Where we are

• Random model

• Flow-based model

• Traffic models

• Trace based models 

Urban Mobility Models for VANETs

• Limitations and short comings of current models
• Understanding the sensitivity of traffic models to various modelling details
• Present 3 new models:
  • Stop Sign Model (SSM)

  • Probabilistic Traffic Sign Model (PTSM) 

  • Traffic Light Model (TLM) 

• Compare these models with previous ones

• Reason as to take conclusions regarding the sensitivity of traffic models to various modelling details

Limitations of Current Models

• Tend to ignore real-world constraints:
  • Street Layouts
  • Traffic Signs
• Results in simulations that do not accurately represent the protocol's performance

Factors Affecting Mobility in VANETs

• Street Layouts
  • Well defined paths
  • Affects connectivity
• Block Size
  • Determines the number of intersections
  • Larger blocks -> more clustering
• Traffic-control Mechanisms
  • Traffic lights, stop signs
  • Affects the average speed
• Independent Vehicular Motion
  • Microscopic Models
• Average Speed
  • Rate of network topology change

Stop Sign Model

• Every street at an intersection has a stop sign. 
• All the vehicles must stop
• A vehicle is conditioned by the one in front of him
  • No overtakings unless considering a multi-lane model
• The model does not illustrate the real world at 100% but it's a good starting point towards understanding the dynamics of mobility and its effect on mobility performance.

Probabilistic Traffic Sign Model (PTSM) 

• Refinement of SSM
  • Stop signs replaced with traffic lights
• ​No coordination between traffic lights of different directions
• When a node reaches in intersection
  • Stop - red light, probability p
  • Continues - green light, probability 1-p
• Cars arriving at a queue wait the time of the previous node plus 1 second (startup delay)
• This model avoids excessive stopping. 

Traffic Light Model

• Next iteration of PTSM
• New features:
  • Coordinated traffic lights
  • Acceleration and deceleration
  • Multiple Lanes
• The goal is to "regulate knobs" and find the best way to get relevant results.

PERFORMANCE
EVALUATION

Test Conditions

• NS-2 Simulator
• AODV(Ad hoc On-Demand Distance Vector) routing protocol.
  • 100 nodes
  • 250m transmission range
• Comparisons
  • SSM,
  • PTSM,
  • TLM, the Random Waypoint Model (RWM)  
  • Rice University Model (RUM) . 

Varying number of nodes

• The delivery ratio increases with the number of nodes, up to 100 nodes, as the connectivity of the communication graph increases.

• The delivery ratio starts decreasing as the number of nodes increases further.  - Too many control messages

Varying number of nodes

• Acceleration/deceleration led to a significant increase in the delivery ratio because this feature reduces the average speed of vehicles. -> network routes are more stable.

• The performance difference between the single-lane and multilane models is not noticeable below 100 nodes.

• Once the acceleration/deceleration is enabled, the difference between the single-lane and multi-lane models becomes negligible.  

Varying Number of Constant Bit Rate Sources 

• When the number of sources increases beyond 15, there is an increase in the end-to-end delay by an order of magnitude and a significant drop in the delivery ratio.
• When the number of CBR sources increases, there is an increase in the number of packets contending for a common wireless channel, which leads to more collisions and packet drops.  

• Respecting the speed limit of each road
• Effects are hardly noticeable

Varying Vehicle Speeds 

• As the block size increases, the delivery ratio decreases. 

• With large block sizes, vehicles spend more time in traversing between intersections; thus, nodes are mobile more often.

• This increased mobility leads to a weakened connectivity in the network, and a corresponding drop in the delivery ratio. 

Effect of Block Sizes

Real Map Results

• The delivery ratio for each model increased with the number of nodes up to 100 nodes, followed by a rapid degradation in performance thereafter.
• However, the performance using TLM remained constant up to almost 200 nodes. 
• These results confirm the correlation between topology and mobility, and between the mobility and performance of the simulated VANETs. 

Conclusions

• Mobility models play a critical role in accurate simulation of routing protocol performance in Vehicular Ad Hoc Networks (VANETs). 
• The authors propose three new but related vehicular mobility models – the Stop Sign Model, the Traffic Sign Model, and the Traffic Light Model 
• The clustering effect of vehicles waiting at intersections and acceleration/deceleration of vehicles are significant factors that affect the delivery ratio and packet delays in VANETs.
• The simulation of multiple lanes and coordinated traffic lights has only a marginal impact on the ad hoc routing performance.  

My thoughts 

• Bullet One
• Bullet Two
• Bullet Three