By: Masoud Moshref [email protected]
MOBISIM : A FRAMEWORK FOR SIMULATION OFMOBILITY MODELS IN MOBILE AD-HOC
NETWORKSV. 3.0
Project homepage: http://sourceforge.net/projects/mobisim/Publication: S. M. Mousavi, H. R. Rabiee, M. Moshref, A. Dabirmoghaddam, " MobiSim : A Frameworkfor Simulation of Mobility Models in Mobile Ad-Hoc Networks", The 3rd IEEE International Conferenceon Wireless and Mobile Computing, Networking and Communications (IEEE WiMob 2007), New York,USA, October 8-10, 2007
2007-2011
Mobile Ad-Hoc Networks
Wireless
Mobile
Self-configurable
No initial platform
INTRODUCTION
MobiSim v. 3.0 2
* Images in this slide are from http://www.hitachi.com/rd/research/sdl/07/soc_02.html
2007-2011
Where to find mobility traces?
Real traces
Mobility model generators
Trace granularity
Cell, Ex: Cell phone network
GPS based, Ex: Ad-Hoc networks, Vehicular networks
INTRODUCTION
MobiSim v. 3.0 3
2007-2011
MOBISIM
MobiSim v. 3.0 4
2007-2011
Modes:
Graphical
Simulation
Visualization
With Various way of presentation
Batch (Scenario based)
Simulation
Scenario configuration
MOBILITY GENERATOR
MobiSim v. 3.0 5
2007-2011
STRUCTURE OF A CONFIGURATION
MobiSim v. 3.0 6
2007-2011
Location Initializer Predefined (fixed)
Random
Various distributions
Repeatable scenarios using the defined seed
Model specific initializer Map
Reflective Square
Reflective Disk
Obstacle
Model specific maps
• Range
GENERAL PROPERTIES OF A MODEL
MobiSim v. 3.0 7
Disk map The path inFreeway map
Obstacle map and the spatial distribution of nodes inRandom Walk model on it
2007-2011
A set of handles will define the properties of each object (Ex: scale, rotation ofan obstacle)
Can add, delete, or move a handle A handle can have its properties A model may handle it on behalf of its map A map/model must validate the position of handles
MAP EDITOR
MobiSim v. 3.0 8
2007-2011
MAP EDITOR: SNAPSHOT
MobiSim v. 3.0 9
2007-2011
Random Waypoint [1] Random Direction [2] Random Walk [3] Levy Walk [4] Tortoise
MOBILITY MODELS: RANDOM MODELS
MobiSim v. 3.0 10Foot print in Levy walk Spatial distribution in Levy walk
Spatial distribution in Random Waypoint
Spatial distribution in Random Direction
2007-2011
A sequence and loop of
Transitions with fixed properties
Transitions with random properties
A sequence/loop of transitions (make a complex pattern)
RANDOM MODELS: TORTOISE MODEL
MobiSim v. 3.0 11
2007-2011
Manhattan Grid [3]
Use Mapeditor to manage the grid
Freeway [3]
Use Mapeditor to draw any path
Special parameters:
Safe distance ratio
Positive acceleration ratio
Lane numbers
MOBILITY MODELS: URBAN MODELS
MobiSim v. 3.0 12
Spatial distribution for Manhattan Grid
2007-2011
Gauss-Markov model [5]
New speed and direction:
Probabilistic Random Walk [6]
Change x/y based on a Markov chain
Exponential correlated model
MOBILITY MODELS: TEMPORAL DEPENDENTMODELS
13MobiSim v. 3.0
2007-2011
Select Leaders’ Model from any models even group models(Usually Random Walk)
Picks a reference point [7] for each group member and defineits behavior based on it (with some randomness).
