mobility Modeling in Mobile Ad Hoc Networks with ... - Semantic Scholar
54
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Mobility Modeling in Mobile Ad Hoc Networks with Environment-Aware Gang Lu Department of Computer Science, University of Sheffield, UK [email protected]
Gordon Manson and Demetrios Belis Department of Computer Science, University of Sheffield, UK { g.manson, dimi }@dcs.sheffield.ac.uk
Abstract—Simulation is the most important and widely used method in the research of Mobile Ad hoc NETworks (MANET). The topology of MANET and the mobility of mobile nodes are the key factors that have an impact on the performance of protocols. However, most of the existing works are based on random movement and the fact that the network topology is highly related to the environment of MANET is overlooked. In this paper, we propose a novel Environment-Aware Mobility (EAM) model which models a more realistic movement of mobile nodes. Environment objects such as Route and Hotspot are introduced to represent the environment components which are rendered by Scalable Vector Graphics (SVG). It is considered to be a complex model with a combination of existing conventional mobility models and network environments. This paper shows that various MANET environments can be modeled based on this work. A sample environment is also simulated and the results show that the intrinsic characteristics and properties of the environments have a significant influence on the performance of MANET protocols. Index Terms—Mobility Modeling, Environment-Aware Mobility Model, Routing Performance, MANET.
I.
INTRODUCTION
Without any central administration, MANET is considered to be the dominating form of wireless network and has drawn more and more attraction from the industry. Using self-organised mobile nodes, MANET can be deployed in any environment, such as a gallery, a theatre, a shopping mall and even a street. Many researchers have shown interest in the fields of MANET and all sorts of protocols aiming at different issues have been proposed in recent years. The performance of these protocols needs to be carefully evaluated before they are ready fore the commercial market. Therefore, network simulation plays a very important role in this area, and is widely used by researchers as a key method to better Based on “A Novel Environment-Aware Mobility Model for Mobile Ad Hoc Networks” by Gang Lu, Demetrios Belis, and Gordon Manson which appeared in the Proceedings of the 1st International Conference MSN 2005, pp. 654-665, Wuhan, China, Dec. 13-15.
© 2006 ACADEMY PUBLISHER
understand the overall performance of MANET. Self-organization is the essential characteristic of this mobile wireless network. The mobile nodes within the network can be any moving or fix-position objects equipped with antennas. They can be either humans moving in a mall or gallery, or insects and animals in nature environments. The movement behaviors are highly dependent on their own mobility factors and the areas they are located. The mobility model is therefore regarded as an important component in ad hoc network simulation. In this paper, the mobility modeling is discussed by proposing a novel realistic model which is incorporated with the environments where a MANET is deployed. Environments are defined by introducing some environment objects, such as a Route, a Junction, an Accessible Area, and a Hotspot etc. The mobile nodes are modeled with the concern of node heterogeneity. The signal-blocking issues are also covered in this model by introducing the ClosedArea whose boundaries are considered to be obstacles with no radio penetration. In addition, with the use of Scalable Vector Graphics (SVG) [5], environments can be easily designed and the movements in the environment can be visually observed. The remainder of this paper is organized as follows. Section II details some related work in the research of mobility models. Section III describes the detailed information about how the Environment-Aware Mobility model works. The details of the evaluation are presented in Section IV, and the results are presented and discussed in the last section. II.
RELATED WORKS
There exists many mobility models proposed for mobile wireless networks. Mobility models are used to describe the movement pattern of mobile nodes. In this section, some of the popular models used in MANET are briefly described. The relevancy between EAM and those models is clarified afterwards. The category of conventional models is illustrated in Fig. 1. The Entity Models are used to model the movement behavior of an individual mobile node. The
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 1. Category of Mobility Models.
Random Waypoint model [1] is widely implemented by network simulators such as ns2 [4]. In each movement epoch, the mobile node picks a position within the simulation area and moves towards it with a speed distributed in the range [νmin, νmax]. Instead of moving to the next destination immediately as soon as it arrives at the current destination, the mobile node pauses for a specified time then repeats the procedure. The major drawback is that the nodes tend to move around the centre area so that they are not really distributed into the entire simulation area. The Random Direction [1] model is proposed to avoid this distribution problem. It forces the mobile node to move until it reaches the border of area before it starts its next epoch. The models introduced previously are all considered to be memory-less models since the next movement segment has no dependency on the previous movement regarding either speed or direction. This memory-less feature causes frequent sharp change in speed and direction of movement which is obviously not applicable in a realistic world. Some memory models are proposed. The Gauss-Markov model was first introduced in [6]. In this model, the velocity of the mobile node is assumed to be correlated over time and modeled as a Gauss-Markov stochastic process. The Boundless model [1] is another example of a memory model but with no geographical restriction. Mobile nodes are allowed to cross the boundaries and appear at the other side of the area. The resulting effect is that the simulation area is modeled as a torus instead of a flat area. In ad hoc networks, there are many situations where the movements of mobile nodes have some correlation with each other, i.e. mobile nodes have some group behavior in common. To present this characteristic, Group Models have been proposed. One typical example is the Reference Point Group Mobility model (RPGM) [3]. In RPGM, a logical centre of a group is defined and its movement is used to direct the group-wide movement. Individual members of the group move not purely on a random basis, their movements are also affected by the group movement. RPGM is popularly used in research to depict some scenarios with group behaviors such as avalanche rescue. Other group models can also be found in [1]. The previous models all assume that the simulation area is a free space area where mobile nodes can move anywhere inside. They expose the self-organization feature of mobile nodes, but they are not generally applicable and the geographic factors have to be considered. The Pathway models [7] and the Obstacle models [8] partially overcome this disadvantage. The Pathway Model forces each mobile node to move along the shortest path towards its destination. Similar behavior © 2006 ACADEMY PUBLISHER
55
is also modeled in the Freeway mobility model and in the Manhattan mobility model in [9]. The Obstacle mobility model was first introduced in [12]. Unlike the Pathway model, the Obstacle model defines some obstacles in the simulation area. These obstacles have a signal-blocking effect on the communication of mobile nodes. This model also allows the nodes to change this movement trajectory when obstacles are encountered. Geographic models support more realistic scenarios than Entity and Group models. Several works have brought the idea of real-world simulation into MANET simulations. The realistic environment in which the network exists needs to be constructed before running the simulation. Ref. [10] provides a way to achieve this by using Auto-CAD, and the realistic world is constructed by implementing the knowledge of the Voronoi diagram in [12]. The movements of mobile nodes are restricted by the predefined arbitrary obstacles and pathways. However, this real-world environment model does not provide a flexible approach to model the mobility variations. The movement trajectory of a mobile node might change over time. For example, the mobile nodes move along the pathway then may enter a particular terrain, such as a park where their movements become less deterministic, therefore they might walk in any direction, i.e. the pathway restraints are no longer applied. As a conclusion, every model mentioned above has some abilities to model the mobility of mobile node in MANET. They can be applied to particular situations, but none of them is flexible and suitable enough for modeling more realistic scenarios. As a mater of fact, Entity Models and Group Models can co-exist in some scenarios with obstacles. The Obstacle model seems to give a good solution for signal-blocking but it leaves entity and group mobility properties unconcerned. Obstacles and Paths are detected but the trajectories to deal with obstacles are simple and not general. Depend on the roles they play in the environment, the mobile nodes can have different mobile characteristics. The fact that environment factors can affect the movement is also overlooked by all of them. By taking the environment into account as illustrated in Fig. 1, the EAM model proposed in this paper provides a more general approach to model more realistic ad hoc networks. III.
ENVIRONMENT-AWARE MOBILITY MODEL
The EAM model proposed in this paper is designed to model the movement behavior of mobile nodes in the environments of realistic ad hoc networks. By studying the possible environment where MANET is located, different sub-areas within the entire simulation area are abstracted to several environment objects, such as a Route, Junction, Hotspot, etc. The movement trajectory of the mobile node is correlated with the sub-area that it is located and is allowed to be changed during the simulation. The node heterogeneity is also concerned for better describing the mobility of mobile nodes. The obstruction of radio propagation is also implemented in this model, and some of the conventional models are
56
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
TABLE I.
DESCRIPTION OF ENVIRONMENT OBJECTS
Name
NORAA (N) Figure 2. Workflow for mobility modeling using Environment-Aware Mobility Model. Accessible Area (AA)
Lane (L)
Path (P)
Figure 3. Structure of Environment Objects (EOs).
A. Environment Objects Different types of sub-areas are abstracted by Environment Objects (EOs). The EOs can be classified into two categories: Non-Accessible Area (NAA) and Accessible Area (AA). NAA represents the restricted area where no movement is allowed. AA represents some areas where mobile nodes can move inside and may move in and out. An AA can be any of the Lane, Path, Route, Junction, NORmal Accessible-Area (NORAA) and Hotspot objects. Some hierarchical relationships also exist in EOs. For instance, a Route is composed of Paths, Junctions and even NORAAs, a Path can be composed of Lanes, and NORAAs can be a container area for Hotspots. 0Fig. 3 illustrates the structure of EOs. With the introduction of EOs, real complex environments can be simplified and easily constructed.
© 2006 ACADEMY PUBLISHER
Route (R) Hotspot (H) Non-Accessible Area (NAA)
modified to take into account the environmental effects and to cope with the obstacles. In our proposal, the environment including all the environment objects is designed with SVG. The relationships between sub-areas and various properties of each sub-area are given by a text-formatted configuration file (Environment Configuration File). A SVG file is generated after the simulation is finished. This animationsupported SVG file can be very useful for researchers to explicitly observe the movements in the simulated scenario. The details shown in this SVG file can be configured by the users by modifying a text-formatted configuration file (Output Configuration File). When the simulation starts, this model reads all three files to become aware of where the environment objects are and their properties. The mobile nodes are distributed into the accessible sub-areas. The mobility factors such as speed, direction and the mobility model attached to each node are determined by the properties of the areas they are located. All the movements are recorded and converted to ns2 compatible movement scenario file at the end of the simulation. ns2 is also modified in order to be able to read an obstacle boundaries description file to block radio signal. Fig. 2 illustrates the work flow of this model.
