1. Mobility Models and Traces
Wei-jen Hsu
Advised by Dr. Ahmed Helmy
Presented in CIS6930 class, Spring 2008

2. Outline Simulation of user mobility
Within NS-2
As a stand-alone code
Traces from existing wireless networks

3. Simulation of User Mobility
Why?
Mobile ad hoc networks are still not widely deployed, even though a lot of research has been done.
But it is a fundamental factor that influences network protocol performance.
Use simulation as a way to perform experiments in a controlled environment.

4. Simulation of User Mobility
How?
Use existing network simulation tools (e.g., NS-2) with extensions to handle user mobility.
Build you tool from scratch.

5. Simulation of User Mobility
How to choose a suitable approach?
NS-2
Powerful simulation tool, with many existing protocols implemented for it.
Hence complex, with high overhead
You have to fit your idea into its structure
You own tool
You can do anything you want, focusing on the part that matters to you
Clean-slate implementation
You have to re-build everything; credibility

6. Simulation of User Mobility
What do you want from the simulation?
Mobility metrics or statistics for the mobility model
Performance of routing protocols under the mobility model

7. NS-2 Mobility Interface
Protocol performance
Mobility metric
Mobility generator
Protocol
Simulation
(NS)
NS trace file format

8. Ns-2 Mobility file format
Mobility trace file
Format (the line you need to instruct how a node moves):
$ns_ at <time> “$node_(<id>) setdest <x> <y> <speed>”
Example: $ns_ at "$MN2 setdest “
Why is this format chosen??

9. Ns-2 Mobility file format

10. Ns-2 Mobility file format
NS provides only linear movement
What if I want a movement trajectory of arbitrary curve?
B(t1)
A(t0)

11. Plugging scenario files into NS
It’s simple, just add these lines in your tcl script
Create tcl variables for file location
set val(cp) "../mobility/scene/cbr-3-test"
set val(sc) "../mobility/scene/scen-3-test"
Include these files in simulation
source $val(cp)
source $val(sc)

12. How do I generate these files?
Default mobility file generator is RWP model. /indep-utils/cmu-scen-gen/setdest/setdest
Default traffic file generator is “random UDP flows” /indep-utils/cmu-scen-gen/cbrgen.tcl
Use IMPORTANT tool to generate more complex mobility scenario
Or… Write your own stand alone code!

13. How to use IMPORTANT tool?
Remember that it is a stand-alone program, with nothing to do with NS except that it’s output scenario files are NS-compatible.
So understand the program, compile it, provide parameters you want, and run it! 

14. What mobility models does IMPORTANT tool provide?
Temporal
Dependence
Spatial
Dependence
Geographic
Restriction
Application
Random
Waypoint
Model
General No No No
Group Mobility
Model
Battlefield No
Yes No
Freeway Mobility Model
Metropolitan
Traffic
Yes
Yes
Yes
Manhattan Mobility Model
Metropolitan
Traffic
Yes No
Yes

15. Random waypoint Random Waypoint Model
Each node chooses a random destination and moves towards it with a random velocity chosen from [0, Vmax]
After reaching the destination, the node stops for a duration defined by the “pause time” parameter
After this duration, it again chooses a random destination and repeats the whole process again until the simulation ends
Parameters: Max Velocity Vmax, Pause time T

16. Setdest utility Format Command $node(<id>) set X_ <x0>
$node(<id>) set Y_ <y0>
$node(<id>) set Z_ <z0>
$ns_ at <time> “$node_(<id>) setdest <x> <y> <speed>”
Command
./setdest –n <num_of_nodes> -p <pause_time> -M <max_speed> -t <simu_time> -x <max_x> -y <max_y> > <trace_filename>

17. Reference Point Group Mobility
Reference Point Group Model
Each group has a logical center (group leader) that determines the group’s motion behavior
Each nodes within group has a speed and direction that is derived by randomly deviating from that of the group leader
Parameter:
Angle Deviation Ratio(ADR) and Speed Deviation Ratio(SDR)
Max_velocity

18. Group Mobility Generator
In simulation, we use two sets of trace files
Single group: all nodes move within one group
Multiple group: each group moves independent of each other and in an overlapping fashion
Input
Mobility trace file of group leaders
Output
Mobility trace file of all nodes SG MG

19. Freeway Model Freeway Model
Each mobile node is restricted to its lane on the freeway
The velocity of mobile node is temporally dependent on its previous velocity
If two mobile nodes on the same freeway lane are within the Safety Distance (SD), the velocity of the following node cannot exceed the velocity of preceding node