Variations in selecting the point makes different behaviors:
Nomadic: Pick a point around the leader
Pursue: The current leader position is the reference point
String: The previous node’s position will be the reference point
Row: Choose the point to keep the row structure in the group
MOBILITY MODELS: GROUP MODELS
MobiSim v. 3.0 14Pursue model with a disk map
Footprint of Nomadic model
2007-2011
Multi Model: Compose models
Use Mapeditor to manage map of internal models
ThreeDizer: Add z value based on a ground
Use Mapeditor to define the highs and lows
Use 3D resimulator to view in 3D environment (needs Java3d)
File Model: load a trace file as an internal model or group leader’s model
MOBILITY MODELS: COMPLEX MODELS
MobiSim v. 3.0 15
Set z value for girdand customizedpoints to describethe ground
Compose differentmodels with differentmap shapes
2007-2011
A scenario has a simulation
With variable parameters
Runs multiple times
Used in batch mode to generate a mass of traces
SCENARIOS
MobiSim v. 3.0 16
2007-2011
Types:
Text: Tab separated table with two top lines as configuration and table headers
XML: Text format converted to a more readable xml format
NS2
Models can add trace columns. Ex: group number
TRACE FORMATS
MobiSim v. 3.0 17
2007-2011
Input: multiple trace files Output: a table of evaluations Can copy some parameters of each trace file (Ex: model name) General properties for an evaluator
Sample time
Sort priority
EVALUATORS
MobiSim v. 3.0 18
2007-2011
Spatial dependency [3] Temporal dependency [3] Relative speed [3] Repetitive behavior [8] Transition characteristics
Time
Length
Speed
Degree change
Location Distribution Variance
MOBILITY EVALUATORS
MobiSim v. 3.0 19
2007-2011
Average degree Average range Clustering coefficient [8] Graph connectivity
# of disconnection
Duration of disconnection
Average size of the biggest cluster Interference Link evaluation
Link duration
Neighborhood instability [8]
Intermeeting time Network diameter
GRAPH EVALUATORS
MobiSim v. 3.0 20
2007-2011
Steps:1. Select input evaluation file and output Excel file
2. Select the input columns for making rows. Ex: model name
3. Select the input columns to create output table columns (variables) Ex: Min/Max speed
4. Select the evaluation column to fill table cells with (will be average value if multiple input rowsmap into a cell)
5. Can divide tables using other columns onto several shits
6. Use “Create Diagram” in Excel to draw diagrams using a script
EXPORT EVALUATION FILE TO MS EXCEL
MobiSim v. 3.0 21
2007-2011
Features: input from evaluations Distance metrics
Euclidean
Gaussian: considers data dispersion by standard deviation as a weight for distance of a pointfrom the center of a group/group member.
Mahalanobis: considers data distribution by finding the dispersion in n axis defined by SVG Classification algorithms
Nearest Center
KNN
Fuzzy Nearest Center
Fuzzy KNN Evaluating classification algorithms
Accuracy
Average right membership
CLASSIFICATION
MobiSim v. 3.0 22
2007-2011
General urban model New mobility models New trace types Dynamic range management Power models Signal attenuation on obstacles in the graph building algorithm
FUTURE WORK
MobiSim v. 3.0 23
2007-2011
1. Broch, J., et al., "A performance comparison of multi-hop wireless ad hoc networkrouting protocols." Mobicom98, ACM, , October 1998.
2. Royer, E. M., Melliar-Smith, P. M. and Moser, L. E., "An Analysis of the OptimumNode Density for Ad hoc Mobile Networks.“, ICC, Finland, June 2001
3. Bai, Fan, Sadagopan, Narayanan and Helmy, Ahmed., "The Important frameworkfor analyzing the Impact of mobility on performance Of Routing Protocols forAdhoc Networks." Elsevier Journal of Ad Hoc Networks, 2003, pp. 383-403
4. Chong, Injong Rhee and Minsu Shin and Seongik Hong and Kyunghan Lee andSong., "On the Levy-Walk Nature of Human Mobility." INFOCOM, 2008. pp.924-932.
5. Liang, B. and Haas, Z., "Predictive distance-based mobility management for PCSnetworks.“ INFOCOM, March 1999
6. Chiang, C., Wireless Network Multicasting. Los Angeles : PhD thesis, University ofCalifornia, 1998
7. Hong, X., et al., "A group mobility model for ad hoc wireless networks,." MSWiM,August 1999
8. Theoleyre, F., Tout, R. and Valois, F., "New metrics to evaluate mobility modelsproperties”, ISWPC 07. , Feb. 2007
REFERENCES
MobiSim v. 3.0 24