Junction (J)
NAA
Description If there are no sub-areas located inside the NORAA, the mobile nodes can move freely inside its area; However, a NORAA can contains Hotspot and NAA, in which case the behaviors of mobile nodes will be affected and restricted. One pair of Entry and Exit must be defined for a Lane. Once the mobile node moves into a Lane from the Entry, the node has to move towards the Exit. A Path can consist of one Lane or multiple parallel Lanes. The mobile nodes moving in a Path are not allowed to change its Lane and they must move from the Entry towards the Exit of the Lane where it is located. It is used to connect two Paths, or two NORAA, or one Path and NORAA. It is the resulting EO if the Path, Junction and Route are joined together. It is used to represent the attracting area where the mobile nodes may visit frequently. It is the area where no mobile node can move inside. The boundary effect will occur if the mobile node encounters a NAA in the way.
Figure 4. An example snippet of Environment Layout File.
Each EO is given some intrinsic characterises which have influence on the mobility of the mobile nodes. Typically, the mobile node located in the Lane must follow the Lane to the exit with a maximum speed limit; if a Junction is encountered, a mobile node is forced to choose one of the exits to enter the next adjacent EO. NORAA is a very flexible EO because it can be used as a container area. If it is a free space area, a mobile node can move using its conventional mobility model. If it contains Hotspots, mobile nodes will be forced to commute among the Hotspots. If it contains a NAA, the accessible space for
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
57
TABLE II. THE PROPERTIES OF ENVIRONMENT OBJECTS TABLE III. THE ASSOCIATIONS OF THE ENVIRONMENT OBJECTS AND THE PROPERTIES
Properties
Hotspot
Junction
Lane
Path
Route
Value
The maximum speed at which the mobile nodes can move when they are located in this area. The speed in meter/second (m/s). Any speed not less than 0.0m/s. Example: 3.5(m/s)
NORAA
Description
NAA
MaxSpeed
MaxDirection Description Value
It is the maximum direction in degree towards which the mobile nodes can move when they are located in this area. The angle in degree, ranges from -180.0 to 180.0 degree. Example: 180 (degree)
MinDirection Description Value
It is the minimum direction in degree towards which the mobile nodes can move when they are located in this area. The direction in degree, ranges from -180.0 to 180.0 degree. Example: -180 (degree)
InboundAreas Description Value
It is used to specify the entry areas where the mobile nodes can move into this area from. The name of the EO. If more than one area exist, the names of EOs must be separated by ‘’. Example: J2-A1
MaxSpeed MaxDirection MinDirection InboundAreas OutboundAreas Capacity MaxDuration Components ClosedArea MobilityModel
√ √ √ √ √ √ √ √ √ √
X X X X √ √ X X X X X √ X X X X X √ X X X √ √ √ X X X √ √ √ X X X X X √ X X X X X √ X X X X √ √ X X X X X √ X X X X X √ √ = the property is applicable to the EO X = the property is NOT applicable to the EO
OutboundAreas Description
Value
It is used to specify the exit areas where the mobile nodes can move into from the area it is located. The name of the EO. If more than one area exist, the names of EOs must be separated by ‘’. Example: J1-A2
Figure 5. Typical configurations for a Normal-Accessible Area.
Capacity Description Value
The maximum number of mobile nodes the area can accommodate Any positive integer. Example: 10
MaxDuration Description Value
The maximum duration of time the mobile nodes can stay inside the area The duration in second (s). Example: 40 (s)
Components Description
Value
It is used to specify the sub-areas located inside this configured area or the areas that it consists of. The name of the EO. If more than one area exist, the names of Eos must be separated by ‘’. Example: H1-H2
ClosedArea Description Value
It is used to indicate if this area is considered as a signal-blocking area. If this is set to be true, its boundaries will be regarded as obstacles. Boolean value, True/False. Example: true
MobilityModel Description
Value
It is used to specify which mobility model the mobile nodes should use once they move into the area. The name of mobility model. The mobility model can be any one of Random Walk, Random Waypoint, Random Direction, GaussMarkov, Boundless, Hotspot and RPGM. Example: randomwalk
the mobile nodes is reduced. The full description of the EOs is given in TABLE I.
© 2006 ACADEMY PUBLISHER
B. Environment Layout Design Scalable Vector Graphic (SVG) has been widely accepted as a graphic standard, and it is the key approach used by the Environment-Aware Model to produce the simulation environment. SVG uses XML to describe 2dimention graphics. SVG contains a set of basic shape elements, such as the rectangle, circle, line, and polygon. These elements are used to represent the environment objects with arbitrary shapes. SVG also supports the abilities to change the vector graphic over time which provides the capability of visualizing the movements of mobile nodes. The full specification of SVG can be found in [5]. A typical snippet of an SVG Environment Layout File (ELF) is attached in Fig. 4. Note that every Environment Object is identified by the ID of the SVG shape element. A simple convention is set: the ID must start with AA (for Accessible Area) or NAA (for Non-Accessible Area) then followed by an underscore then its name which is composed by the object’s type plus the index. For example, if an element has an ID of “AA_L1” (‘L1’ is the name of this object), it means the sub-area represented by this element is an accessible Lane with index 1. C. Environment Configuration The SVG layout file provides the model with the location of each EO and helps identify their types. The relationships among EOs mentioned along with their properties are configured in a text-format Environment Configuration File (ECF). It is also assumed in EAM that
58
some of the EOs might have some properties in common. They may have restraints on the speed, the direction and capacity which is the maximum number of mobile nodes the EO can accommodate. TABLE. II gives the full description of all properties defined in the EAM model and TABLE. III shows the association of the properties and the EOs. All these properties are configurable in the ECF. As an example, a typical configuration for the NormalAccessible Area with ID=AA_A1 which exists in the environment of Fig.4 is given in Fig. 5. Referring to the description in TABLE II, it is very easy to derive that the area A1 is adjacent to Junction J2 which is its entry area. A1 can be considered as a room since it is set to be a Close area, it contains three Hotspots (H1, H2 and H3) and maximum 10 mobile nodes are allowed to stay in at the same time. As for the mobility factors, the mobile nodes located inside can move in any direction but with speed up to 7m/s, and the general movements are modeled using the Hotspot Model. D. Environment-Aware Movement The EAM can integrate many Entity Models such as the Random Waypoint model, the Boundless model and the Group Models like RPGM with some modifications. In order to cope with the complex environments composed by the EOs, two sub-models are proposed: the Route model and the Hotspot model. These two models are described in detail and intensively studied in [13]. Boundary Effects are implemented to cope with the situation when NAAs are encountered. y Movement in the Route model: The Route model can model some complex environments, such as a city area or an urban road traffic network. The mobile nodes moving in a Route travel through all its components in a specified order given in the ECF file. When they reach the end of a Route, the mobile nodes will appear at the entry of a new Route in order to keep the number of nodes constant during the simulation. The Route Mobility model can be used to model the movement described in the Pathway model, the Manhattan model, the Virtual Track model, and the Obstacle model. Nevertheless here the movement in a route can use the Random Walk model as well since the Route model allows the existence of a NORAA to bridge two Paths. Movements can also be restrained by the route. For instance, the speed may not exceed the maximum allowed speed of a lane. y Movement in the Hotspot model: A hotspot is a place where most of the mobile nodes are likely to visit. If hotspots exist in an area, the strategies with which the mobile nodes move between the hotspots can be varied because a hotspot might have a capacity to accommodate a maximum number of visitors and a maximum duration that the mobile nodes are allowed to stay. A number of different strategies are introduced in [13]. In this paper, a simple strategy is applied: the mobile nodes always adjust their speed to make sure that the hotspot can allow at least one more nodes to be added when it arrives at the hotspot.
© 2006 ACADEMY PUBLISHER
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 6. Boundary Effects applied to different models.
Figure 7. Example of Obstacles Description File.
y Boundary Effects As mentioned earlier, various models are modified and given environment-aware capability. For example, different behaviors are designed for the Random Walk model and the Random Waypoint model when a NAA is encountered. Bouncing behavior is applied to the Random Walk model. Fig. 6(a) illustrates the effect: the node bounces to another random direction at boundary e1. Compared with the Random Walk model, the Random Waypoint model features a target-driven behavior so that Bouncing is not applicable. Thus, a different boundary effect called Surrounding is proposed. With the Surrounding behavior, once the next destination is determined, the mobile node will try to get there even if there are obstacles blocking the expected path. Fig. 6(b) illustrates this effect: the node starting at S moves towards D, it actually travels along the boundaries e3-e4e1, and then gets to D. E. Signal Obstruction Radio propagation could be one of the major issues that influence the performance of MANET. The received signal is very vulnerable to suffer significant power dropdown due to attenuation, multipath fading, etc. The TwoRay Pathloss propagation model is used to calculate the power of the received signal. In EAM, it is assumed that if there is a boundary of an obstacle geographically located between the LOS range of the pair of mobile nodes, then the signal transmitted from either of them will not reach the other node. OpenAreas and ClosedAreas are defined to help the model identify which EO should be considered as an
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 8. Illustration of Bus-aided IVC System.
Figure 9. Illustration of RoadLamp-aided IVC System.
