1. A Social Approach for the Spreading of Messages in Vehicular Ad Hoc Networks
Alexandra Stagkopoulou, Pavlos Basaras, Dimitrios Katsaros University of Thessaly, Greece, Volos

2. Outline Introduction-Motivation Related approaches
From Control Centrality to pCoCe
Experimental Setup & Evaluation
Conclusion

3. Introduction-Motivation
V2V, V2I, I2V and I2I communications are important components of ITS that find many applications in real life
The increasing number of car accidents, the high traffic volumes combined with traffic congestions or the environmental impact in CO2 emissions urge for the use of such communications to increase the living quality in urban cities

4. One-to-all Communication
Broadcasting is a broadly used method for urgent notification events
Goal: inform the entire vehicular network
Vehicles informed of unfavorable road conditions, (of blocked roads, traffic jams or accidents) will take prompt actions and alternate their route
thus save time, fuel and money
However:
Broadcast storm problem, exhaustion of wireless resources, redundant retransmissions etc..

5. Relevant Methods Costly Strategies: Flooding approaches:
Connected Dominating Sets (CDSs)
Grouping vehicles in Clusters
Flooding approaches:
Simple flooding
Probabilistic flooding
Relay vehicles: select vehicles by some criterion who on behalf of a sender will further rebroadcast a message

6. Relay Vehicles

7. Intuition of Influential spreaders
Influential spreaders in complex networks i.e. node entities able to influence-propagate a message to a sufficiently large part of a complex network
Vehicular networks are complex networks
Intuition: select as relay nodes the influentials

8. Control Centrality(CnC) - Definitions
A stem in a directed network is a path of n nodes and n-1 connections among them such as no node appears more than once, e.g. k->l->m->n
A cycle is noted as a stem ending on the initial node, e.g. k->l->m->k
A stem-cycle disjoint subgraph, is a subgraph of the directed network where stems and cycles have no nodes in common

9. Control Centrality (CnC)
The largest number of connections (including the input) among all possible stem-cycle disjoint subgraphs constitutes the CnC value of a node- vehicle, e.g. CnC = 6.

10. From CnC to pCoCe Algorithm
Link connections between vehicles are bidirectional, thus we include the vehicles directions to built in-out neighborhoods.
A vehicle x is considered an out of y if y is:
in front of x and on the same direction
Moving away from y in a different direction

11. In-out Neighbors
For vehicle 7: out-neighbors{1,6,3} || in-neighbors{the rest}

12. Cycles and Overestimation
The use of cycles to enhance a vehicle’s importance in a vehicular network is very likely to overestimate the vehicle’s ability for disseminating a message
Cycles stand for revisited or repeated regions
Hence cycles created from vehicle paths are not used to enhance a vehicle’s potential in our method and thus we account only for vehicle-stems

13. Nature of Network
CnC is built on the knowledge of the entire network topology
VANET networks however are dynamic in nature with constantly changing topology and connection patterns:
Neighbors increase or decrease their distance in and out of a communication range
In neighbors may become outs’ and vise versa

14. Focus On One’s Local Neighborhood
Due to great cost for maintenance we focus on local distance i.e. 2 hop stems from the originator: a simple scenario of 2pCoCe
Blue Cars: out-Neighbors, Grey cars: in-Neighbors

15. Strength of Connections
Connections between vehicles may vary in ‘strength’ due to potential obstacles (e.g. buildings) or due to a great distance between them etc..
pCoCe includes a parameter that accounts for the strength of connections and thus the strength of the corresponding stem in [0,1]
values close to 1 depict a perfect communication link
while values close to 0 depict an almost absent connection
Weights close to 1 depict a perfect communication link whereas values close to 0 depict an almost absent connection.

16. Strength of Connections

17. Our Approach, pCoCe
In our experimentation the communication links are considered perfect, i.e. PW= 1

18. pCoCe Relay Selection Goal: cover the entire two-hop out neighborhood
The initial sender (S) requests from his neighbors to commute and send their pCoCe values
When a neighbor computes its pCoCe index, excludes any potential stems incoming to S
i.e. final vehicle of the stem is an incoming neighbor of S.
The next highest pCoCe value is included in the set when covering additional two-hop out neighbors and so on until the goal is reached

19. pCoCe Relay Selection When a vehicle
Vector[0]: is the vehicle with the highest pCoCe value, etc..

20. OLSR Protocol Proactive routing protocol
maintains a list of destinations and routes by periodically exchanging topology messages
Makes use of relay nodes called multipoint relays (MPRs)

21. Olsr Relay Selection Goal: cover the entire two-hop out neighborhood
Select as relays those who provide unique access to some vehicle(s)
If the are still vehicles uncovered by the current MPR set select the node which covers the most of the remaining out two hops and so on until the goal is reached

22. Olsr Relay Selection

23. Setup and Evaluation
Grid network of the city of Volos and sumo map simulation

24. Simulation Setup
Veins composed of the network simulator OMNET++ and the traffic simulator SUMO
Bidirectional grid road network (3X3) with intersections every 2km
DSRC with communication range 250,500m
Vehicle speed: 50,70,90 km/h
Density: a vehicle is introduced in the simulation every 1,5,10 and 15 seconds ranging from very dense to very sparse scenarios
Average number of vehicles with respect to frequency: 950, 250, 170, 120

25. Simulation Setup
Beacon messages are exchanged every 1 second in order for vehicles to be aware of their vicinities
A notification message event is triggered by a random vehicle at random position
Results are averaged over 10 different notification events

26. Speed 50km/h, Com. Range 500m

27. Speed 70km/h, Com. Range 500m

28. Speed 90km/h, Com. Range 500m

29. Increasing stems to 3 hops: 2pCoCe VS 3pCoCe

30. Differences in the size of selected relay sets

31. Com. Range 250m, Frequency 1s

32. Conclusion pCoCe Olsr Small additional communication cost
Performs extensively well when dealing with a large number of potential relay choices
Olsr
Often fails to cover a sufficiently large portion of the vehicular network
Its performance decreases when dealing with large number of potential relays

33. Thank You
Questions ??