20. Implementation Parameters Input: map format Output Key
Map and Max_velocity
Input: map format
<freeway id> <lane id> <x0,y0> <x1,y1>
Output
Trace file for all nodes
Key
Link list to maintain the order of nodes on the same lane
Randomly insert the nodes into various lane

21. Manhattan Model
Similar specification with freeway, but it allows node to make turns at each corner of street
At each intersection
Probability of moving on the same street is 0.5
Probability of turning right is 0.25
Probability of turning left is 0.25
Parameter
Map
Max_velocity

22. Manhattan Map Input: map
Street: <street_id> <lane_id> <direction> <x0,y0> <x1,y1>
Corner: <ver_str_id> <hrn_str_id> <x,y>

23. Time-Variant Community Model
Available at
User manual and mobility trace generator.
It generates mobility trace in 2 formats – NS-2 format and (x, y, t) format
You can configure the time period structure and communities for each node with full freedom

24. Issues with Mobility Models
Open network v.s. closed network
Each node has an ID, there is a given number of nodes, and they don’t leave the simulation area
This does not really capture the dynamics for a highly given environment
Potential solution: the “black hole”
Take care of the dynamics, but it is still the same node
Reuse the same node for a different virtual ID?

25. Issues with Mobility Models
Boundary Effects
The boundary can be also “torus” or “reflective”
Torus is unrealistic but analytically nice. Reflective is more realistic.
You need to make decisions on all these small things in the mobility model!!

26. Traces
Why traces?
A realistic measurement of user’s behavior in wireless networks
For example, “re-play” the trace as a mobility input for users in the simulation
Note this is a very different approach from mobility modeling (philosophy and granularity)

27. Where to get the traces? Collect it yourself
Sniffers, bluetooth encounters, etc.
Monitor the existing infrastructure
WLANs? Cellular phone networks? Internet traffic?
Go to the trace archives

28. The UF WLAN Trace
Collected from the on-campus WLAN in UF, at two different levels
Access points report syslogs
Authentication servers report syslogs
The trace is being uploaded to our server in the lab at almost “real time”

29. Access Points Syslogs Users are reported by MAC addresses
When they associate with a AP
When they disaccosiate from a AP
When they roam away from a AP
When some other event happens (error in packet checksum, max retry for a packet reached, etc.)

30. Authentication server syslogs
The authentication server reports the following events
DHCP lease – IP xxx is given to MAC yyy
User log in – User Gatorlink-ID logs in from MAC yyy
User log out – User Gatorlink-ID logs out, and it has been online for time ttt, sent/received bbb bytes
Every 30 minutes, each online user is reported for its traffic usage in the past 30 mins

31. What is new?
The differentiation of “having the intention to use wireless network” v.s. “just turning on the computer”
The capability to analyze the data at “device level (MAC address)” and “user level (Gatorlink ID)”
Getting the trace at “almost real time”
Traffic summary with location information

32. Other Popular WLAN traces
USC trace
Collected from summer 2005 to today
Time/location information of the user (association) at switch port level. The mapping between switch port and building is available but messy (changing a lot)
Dartmouth trace
Collected from 2001 to 2004
Time/location information of user, SNMP logs from access points, tcpdump traffic headers

33. Other Popular WLAN traces
UCSD – PDA experiment
~275 PDA users
Time/location information, each PDA logs all APs in communication range
MIT/IBM
Corporate users from 3 buildings
Time/location information, SNMP logs

34. Traces of Other Types Encounter traces Cellphone traces
The Intel/Cambridge Haggle/Pocket Switch Network project
The U of Toronto PDA-based encounter experiments
Cellphone traces
MIT Reality Mining: encounter, location of users (by cellphone tower/bluetooth), call log

35. Usage of the Traces To understand existing systems (WLANs)
How, when, where do users use WLAN?
Can users be classified based on their behaviors?
Can we build models of users?
To leverage it as a ground for not-yet-deployed services/protocols
IF users behave like this, what if we do….
IF users continue to develop a trend, where will we be in year 201X?

36. Final Notes Downside of trace-based work Concerns
Unable to access the ground truth
Using the current user data to speculate the future
We have plenty data sets for the normal scenarios, but should they be the focus?
Concerns
Privacy
What’s a real-world application that can’t be done with today’s the great Internet?

37. Potential Directions Combing the trace with small-scale experiments
Build new testbeds and prototype new services
Come up with a specific scenario to provide solutions for
Challenge the established knowledge

38. Tricks of Trace Processing
Identify a common format that you can convert multiple traces into
I use one file for each user, within each file, each line represents “time location duration”
Abuse your hard drive
Keep intermediate results if they take long time to generate.... You will thank your former self years after you generated those files
Learn handy tools
Shell script, perl, database (Udayan’s turn)