Obstacle. A text-formatted Obstacle Description File (ODF) is generated for the environment. If one EO is configured to be a ClosedArea, Its boundaries locations are added into the ODF which will be processed in the physical layer of ns2. Fig. 7 gives an example environment and its relevant ODF. F. Node Heterogeneity The mobile nodes can have different behavior depend on the roles they play in the environment. In Environment-Aware Mobility mode, the mobile nodes are classified into five categories based on their mobile characteristics. They include High Mobility Node, Low Mobility Node, Restricted Nodes, Bus Node, and Static Node. A High Mobility Node (HMN) is the most common node. The nodes belong to this category normally can have relatively higher range of speed compared to other nodes. Another important characteristic is that all the HMNs have certain randomness. Every node is moving towards a random destination through one of the available routes; they can have different random speed, but they must have more or less the same average speed depend on the traffic load. One typical High Mobility Node is the car. A Low Mobility Node (LMN) is actually a special case of the Normal Node. The nodes belonging to this category have the same level of randomness as Normal Node, but they usually are moving with much slower speed. A typical example of this category is the pedestrians.
© 2006 ACADEMY PUBLISHER
59
A Restricted Node (RN) is normal located inside certain obstacles such as a building, a stadium and so on. The nodes belonging to this category normally exhibit lowest speeds inside a relatively small sub-area of the environment. It is noted that the behavior can be changed to be a Restricted Node once the High Mobility Node and Low Mobility Node move into a restricted area. A visitor to an Art Gallery, for example, can be considered as a good representative of this category. A Bus Node (BN) is defined to represent mobile nodes such as a bus or supertram. The nodes belonging to this category normally have the middle speed between the LMN and HMN, but more importantly, the node’s average speed is relatively constant in order to catch up with its timetables and it is normally moving in a loop along a pre-defined route and pauses for a certain time at some locations along the route. A Static Node (SN) is defined to represent the nodes which are not moving during the observation period. The speed of a Static Node is considered as 0m/s. The traffic lights located at the junctions and the roadlamps located along the routes can be treated as typical examples, assuming they are equipped with radio transceivers. With the consideration of the node heterogeneity, some particular mobile ad hoc networks can be simulated in various environments, for example the so-called Busaided Inter-Vehicle Communication (Bus-aided IVC) system as illustrated in Fig. 8. It takes the advantage of the regular movement of the bus helping the connection of two physically out-of-communication-range vehicles; The RoadLamp-aided IVC system shown in Fig. 9 is another particular communication system where some of the roadlamps or traffic lights are assumed to be equipped with wireless transponders and serve as the access points. IV.
SIMULATIONS
The Environment-Aware Mobility model is very flexible and can be applied to various environments. The main purpose of the simulation is to show that the performance of the ad hoc networks protocols is not only influenced by mobility models, but also significantly affected by the Environment Objects which are the fundamental sub-environments abstracted from the real environments. With the presence of EOs, the movement behaviors of mobile nodes modeled by the mobility model are also restricted by the properties of the EOs. Moreover, by introducing Obstacles, the network connectivity is also highly influenced. The resulting topology change is evaluated by the measurement of the average numbers of connection changes at different speeds. One popular routing protocol AODV [11] is simulated. Three typical metrics, Control Packet Overhead, End-toEnd Delay and Data Packet Delivery Rate are used to evaluate the routing performance. For comparison, all the metrics are measured repeatly with increasing speed from 0m/s to 9m/s. In order to show the impact caused by the environment, two ECF files are used and simulated individually. Both the Obstacle and Obstacles-less environment are
60
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 10. Environment Layout of the simulation.
Figure 11. Snapshot of the running scenario.
simulated and compared. The SVG Layout is shown in Fig. 10. A. Environment Description A 600x300m area is used to represent the simulation area. Two different MANET environments, The Art Gallery and Festival Event are used. The Art Gallery: In this environment, Area 1 and Area 2 can be considered to be two rooms separated by a wall (NAA). All the arts exhibited in the rooms are presented by Hotspots. Visitors enter this area from the entry. They may continue moving to the next exhibition section through Exit 1 or enter one room (Area 1 or Area 2). Once the visitors enter the room, they will move towards the art object that they find most attractive then the next and so on. If a visitor notices that his art object has too many visitors (up to the capacity of Hotspot), they will go to the next most interesting object. Once they have visited all the art objects or if they have been in the room too long (up to maximum duration allowed to stay in the area), they will leave the room and continue the visit through Exit 2 or Exit 3 as shown in Fig. 10. In this case, both areas are ClosedArea because the signal is very likely to be blocked by the walls. The Festival Event: Compared to the Art Gallery environment, this environment exists in some public places, such as a park so that the two areas can be assumed to be OpenAreas. In this case, when mobile nodes move into Area 1 or Area 2, they may just walk around freely for a while then leave. Note that if some entertainments or shops are distributed inside the areas as hotspots, the behaviors of nodes could be similar to the previous one. B. Simulation Description All the simulations are done by ns2. 40 mobile nodes with 100m omni-directional antenna transmission range are distributed over the simulation area. The Hotspot model is used for the modeling the Art Gallery environment and the random movements in the Festival Event are modeled by the Random Waypoint model. The simulation is repeated with different average nodes speeds ranging from 0m/s to 9m/s and the final result is an average of the simulation results from 40 runs. To evaluate the impact of Obstacles on different mobility models, both environments are simulated with Obstacles and without Obstacles. Area 1 and Area 2 are
© 2006 ACADEMY PUBLISHER
Figure 12. Setting for Bus Nodes in Output Configuration File (OCF).
assumed to contain minimum of 15 mobile nodes inside, and the other 10 nodes are distributed randomly along the routes. If any of the 10 nodes are located inside the two sub-areas, they will stay there for a certain time then leave. If they are located in Lanes, they will randomly choose a route. For the Random Waypoint model, a 10 seconds pause time is spent by each node when it gets to the current destination and the pause time in the Hotspot model is depend on the property of each Hotspot. In the Art Gallery environment, nodes move about with a maximum speed of 7m/s inside the two rooms, but there is no limitation for the Festival Event because it is considered to be an out-door event. In order to keep the number of mobile nodes constant during the simulation, the nodes that arrive at any of the Exits will appear at the Entry to start a new journey. Each scenario is simulated for 500 seconds. The measurement is started after 50 seconds to ensure all the nodes have been distributed throughout the areas. The number of connectivity between neighbors is calculated every second. The average changes of connectivity at each speed are measured as the metric to evaluate the influence on the network topology caused by the mobility models and the environments. Environment Layout AODV routing protocol is used to evaluate the impact on the routing performance. The CBR data packet size is set to 512 bytes and the sending rate is set to 4 packets per second. A maximum of 20 data connection is allowed at one time. A node will act as either a sender or receiver at any time during the simulation. V.
RESULTS
A. Scenario Visualization By utilizing the animation elements of SVG [5], the simulated scenarios can be visualized with animation by any SVG-viewer or SVG-supported web browser. Every
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
61
Data Packets Delivery Rate in Environment-Aware Model
Number of Connectivity Changes in Environment-Aware Model 1.000
9.0
EAM with Route sub-model (20nodes) with Obstacles EAM with Route sub-model (10nodes) with Obstacles EAM with Route sub-model (20nodes) without Obstacles EAM with Route sub-model (10nodes) without Obstacles
8.6 8.2
0.970 0.955
Data Packets Delivery Rate (AODV)
Number of Connectivity Changes
7.8
0.985
7.4 7.0 6.6 6.2 5.8 5.4 5.0 4.6 4.2 3.8 3.4 3.0 2.6 2.2
0.940 0.925 0.910 0.895 0.880 0.865 0.850 0.835 0.820 0.805 0.790 0.775 0.760 0.745
1.8
EAM with Hotspot sub-model with Obstacles EAM with RandomWaypoint with Obstacles EAM with Hotspot sub-model without Obstacles EAM with RandomWaypoint without Obstacle
0.730
1.4
0.715
1.0 0
1
2
3
4
5
6
7
8
0.700
9
0
Average Speed of Mobile Nodes (meters/sec)
1
© 2006 ACADEMY PUBLISHER
3
4
5
6
7
8
9
Average Speed of Mobile Nodes (meters/sec)
Figure 13. Average number of connectivity changes.
Figure 14. Average Data packets delivery rate.
Number Of Control Packets (AODV)
Number Of Control Packets in Environment-Aware Model 22200 21500 20800 20100 19400 18700 18000 17300 16600 15900 15200 14500 13800 13100 12400 11700 11000 10300 9600 8900 8200 7500 6800 6100 5400 4700 4000
EAM with Hotspot sub-model with Obstacles EAM with RandomWaypoint with Obstacles EAM with Hotspot sub-model without Obstacles EAM with RandomWaypoint withoutObstacles
0
1
2
3
4
5
6
7
8
9
Average Speed of Mobile Nodes (meters/sec)
Figure 15. Average number of control packets sent. Data Packets End-to-End Delay in Environment-Aware Model
Data Packets End-to-End Delay (AODV)(sec)
movement of each mobile node is recorded into the Mobility Scenario File (MSF) in SVG format. Fig. 11 shows a snapshot of the running scenario. Provided the Output Configuration File (OCF) which is indicated in Fig. 2, the MSF is also configurable to visualize the ad hoc network on demand. Fig. 12 shows some of the setting for the Bus Node. With those settings, the BNs will be visible and rendered with blue color. The transmission range of the antenna equipped by the BN is set to 150m, but it will not be shown during the animation. The moving path of BNs will be rendered on MSF in pink color. With the help of MSF and OCF, the network can be better understood and studied. B. Network Topology The average number of connectivity changes with increasing speed is shown in Fig. 13. With the existence of Hotspots and NAA, the effective space in the simulation area is highly reduced. In the route environment, the movements of mobile nodes are also restricted. There are 16 hotspots located in the environment. Nodes gathered in one of the hotspots, generates very high connectivity, but this established connectivity can only be maintained for a short time because later they will head for another hotspot. The Obstacles partition the area into several sub-areas so that the possibility of connectivity loss is increased as the nodes move in and out of the sub-areas. It is shown that the Obstacles cause higher connectivity change and the Hotspot model normally have much higher connectivity change than the Random Waypoint mode in which nodes are more distributed throughout the area. An interesting point can be pointed out here is that for the Hotspot submodel the connectivity change without Obstacles is increased sharply when Obstacles are introduced. At higher speed without Obstacles, the frequency of moving in and out is increased and the nodes moving along the routes are acting more likely as communication bridges to connect the nodes of two sub-areas. On the other hand, the connectivity created by those bridge nodes can not be kept for a long time at higher speed, which increases the change. This explanation is also applied to the Random Waypoint case. It can be seen that the two curves are jointed at 9m/s and the without-Obstacles situation has a faster increase. Most of the hotspots are located close to the boundaries which are adjacent to the route, therefore
2
0.115 0.110 0.105 0.100 0.095 0.090 0.085 0.080 0.075 0.070 0.065 0.060 0.055 0.050 0.045 0.040 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000
EAM with Hotspot sub-model with Obstacles EAM with RandomWaypoint with Obstacles EAM with Hotspot sub-model without Obstacles EAM with RandomWaypoint sub-model without Obstacles
0
1
2
3
4
5
6
7
8
9
Average Speed of Mobile Nodes (meters/sec)
Figure 16. Average data packets end-to-end delay.
more nodes in the two sub-areas can be connected and also disconnected by movement of the bridge nodes. This explains why this effect is more significant in the Hotspots environment. C. Routing Performance The data delivery rate measured in different environments is shown in Fig. 14. Since the Obstacles obstruct the transmission, the mobile nodes located in different sub-areas are not communicable if a boundary of Closed Area detected, which obviously deteriorates the delivery rate. The hotspots generate very high density which also causes much higher possibility of collision and packets drop due to the mechanism of the MAC layer of MANET. The delivery rate in both Random Waypoint environments is much more stable than the Hotspot environments. There is a Faster drop-down in the Hotspots model without Obstacles (10% dropdown) than
62
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
with-Obstacles (5% dropdown). This is due to the network topology change as explained in the previous subsection. The number of control packets sent for routing is plotted in Fig. 15. This result is highly correlated to the packet reception showed before. Low delivery rate in the Hotspot environments causes the nodes to send more control packets to discover the routes as the requirement of AODV. The high density of nodes in the Hotspot environment cause frequent transmission collision and contention as shown in Fig. 16. Consequently, the packets are very likely to be delayed until an idle radio channel is found. Due to the higher topology change in the Hotspot without-Obstacles environment, the time spent in route discovery is much longer so that the latency of packets transmission is higher. The random distribution and lower topology change guarantees that the data packets can be sent and arrive at their destinations quickly in the Random Waypoint environment when no obstacles exist. VI.
CONCLUSIONS
This paper discusses the mobility modeling in MANET by introducing a novel mobility model, EnvironmentAware Mobility model. This model is an integration of current existing models to the complex environments where the networks are deployed. With the aid of SVG, the layout of the environment can be easily designed. Some abstract Environment Objects are introduced to represent the various areas existing in the networks. It has been showed that the movements are highly influenced by both intrinsic characteristics and some properties of EOs. The behaviors of the mobile nodes are not only dependent on which mobility models they use but also on the area in which they are located. The signal-obstruction effect is also incorporated in this model with the introduction of the Open or Closed properties of EOs. The signal is assumed to be blocked if the transmission needs to go through any boundary of the Closed EOs. Two sub-models, Route model and Hotspot model are also proposed to model the mobility observed in some particular environments. The characteristics of the movements of mobile nodes are studied and the mobile nodes are classified into five categories. With the concern of the node heterogeneity, the particular ad hoc networks in more complex environments, for example the Busaided IVC system and RoadLamp-aided IVC system can be modeled. The simulations are done with different mobility models given the same geographical layout of network but with different environmental configurations. The scenarios can be visualized on demand in order to study the performance of the network in depth. The average connectivity changes are measured and the throughput, end-to-end delay and delivery rate of AODV routing protocol are evaluated. The results prove that both the network topology and the performance of protocols are significantly influenced by the mobility models and environmental factors.
© 2006 ACADEMY PUBLISHER
With this Environment-Aware Mobility model, more realistic MANET environments can be easily modeled and better understanding of MANET can be achieved. REFERENCES [1] T. Camp, J. Boleng, and V. Davies, “A Survey of Mobility Models for Ad Hoc Network Research,” in Wireless Communication and Mobile Computing (WCMC): Special issue on Mobile Ad Hoc Networking: Research, Trends and Applications, vol. 2, no. 5, pp. 483-502, 2002. [2] V. Tolety. Load reduction in ad hoc networks using mobile servers. Master’s thesis, Colorado School of Mines, 1999. [3] X. Hong, M. Gerla, G.Pei, and C. Chiang. “A group mobility model for ad hoc wireless networks,” In Proceedings of the ACM International Workshop on Modeling and Simulation of Wireless and Mobile Systems (MSWiM), August 1999. [4] The Network Simulator 2 .http://www.isi.edu/nsnam/ns. [5] Scalable Vector Graphics. http://www.w3.org/TR/SVG/ [6] B.Liang, Z. J. Haas, “Predictive Distance-Based Mobility Management for PCS Networks,” in Proceedings of IEEE Information Communications Conference (INFOCOM 1999), Apr. 1999. [7] J.Tian, J. Hahner, C. Becker, I. Stepanov and K. Rothermel, “Graph-based Mobility Model for Mobile Ad Hoc Network Simulation,” in the Proceedings of 35th Annual Simulation Symposium, in cooperation with the IEEE Computer Society and ACM. San Diego, California. April 2002. [8] P. Johansson, T. Larsson, N. Hedman, B. Mielczarek, and M. Degermark, “Scenario-based performance analysis of routing protocols for mobile ad-hoc networks,” in International Conference on Mobile Computing and Networking (MobiCom'99), 1999, pp. 195-206. [9] F. Bai, N.Sadagopan, and A. Helmy, “Important: a framework to systematically analyze the impact of mobility on performance of routing protocols for ad hoc networks,” in Proceedings of IEEE Information Communications Conference (INFOCOM 2003), San Francisco, Apr. 2003. [10] Subodh Shah, Edwin Hernandez, Abdelsalam Helal, “CAD-HOC: A CAD-Like Tool for Generating Mobility Benchmarks in Ad-Hoc Networks,” SAINT 2002: 270-280. [11] C. E. Perkins, E. Royer, and S. R. Das, “Ad hoc On Demand Distance Vector (AODV) Routing,” In 2nd IEEE WorkShop on Mobile Computing Systems and Applications (WMCSA'99), pages 90-100, February 1999. [12] A.P. Jardosh, E.M.Belding-Royer, K.C. Almeroth, and S. Suri, “Towards Realistic Mobility Models for Mobile Ad hoc Networks,” In Proceedings of ACM MOBICOM, pages 217-229, San Diego, CA, September 2003. [13] Gang Lu, Demetrios Belis, Gordon Manson, “Study on Environment Mobility Models for Mobile Ad Hoc Network: Hotspot Mobility Model and Route Mobility Model”, WirelessCom2005, Hawaii, USA.
Gang Lu was born in Chengdu, China in 1978. He got his BSc with first class degree from East China Normal University in information and communications system in 2000. He came to the UK in 2001 and gained the MSc in data communications from the University of Sheffield in 2002. He started his PhD in mobile ad hoc networks from 2002 in the Center for Mobile Communications Research Group.
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
His major research interests are self-positioning, locationbased routing, mobility modeling in mobile ad hoc networks, mobile ad hoc networks emulation and wireless sensor networks. He is also a lecturer of IT courses in the Institute for Lifelong Learning, the University of Sheffield, UK. He joined R&D department of Dialogue Communications Ltd. in 2003, UK. His current work role is the major researcher and developer of multimedia messaging and 3G applications. Mr. Lu is the translator of the book Data Networks (2nd Edition, by D. Bertsekas and R.Gallager, Chinese version), ISBN 7-115-12292-X/TP, Posts & Telecom Press, 2004. He is also the main author of the ongoing book Wireless Sensor Networks which is planned to be published in China in end of 2006.
Gordon Manson was born in Glasgow, Scotland in 1949. He was educated in Glasgow and in Irvine Ayrshire where he attended Irvine Royal Academy. He graduated from Strathclyde University in Glasgow with a 1st class honours degree in mathematics in 1971 and gained a PhD in applied mathematics in 1977. He lectured for one year at Strathclyde University and joined the Applied Mathematics department in the University of Sheffield in 1975. In 1982 he changed to the Computer Science department at Sheffield and he gains an MSc in Computer Science from the University of Manchester in 1984. He became a senior lecture in 1992 and this is his current position. His main research area was parallel processing using Transputers and he was the Assistant Director of the National Transputer Support Centre from 1987-95. Currently he is interested in applying ad-hoc techniques in swarming intelligent UAVs. Dr. Manson has published over 100 papers and been a coauthor in 2 books and a co-grant holder of 16 grants funded by industry, EPSRC and the EU. He was jointly awarded the best paper in Journal of Information Management and Computer Security in 2002.
Demetrios Belis was born in Piraeus, Greece. He graduated from the Technological Educational Institute of Piraeus in Greece with a degree in Automatic Control Systems in 1997. He obtained an MSc (Eng) in Automatic Control and Systems Engineering in 2000 and a PhD in Computer Science in 2006, both at the University of Sheffield, United Kingdom. His is specialized in the areas of Artificial Life and telecommunications, particularly in the fusion of ad hoc networks and autonomous agents. Currently he is working for learndirect (University for Industry, Ufi Ltd) at Sheffield, UK. He is also a lecturer of IT courses at the Institute for Lifelong Learning, University of Sheffield, UK.
© 2006 ACADEMY PUBLISHER
63
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Mobility Modeling in Mobile Ad Hoc Networks with Environment-Aware Gang Lu Department of Computer Science, University of Sheffield, UK [email protected]
Gordon Manson and Demetrios Belis Department of Computer Science, University of Sheffield, UK { g.manson, dimi }@dcs.sheffield.ac.uk
Abstract—Simulation is the most important and widely used method in the research of Mobile Ad hoc NETworks (MANET). The topology of MANET and the mobility of mobile nodes are the key factors that have an impact on the performance of protocols. However, most of the existing works are based on random movement and the fact that the network topology is highly related to the environment of MANET is overlooked. In this paper, we propose a novel Environment-Aware Mobility (EAM) model which models a more realistic movement of mobile nodes. Environment objects such as Route and Hotspot are introduced to represent the environment components which are rendered by Scalable Vector Graphics (SVG). It is considered to be a complex model with a combination of existing conventional mobility models and network environments. This paper shows that various MANET environments can be modeled based on this work. A sample environment is also simulated and the results show that the intrinsic characteristics and properties of the environments have a significant influence on the performance of MANET protocols. Index Terms—Mobility Modeling, Environment-Aware Mobility Model, Routing Performance, MANET.
I.
INTRODUCTION
Without any central administration, MANET is considered to be the dominating form of wireless network and has drawn more and more attraction from the industry. Using self-organised mobile nodes, MANET can be deployed in any environment, such as a gallery, a theatre, a shopping mall and even a street. Many researchers have shown interest in the fields of MANET and all sorts of protocols aiming at different issues have been proposed in recent years. The performance of these protocols needs to be carefully evaluated before they are ready fore the commercial market. Therefore, network simulation plays a very important role in this area, and is widely used by researchers as a key method to better Based on “A Novel Environment-Aware Mobility Model for Mobile Ad Hoc Networks” by Gang Lu, Demetrios Belis, and Gordon Manson which appeared in the Proceedings of the 1st International Conference MSN 2005, pp. 654-665, Wuhan, China, Dec. 13-15.
© 2006 ACADEMY PUBLISHER
understand the overall performance of MANET. Self-organization is the essential characteristic of this mobile wireless network. The mobile nodes within the network can be any moving or fix-position objects equipped with antennas. They can be either humans moving in a mall or gallery, or insects and animals in nature environments. The movement behaviors are highly dependent on their own mobility factors and the areas they are located. The mobility model is therefore regarded as an important component in ad hoc network simulation. In this paper, the mobility modeling is discussed by proposing a novel realistic model which is incorporated with the environments where a MANET is deployed. Environments are defined by introducing some environment objects, such as a Route, a Junction, an Accessible Area, and a Hotspot etc. The mobile nodes are modeled with the concern of node heterogeneity. The signal-blocking issues are also covered in this model by introducing the ClosedArea whose boundaries are considered to be obstacles with no radio penetration. In addition, with the use of Scalable Vector Graphics (SVG) [5], environments can be easily designed and the movements in the environment can be visually observed. The remainder of this paper is organized as follows. Section II details some related work in the research of mobility models. Section III describes the detailed information about how the Environment-Aware Mobility model works. The details of the evaluation are presented in Section IV, and the results are presented and discussed in the last section. II.
RELATED WORKS
There exists many mobility models proposed for mobile wireless networks. Mobility models are used to describe the movement pattern of mobile nodes. In this section, some of the popular models used in MANET are briefly described. The relevancy between EAM and those models is clarified afterwards. The category of conventional models is illustrated in Fig. 1. The Entity Models are used to model the movement behavior of an individual mobile node. The
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 1. Category of Mobility Models.
Random Waypoint model [1] is widely implemented by network simulators such as ns2 [4]. In each movement epoch, the mobile node picks a position within the simulation area and moves towards it with a speed distributed in the range [νmin, νmax]. Instead of moving to the next destination immediately as soon as it arrives at the current destination, the mobile node pauses for a specified time then repeats the procedure. The major drawback is that the nodes tend to move around the centre area so that they are not really distributed into the entire simulation area. The Random Direction [1] model is proposed to avoid this distribution problem. It forces the mobile node to move until it reaches the border of area before it starts its next epoch. The models introduced previously are all considered to be memory-less models since the next movement segment has no dependency on the previous movement regarding either speed or direction. This memory-less feature causes frequent sharp change in speed and direction of movement which is obviously not applicable in a realistic world. Some memory models are proposed. The Gauss-Markov model was first introduced in [6]. In this model, the velocity of the mobile node is assumed to be correlated over time and modeled as a Gauss-Markov stochastic process. The Boundless model [1] is another example of a memory model but with no geographical restriction. Mobile nodes are allowed to cross the boundaries and appear at the other side of the area. The resulting effect is that the simulation area is modeled as a torus instead of a flat area. In ad hoc networks, there are many situations where the movements of mobile nodes have some correlation with each other, i.e. mobile nodes have some group behavior in common. To present this characteristic, Group Models have been proposed. One typical example is the Reference Point Group Mobility model (RPGM) [3]. In RPGM, a logical centre of a group is defined and its movement is used to direct the group-wide movement. Individual members of the group move not purely on a random basis, their movements are also affected by the group movement. RPGM is popularly used in research to depict some scenarios with group behaviors such as avalanche rescue. Other group models can also be found in [1]. The previous models all assume that the simulation area is a free space area where mobile nodes can move anywhere inside. They expose the self-organization feature of mobile nodes, but they are not generally applicable and the geographic factors have to be considered. The Pathway models [7] and the Obstacle models [8] partially overcome this disadvantage. The Pathway Model forces each mobile node to move along the shortest path towards its destination. Similar behavior © 2006 ACADEMY PUBLISHER
55
is also modeled in the Freeway mobility model and in the Manhattan mobility model in [9]. The Obstacle mobility model was first introduced in [12]. Unlike the Pathway model, the Obstacle model defines some obstacles in the simulation area. These obstacles have a signal-blocking effect on the communication of mobile nodes. This model also allows the nodes to change this movement trajectory when obstacles are encountered. Geographic models support more realistic scenarios than Entity and Group models. Several works have brought the idea of real-world simulation into MANET simulations. The realistic environment in which the network exists needs to be constructed before running the simulation. Ref. [10] provides a way to achieve this by using Auto-CAD, and the realistic world is constructed by implementing the knowledge of the Voronoi diagram in [12]. The movements of mobile nodes are restricted by the predefined arbitrary obstacles and pathways. However, this real-world environment model does not provide a flexible approach to model the mobility variations. The movement trajectory of a mobile node might change over time. For example, the mobile nodes move along the pathway then may enter a particular terrain, such as a park where their movements become less deterministic, therefore they might walk in any direction, i.e. the pathway restraints are no longer applied. As a conclusion, every model mentioned above has some abilities to model the mobility of mobile node in MANET. They can be applied to particular situations, but none of them is flexible and suitable enough for modeling more realistic scenarios. As a mater of fact, Entity Models and Group Models can co-exist in some scenarios with obstacles. The Obstacle model seems to give a good solution for signal-blocking but it leaves entity and group mobility properties unconcerned. Obstacles and Paths are detected but the trajectories to deal with obstacles are simple and not general. Depend on the roles they play in the environment, the mobile nodes can have different mobile characteristics. The fact that environment factors can affect the movement is also overlooked by all of them. By taking the environment into account as illustrated in Fig. 1, the EAM model proposed in this paper provides a more general approach to model more realistic ad hoc networks. III.
ENVIRONMENT-AWARE MOBILITY MODEL
The EAM model proposed in this paper is designed to model the movement behavior of mobile nodes in the environments of realistic ad hoc networks. By studying the possible environment where MANET is located, different sub-areas within the entire simulation area are abstracted to several environment objects, such as a Route, Junction, Hotspot, etc. The movement trajectory of the mobile node is correlated with the sub-area that it is located and is allowed to be changed during the simulation. The node heterogeneity is also concerned for better describing the mobility of mobile nodes. The obstruction of radio propagation is also implemented in this model, and some of the conventional models are
56
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
TABLE I.
DESCRIPTION OF ENVIRONMENT OBJECTS
Name
NORAA (N) Figure 2. Workflow for mobility modeling using Environment-Aware Mobility Model. Accessible Area (AA)
Lane (L)
Path (P)
Figure 3. Structure of Environment Objects (EOs).
A. Environment Objects Different types of sub-areas are abstracted by Environment Objects (EOs). The EOs can be classified into two categories: Non-Accessible Area (NAA) and Accessible Area (AA). NAA represents the restricted area where no movement is allowed. AA represents some areas where mobile nodes can move inside and may move in and out. An AA can be any of the Lane, Path, Route, Junction, NORmal Accessible-Area (NORAA) and Hotspot objects. Some hierarchical relationships also exist in EOs. For instance, a Route is composed of Paths, Junctions and even NORAAs, a Path can be composed of Lanes, and NORAAs can be a container area for Hotspots. 0Fig. 3 illustrates the structure of EOs. With the introduction of EOs, real complex environments can be simplified and easily constructed.
© 2006 ACADEMY PUBLISHER
Route (R) Hotspot (H) Non-Accessible Area (NAA)
modified to take into account the environmental effects and to cope with the obstacles. In our proposal, the environment including all the environment objects is designed with SVG. The relationships between sub-areas and various properties of each sub-area are given by a text-formatted configuration file (Environment Configuration File). A SVG file is generated after the simulation is finished. This animationsupported SVG file can be very useful for researchers to explicitly observe the movements in the simulated scenario. The details shown in this SVG file can be configured by the users by modifying a text-formatted configuration file (Output Configuration File). When the simulation starts, this model reads all three files to become aware of where the environment objects are and their properties. The mobile nodes are distributed into the accessible sub-areas. The mobility factors such as speed, direction and the mobility model attached to each node are determined by the properties of the areas they are located. All the movements are recorded and converted to ns2 compatible movement scenario file at the end of the simulation. ns2 is also modified in order to be able to read an obstacle boundaries description file to block radio signal. Fig. 2 illustrates the work flow of this model.
Junction (J)
NAA
Description If there are no sub-areas located inside the NORAA, the mobile nodes can move freely inside its area; However, a NORAA can contains Hotspot and NAA, in which case the behaviors of mobile nodes will be affected and restricted. One pair of Entry and Exit must be defined for a Lane. Once the mobile node moves into a Lane from the Entry, the node has to move towards the Exit. A Path can consist of one Lane or multiple parallel Lanes. The mobile nodes moving in a Path are not allowed to change its Lane and they must move from the Entry towards the Exit of the Lane where it is located. It is used to connect two Paths, or two NORAA, or one Path and NORAA. It is the resulting EO if the Path, Junction and Route are joined together. It is used to represent the attracting area where the mobile nodes may visit frequently. It is the area where no mobile node can move inside. The boundary effect will occur if the mobile node encounters a NAA in the way.
Figure 4. An example snippet of Environment Layout File.
Each EO is given some intrinsic characterises which have influence on the mobility of the mobile nodes. Typically, the mobile node located in the Lane must follow the Lane to the exit with a maximum speed limit; if a Junction is encountered, a mobile node is forced to choose one of the exits to enter the next adjacent EO. NORAA is a very flexible EO because it can be used as a container area. If it is a free space area, a mobile node can move using its conventional mobility model. If it contains Hotspots, mobile nodes will be forced to commute among the Hotspots. If it contains a NAA, the accessible space for
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
57
TABLE II. THE PROPERTIES OF ENVIRONMENT OBJECTS TABLE III. THE ASSOCIATIONS OF THE ENVIRONMENT OBJECTS AND THE PROPERTIES
Properties
Hotspot
Junction
Lane
Path
Route
Value
The maximum speed at which the mobile nodes can move when they are located in this area. The speed in meter/second (m/s). Any speed not less than 0.0m/s. Example: 3.5(m/s)
NORAA
Description
NAA
MaxSpeed
MaxDirection Description Value
It is the maximum direction in degree towards which the mobile nodes can move when they are located in this area. The angle in degree, ranges from -180.0 to 180.0 degree. Example: 180 (degree)
MinDirection Description Value
It is the minimum direction in degree towards which the mobile nodes can move when they are located in this area. The direction in degree, ranges from -180.0 to 180.0 degree. Example: -180 (degree)
InboundAreas Description Value
It is used to specify the entry areas where the mobile nodes can move into this area from. The name of the EO. If more than one area exist, the names of EOs must be separated by ‘’. Example: J2-A1
MaxSpeed MaxDirection MinDirection InboundAreas OutboundAreas Capacity MaxDuration Components ClosedArea MobilityModel
√ √ √ √ √ √ √ √ √ √
X X X X √ √ X X X X X √ X X X X X √ X X X √ √ √ X X X √ √ √ X X X X X √ X X X X X √ X X X X √ √ X X X X X √ X X X X X √ √ = the property is applicable to the EO X = the property is NOT applicable to the EO
OutboundAreas Description
Value
It is used to specify the exit areas where the mobile nodes can move into from the area it is located. The name of the EO. If more than one area exist, the names of EOs must be separated by ‘’. Example: J1-A2
Figure 5. Typical configurations for a Normal-Accessible Area.
Capacity Description Value
The maximum number of mobile nodes the area can accommodate Any positive integer. Example: 10
MaxDuration Description Value
The maximum duration of time the mobile nodes can stay inside the area The duration in second (s). Example: 40 (s)
Components Description
Value
It is used to specify the sub-areas located inside this configured area or the areas that it consists of. The name of the EO. If more than one area exist, the names of Eos must be separated by ‘’. Example: H1-H2
ClosedArea Description Value
It is used to indicate if this area is considered as a signal-blocking area. If this is set to be true, its boundaries will be regarded as obstacles. Boolean value, True/False. Example: true
MobilityModel Description
Value
It is used to specify which mobility model the mobile nodes should use once they move into the area. The name of mobility model. The mobility model can be any one of Random Walk, Random Waypoint, Random Direction, GaussMarkov, Boundless, Hotspot and RPGM. Example: randomwalk
the mobile nodes is reduced. The full description of the EOs is given in TABLE I.
© 2006 ACADEMY PUBLISHER
B. Environment Layout Design Scalable Vector Graphic (SVG) has been widely accepted as a graphic standard, and it is the key approach used by the Environment-Aware Model to produce the simulation environment. SVG uses XML to describe 2dimention graphics. SVG contains a set of basic shape elements, such as the rectangle, circle, line, and polygon. These elements are used to represent the environment objects with arbitrary shapes. SVG also supports the abilities to change the vector graphic over time which provides the capability of visualizing the movements of mobile nodes. The full specification of SVG can be found in [5]. A typical snippet of an SVG Environment Layout File (ELF) is attached in Fig. 4. Note that every Environment Object is identified by the ID of the SVG shape element. A simple convention is set: the ID must start with AA (for Accessible Area) or NAA (for Non-Accessible Area) then followed by an underscore then its name which is composed by the object’s type plus the index. For example, if an element has an ID of “AA_L1” (‘L1’ is the name of this object), it means the sub-area represented by this element is an accessible Lane with index 1. C. Environment Configuration The SVG layout file provides the model with the location of each EO and helps identify their types. The relationships among EOs mentioned along with their properties are configured in a text-format Environment Configuration File (ECF). It is also assumed in EAM that
58
some of the EOs might have some properties in common. They may have restraints on the speed, the direction and capacity which is the maximum number of mobile nodes the EO can accommodate. TABLE. II gives the full description of all properties defined in the EAM model and TABLE. III shows the association of the properties and the EOs. All these properties are configurable in the ECF. As an example, a typical configuration for the NormalAccessible Area with ID=AA_A1 which exists in the environment of Fig.4 is given in Fig. 5. Referring to the description in TABLE II, it is very easy to derive that the area A1 is adjacent to Junction J2 which is its entry area. A1 can be considered as a room since it is set to be a Close area, it contains three Hotspots (H1, H2 and H3) and maximum 10 mobile nodes are allowed to stay in at the same time. As for the mobility factors, the mobile nodes located inside can move in any direction but with speed up to 7m/s, and the general movements are modeled using the Hotspot Model. D. Environment-Aware Movement The EAM can integrate many Entity Models such as the Random Waypoint model, the Boundless model and the Group Models like RPGM with some modifications. In order to cope with the complex environments composed by the EOs, two sub-models are proposed: the Route model and the Hotspot model. These two models are described in detail and intensively studied in [13]. Boundary Effects are implemented to cope with the situation when NAAs are encountered. y Movement in the Route model: The Route model can model some complex environments, such as a city area or an urban road traffic network. The mobile nodes moving in a Route travel through all its components in a specified order given in the ECF file. When they reach the end of a Route, the mobile nodes will appear at the entry of a new Route in order to keep the number of nodes constant during the simulation. The Route Mobility model can be used to model the movement described in the Pathway model, the Manhattan model, the Virtual Track model, and the Obstacle model. Nevertheless here the movement in a route can use the Random Walk model as well since the Route model allows the existence of a NORAA to bridge two Paths. Movements can also be restrained by the route. For instance, the speed may not exceed the maximum allowed speed of a lane. y Movement in the Hotspot model: A hotspot is a place where most of the mobile nodes are likely to visit. If hotspots exist in an area, the strategies with which the mobile nodes move between the hotspots can be varied because a hotspot might have a capacity to accommodate a maximum number of visitors and a maximum duration that the mobile nodes are allowed to stay. A number of different strategies are introduced in [13]. In this paper, a simple strategy is applied: the mobile nodes always adjust their speed to make sure that the hotspot can allow at least one more nodes to be added when it arrives at the hotspot.
© 2006 ACADEMY PUBLISHER
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 6. Boundary Effects applied to different models.
Figure 7. Example of Obstacles Description File.
y Boundary Effects As mentioned earlier, various models are modified and given environment-aware capability. For example, different behaviors are designed for the Random Walk model and the Random Waypoint model when a NAA is encountered. Bouncing behavior is applied to the Random Walk model. Fig. 6(a) illustrates the effect: the node bounces to another random direction at boundary e1. Compared with the Random Walk model, the Random Waypoint model features a target-driven behavior so that Bouncing is not applicable. Thus, a different boundary effect called Surrounding is proposed. With the Surrounding behavior, once the next destination is determined, the mobile node will try to get there even if there are obstacles blocking the expected path. Fig. 6(b) illustrates this effect: the node starting at S moves towards D, it actually travels along the boundaries e3-e4e1, and then gets to D. E. Signal Obstruction Radio propagation could be one of the major issues that influence the performance of MANET. The received signal is very vulnerable to suffer significant power dropdown due to attenuation, multipath fading, etc. The TwoRay Pathloss propagation model is used to calculate the power of the received signal. In EAM, it is assumed that if there is a boundary of an obstacle geographically located between the LOS range of the pair of mobile nodes, then the signal transmitted from either of them will not reach the other node. OpenAreas and ClosedAreas are defined to help the model identify which EO should be considered as an
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 8. Illustration of Bus-aided IVC System.
Figure 9. Illustration of RoadLamp-aided IVC System.
Obstacle. A text-formatted Obstacle Description File (ODF) is generated for the environment. If one EO is configured to be a ClosedArea, Its boundaries locations are added into the ODF which will be processed in the physical layer of ns2. Fig. 7 gives an example environment and its relevant ODF. F. Node Heterogeneity The mobile nodes can have different behavior depend on the roles they play in the environment. In Environment-Aware Mobility mode, the mobile nodes are classified into five categories based on their mobile characteristics. They include High Mobility Node, Low Mobility Node, Restricted Nodes, Bus Node, and Static Node. A High Mobility Node (HMN) is the most common node. The nodes belong to this category normally can have relatively higher range of speed compared to other nodes. Another important characteristic is that all the HMNs have certain randomness. Every node is moving towards a random destination through one of the available routes; they can have different random speed, but they must have more or less the same average speed depend on the traffic load. One typical High Mobility Node is the car. A Low Mobility Node (LMN) is actually a special case of the Normal Node. The nodes belonging to this category have the same level of randomness as Normal Node, but they usually are moving with much slower speed. A typical example of this category is the pedestrians.
© 2006 ACADEMY PUBLISHER
59
A Restricted Node (RN) is normal located inside certain obstacles such as a building, a stadium and so on. The nodes belonging to this category normally exhibit lowest speeds inside a relatively small sub-area of the environment. It is noted that the behavior can be changed to be a Restricted Node once the High Mobility Node and Low Mobility Node move into a restricted area. A visitor to an Art Gallery, for example, can be considered as a good representative of this category. A Bus Node (BN) is defined to represent mobile nodes such as a bus or supertram. The nodes belonging to this category normally have the middle speed between the LMN and HMN, but more importantly, the node’s average speed is relatively constant in order to catch up with its timetables and it is normally moving in a loop along a pre-defined route and pauses for a certain time at some locations along the route. A Static Node (SN) is defined to represent the nodes which are not moving during the observation period. The speed of a Static Node is considered as 0m/s. The traffic lights located at the junctions and the roadlamps located along the routes can be treated as typical examples, assuming they are equipped with radio transceivers. With the consideration of the node heterogeneity, some particular mobile ad hoc networks can be simulated in various environments, for example the so-called Busaided Inter-Vehicle Communication (Bus-aided IVC) system as illustrated in Fig. 8. It takes the advantage of the regular movement of the bus helping the connection of two physically out-of-communication-range vehicles; The RoadLamp-aided IVC system shown in Fig. 9 is another particular communication system where some of the roadlamps or traffic lights are assumed to be equipped with wireless transponders and serve as the access points. IV.
SIMULATIONS
The Environment-Aware Mobility model is very flexible and can be applied to various environments. The main purpose of the simulation is to show that the performance of the ad hoc networks protocols is not only influenced by mobility models, but also significantly affected by the Environment Objects which are the fundamental sub-environments abstracted from the real environments. With the presence of EOs, the movement behaviors of mobile nodes modeled by the mobility model are also restricted by the properties of the EOs. Moreover, by introducing Obstacles, the network connectivity is also highly influenced. The resulting topology change is evaluated by the measurement of the average numbers of connection changes at different speeds. One popular routing protocol AODV [11] is simulated. Three typical metrics, Control Packet Overhead, End-toEnd Delay and Data Packet Delivery Rate are used to evaluate the routing performance. For comparison, all the metrics are measured repeatly with increasing speed from 0m/s to 9m/s. In order to show the impact caused by the environment, two ECF files are used and simulated individually. Both the Obstacle and Obstacles-less environment are
60
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
Figure 10. Environment Layout of the simulation.
Figure 11. Snapshot of the running scenario.
simulated and compared. The SVG Layout is shown in Fig. 10. A. Environment Description A 600x300m area is used to represent the simulation area. Two different MANET environments, The Art Gallery and Festival Event are used. The Art Gallery: In this environment, Area 1 and Area 2 can be considered to be two rooms separated by a wall (NAA). All the arts exhibited in the rooms are presented by Hotspots. Visitors enter this area from the entry. They may continue moving to the next exhibition section through Exit 1 or enter one room (Area 1 or Area 2). Once the visitors enter the room, they will move towards the art object that they find most attractive then the next and so on. If a visitor notices that his art object has too many visitors (up to the capacity of Hotspot), they will go to the next most interesting object. Once they have visited all the art objects or if they have been in the room too long (up to maximum duration allowed to stay in the area), they will leave the room and continue the visit through Exit 2 or Exit 3 as shown in Fig. 10. In this case, both areas are ClosedArea because the signal is very likely to be blocked by the walls. The Festival Event: Compared to the Art Gallery environment, this environment exists in some public places, such as a park so that the two areas can be assumed to be OpenAreas. In this case, when mobile nodes move into Area 1 or Area 2, they may just walk around freely for a while then leave. Note that if some entertainments or shops are distributed inside the areas as hotspots, the behaviors of nodes could be similar to the previous one. B. Simulation Description All the simulations are done by ns2. 40 mobile nodes with 100m omni-directional antenna transmission range are distributed over the simulation area. The Hotspot model is used for the modeling the Art Gallery environment and the random movements in the Festival Event are modeled by the Random Waypoint model. The simulation is repeated with different average nodes speeds ranging from 0m/s to 9m/s and the final result is an average of the simulation results from 40 runs. To evaluate the impact of Obstacles on different mobility models, both environments are simulated with Obstacles and without Obstacles. Area 1 and Area 2 are
© 2006 ACADEMY PUBLISHER
Figure 12. Setting for Bus Nodes in Output Configuration File (OCF).
assumed to contain minimum of 15 mobile nodes inside, and the other 10 nodes are distributed randomly along the routes. If any of the 10 nodes are located inside the two sub-areas, they will stay there for a certain time then leave. If they are located in Lanes, they will randomly choose a route. For the Random Waypoint model, a 10 seconds pause time is spent by each node when it gets to the current destination and the pause time in the Hotspot model is depend on the property of each Hotspot. In the Art Gallery environment, nodes move about with a maximum speed of 7m/s inside the two rooms, but there is no limitation for the Festival Event because it is considered to be an out-door event. In order to keep the number of mobile nodes constant during the simulation, the nodes that arrive at any of the Exits will appear at the Entry to start a new journey. Each scenario is simulated for 500 seconds. The measurement is started after 50 seconds to ensure all the nodes have been distributed throughout the areas. The number of connectivity between neighbors is calculated every second. The average changes of connectivity at each speed are measured as the metric to evaluate the influence on the network topology caused by the mobility models and the environments. Environment Layout AODV routing protocol is used to evaluate the impact on the routing performance. The CBR data packet size is set to 512 bytes and the sending rate is set to 4 packets per second. A maximum of 20 data connection is allowed at one time. A node will act as either a sender or receiver at any time during the simulation. V.
RESULTS
A. Scenario Visualization By utilizing the animation elements of SVG [5], the simulated scenarios can be visualized with animation by any SVG-viewer or SVG-supported web browser. Every
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
61
Data Packets Delivery Rate in Environment-Aware Model
Number of Connectivity Changes in Environment-Aware Model 1.000
9.0
EAM with Route sub-model (20nodes) with Obstacles EAM with Route sub-model (10nodes) with Obstacles EAM with Route sub-model (20nodes) without Obstacles EAM with Route sub-model (10nodes) without Obstacles
8.6 8.2
0.970 0.955
Data Packets Delivery Rate (AODV)
Number of Connectivity Changes
7.8
0.985
7.4 7.0 6.6 6.2 5.8 5.4 5.0 4.6 4.2 3.8 3.4 3.0 2.6 2.2
0.940 0.925 0.910 0.895 0.880 0.865 0.850 0.835 0.820 0.805 0.790 0.775 0.760 0.745
1.8
EAM with Hotspot sub-model with Obstacles EAM with RandomWaypoint with Obstacles EAM with Hotspot sub-model without Obstacles EAM with RandomWaypoint without Obstacle
0.730
1.4
0.715
1.0 0
1
2
3
4
5
6
7
8
0.700
9
0
Average Speed of Mobile Nodes (meters/sec)
1
© 2006 ACADEMY PUBLISHER
3
4
5
6
7
8
9
Average Speed of Mobile Nodes (meters/sec)
Figure 13. Average number of connectivity changes.
Figure 14. Average Data packets delivery rate.
Number Of Control Packets (AODV)
Number Of Control Packets in Environment-Aware Model 22200 21500 20800 20100 19400 18700 18000 17300 16600 15900 15200 14500 13800 13100 12400 11700 11000 10300 9600 8900 8200 7500 6800 6100 5400 4700 4000
EAM with Hotspot sub-model with Obstacles EAM with RandomWaypoint with Obstacles EAM with Hotspot sub-model without Obstacles EAM with RandomWaypoint withoutObstacles
0
1
2
3
4
5
6
7
8
9
Average Speed of Mobile Nodes (meters/sec)
Figure 15. Average number of control packets sent. Data Packets End-to-End Delay in Environment-Aware Model
Data Packets End-to-End Delay (AODV)(sec)
movement of each mobile node is recorded into the Mobility Scenario File (MSF) in SVG format. Fig. 11 shows a snapshot of the running scenario. Provided the Output Configuration File (OCF) which is indicated in Fig. 2, the MSF is also configurable to visualize the ad hoc network on demand. Fig. 12 shows some of the setting for the Bus Node. With those settings, the BNs will be visible and rendered with blue color. The transmission range of the antenna equipped by the BN is set to 150m, but it will not be shown during the animation. The moving path of BNs will be rendered on MSF in pink color. With the help of MSF and OCF, the network can be better understood and studied. B. Network Topology The average number of connectivity changes with increasing speed is shown in Fig. 13. With the existence of Hotspots and NAA, the effective space in the simulation area is highly reduced. In the route environment, the movements of mobile nodes are also restricted. There are 16 hotspots located in the environment. Nodes gathered in one of the hotspots, generates very high connectivity, but this established connectivity can only be maintained for a short time because later they will head for another hotspot. The Obstacles partition the area into several sub-areas so that the possibility of connectivity loss is increased as the nodes move in and out of the sub-areas. It is shown that the Obstacles cause higher connectivity change and the Hotspot model normally have much higher connectivity change than the Random Waypoint mode in which nodes are more distributed throughout the area. An interesting point can be pointed out here is that for the Hotspot submodel the connectivity change without Obstacles is increased sharply when Obstacles are introduced. At higher speed without Obstacles, the frequency of moving in and out is increased and the nodes moving along the routes are acting more likely as communication bridges to connect the nodes of two sub-areas. On the other hand, the connectivity created by those bridge nodes can not be kept for a long time at higher speed, which increases the change. This explanation is also applied to the Random Waypoint case. It can be seen that the two curves are jointed at 9m/s and the without-Obstacles situation has a faster increase. Most of the hotspots are located close to the boundaries which are adjacent to the route, therefore
2
0.115 0.110 0.105 0.100 0.095 0.090 0.085 0.080 0.075 0.070 0.065 0.060 0.055 0.050 0.045 0.040 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000
EAM with Hotspot sub-model with Obstacles EAM with RandomWaypoint with Obstacles EAM with Hotspot sub-model without Obstacles EAM with RandomWaypoint sub-model without Obstacles
0
1
2
3
4
5
6
7
8
9
Average Speed of Mobile Nodes (meters/sec)
Figure 16. Average data packets end-to-end delay.
more nodes in the two sub-areas can be connected and also disconnected by movement of the bridge nodes. This explains why this effect is more significant in the Hotspots environment. C. Routing Performance The data delivery rate measured in different environments is shown in Fig. 14. Since the Obstacles obstruct the transmission, the mobile nodes located in different sub-areas are not communicable if a boundary of Closed Area detected, which obviously deteriorates the delivery rate. The hotspots generate very high density which also causes much higher possibility of collision and packets drop due to the mechanism of the MAC layer of MANET. The delivery rate in both Random Waypoint environments is much more stable than the Hotspot environments. There is a Faster drop-down in the Hotspots model without Obstacles (10% dropdown) than
62
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
with-Obstacles (5% dropdown). This is due to the network topology change as explained in the previous subsection. The number of control packets sent for routing is plotted in Fig. 15. This result is highly correlated to the packet reception showed before. Low delivery rate in the Hotspot environments causes the nodes to send more control packets to discover the routes as the requirement of AODV. The high density of nodes in the Hotspot environment cause frequent transmission collision and contention as shown in Fig. 16. Consequently, the packets are very likely to be delayed until an idle radio channel is found. Due to the higher topology change in the Hotspot without-Obstacles environment, the time spent in route discovery is much longer so that the latency of packets transmission is higher. The random distribution and lower topology change guarantees that the data packets can be sent and arrive at their destinations quickly in the Random Waypoint environment when no obstacles exist. VI.
CONCLUSIONS
This paper discusses the mobility modeling in MANET by introducing a novel mobility model, EnvironmentAware Mobility model. This model is an integration of current existing models to the complex environments where the networks are deployed. With the aid of SVG, the layout of the environment can be easily designed. Some abstract Environment Objects are introduced to represent the various areas existing in the networks. It has been showed that the movements are highly influenced by both intrinsic characteristics and some properties of EOs. The behaviors of the mobile nodes are not only dependent on which mobility models they use but also on the area in which they are located. The signal-obstruction effect is also incorporated in this model with the introduction of the Open or Closed properties of EOs. The signal is assumed to be blocked if the transmission needs to go through any boundary of the Closed EOs. Two sub-models, Route model and Hotspot model are also proposed to model the mobility observed in some particular environments. The characteristics of the movements of mobile nodes are studied and the mobile nodes are classified into five categories. With the concern of the node heterogeneity, the particular ad hoc networks in more complex environments, for example the Busaided IVC system and RoadLamp-aided IVC system can be modeled. The simulations are done with different mobility models given the same geographical layout of network but with different environmental configurations. The scenarios can be visualized on demand in order to study the performance of the network in depth. The average connectivity changes are measured and the throughput, end-to-end delay and delivery rate of AODV routing protocol are evaluated. The results prove that both the network topology and the performance of protocols are significantly influenced by the mobility models and environmental factors.
© 2006 ACADEMY PUBLISHER
With this Environment-Aware Mobility model, more realistic MANET environments can be easily modeled and better understanding of MANET can be achieved. REFERENCES [1] T. Camp, J. Boleng, and V. Davies, “A Survey of Mobility Models for Ad Hoc Network Research,” in Wireless Communication and Mobile Computing (WCMC): Special issue on Mobile Ad Hoc Networking: Research, Trends and Applications, vol. 2, no. 5, pp. 483-502, 2002. [2] V. Tolety. Load reduction in ad hoc networks using mobile servers. Master’s thesis, Colorado School of Mines, 1999. [3] X. Hong, M. Gerla, G.Pei, and C. Chiang. “A group mobility model for ad hoc wireless networks,” In Proceedings of the ACM International Workshop on Modeling and Simulation of Wireless and Mobile Systems (MSWiM), August 1999. [4] The Network Simulator 2 .http://www.isi.edu/nsnam/ns. [5] Scalable Vector Graphics. http://www.w3.org/TR/SVG/ [6] B.Liang, Z. J. Haas, “Predictive Distance-Based Mobility Management for PCS Networks,” in Proceedings of IEEE Information Communications Conference (INFOCOM 1999), Apr. 1999. [7] J.Tian, J. Hahner, C. Becker, I. Stepanov and K. Rothermel, “Graph-based Mobility Model for Mobile Ad Hoc Network Simulation,” in the Proceedings of 35th Annual Simulation Symposium, in cooperation with the IEEE Computer Society and ACM. San Diego, California. April 2002. [8] P. Johansson, T. Larsson, N. Hedman, B. Mielczarek, and M. Degermark, “Scenario-based performance analysis of routing protocols for mobile ad-hoc networks,” in International Conference on Mobile Computing and Networking (MobiCom'99), 1999, pp. 195-206. [9] F. Bai, N.Sadagopan, and A. Helmy, “Important: a framework to systematically analyze the impact of mobility on performance of routing protocols for ad hoc networks,” in Proceedings of IEEE Information Communications Conference (INFOCOM 2003), San Francisco, Apr. 2003. [10] Subodh Shah, Edwin Hernandez, Abdelsalam Helal, “CAD-HOC: A CAD-Like Tool for Generating Mobility Benchmarks in Ad-Hoc Networks,” SAINT 2002: 270-280. [11] C. E. Perkins, E. Royer, and S. R. Das, “Ad hoc On Demand Distance Vector (AODV) Routing,” In 2nd IEEE WorkShop on Mobile Computing Systems and Applications (WMCSA'99), pages 90-100, February 1999. [12] A.P. Jardosh, E.M.Belding-Royer, K.C. Almeroth, and S. Suri, “Towards Realistic Mobility Models for Mobile Ad hoc Networks,” In Proceedings of ACM MOBICOM, pages 217-229, San Diego, CA, September 2003. [13] Gang Lu, Demetrios Belis, Gordon Manson, “Study on Environment Mobility Models for Mobile Ad Hoc Network: Hotspot Mobility Model and Route Mobility Model”, WirelessCom2005, Hawaii, USA.
Gang Lu was born in Chengdu, China in 1978. He got his BSc with first class degree from East China Normal University in information and communications system in 2000. He came to the UK in 2001 and gained the MSc in data communications from the University of Sheffield in 2002. He started his PhD in mobile ad hoc networks from 2002 in the Center for Mobile Communications Research Group.
JOURNAL OF NETWORKS, VOL. 1, NO. 1, MAY 2006
His major research interests are self-positioning, locationbased routing, mobility modeling in mobile ad hoc networks, mobile ad hoc networks emulation and wireless sensor networks. He is also a lecturer of IT courses in the Institute for Lifelong Learning, the University of Sheffield, UK. He joined R&D department of Dialogue Communications Ltd. in 2003, UK. His current work role is the major researcher and developer of multimedia messaging and 3G applications. Mr. Lu is the translator of the book Data Networks (2nd Edition, by D. Bertsekas and R.Gallager, Chinese version), ISBN 7-115-12292-X/TP, Posts & Telecom Press, 2004. He is also the main author of the ongoing book Wireless Sensor Networks which is planned to be published in China in end of 2006.
Gordon Manson was born in Glasgow, Scotland in 1949. He was educated in Glasgow and in Irvine Ayrshire where he attended Irvine Royal Academy. He graduated from Strathclyde University in Glasgow with a 1st class honours degree in mathematics in 1971 and gained a PhD in applied mathematics in 1977. He lectured for one year at Strathclyde University and joined the Applied Mathematics department in the University of Sheffield in 1975. In 1982 he changed to the Computer Science department at Sheffield and he gains an MSc in Computer Science from the University of Manchester in 1984. He became a senior lecture in 1992 and this is his current position. His main research area was parallel processing using Transputers and he was the Assistant Director of the National Transputer Support Centre from 1987-95. Currently he is interested in applying ad-hoc techniques in swarming intelligent UAVs. Dr. Manson has published over 100 papers and been a coauthor in 2 books and a co-grant holder of 16 grants funded by industry, EPSRC and the EU. He was jointly awarded the best paper in Journal of Information Management and Computer Security in 2002.
Demetrios Belis was born in Piraeus, Greece. He graduated from the Technological Educational Institute of Piraeus in Greece with a degree in Automatic Control Systems in 1997. He obtained an MSc (Eng) in Automatic Control and Systems Engineering in 2000 and a PhD in Computer Science in 2006, both at the University of Sheffield, United Kingdom. His is specialized in the areas of Artificial Life and telecommunications, particularly in the fusion of ad hoc networks and autonomous agents. Currently he is working for learndirect (University for Industry, Ufi Ltd) at Sheffield, UK. He is also a lecturer of IT courses at the Institute for Lifelong Learning, University of Sheffield, UK.
© 2006 ACADEMY PUBLISHER
63
