From Communication to Traffic Self-Organization in VANETs Gianluigi Ferrari, 1 Stefano Busanelli, 1 Nicola Iotti 2 1 WASN Lab, Dept. of Information Eng., UniParma, Italy 2 Guglielmo Srl, Pilastro (Parma), Italy RD 11 @ Klagenfurt - July 2011
Outline 1 Our Activities: Communications in VANETs 2 3 G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 2 / 19
Outline 1 Our Activities: Communications in VANETs 2 3 G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 3 / 19
Our Activities: Communications in VANETs VANETs: The Big Picture Event-driven Communications (V2V) Data dissemination (I2V) Cluster 1 RSU Cluster 1 G. Ferrari RSU Cluster 2 Distributed Data Collection (V2I) Cluster 2 RD 11 @ Klagenfurt Cluster 3 Cluster 4 July 11-15, 2011 4 / 19
Broadcast Protocols in VANETs Topology information dissemination Periodic one-hop broadcast transmissions Unicast forwarding protocol need topology information Weak QoS requirements Event-driven information dissemination Event-driven multihop broadcast transmissions Most of the information is intrinsically broadcast (i.e., accidents) Large area to be covered in short time and reliably No time to establish unicast communications Broadcast storm problem!! G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 5 / 19
Our Activities: Communications in VANETs Silencing Irresponsible Forwarding: the starting point Our proposal Nodes decide to retransmit in a probabilistic way A node with a scheduled transmission interrupts the backoff when it hears a retransmission from a better placed node (silencing) Intuition: the best relay is the farthest one reachable from the transmitter Location-dependent retransmission probability: The Retransmission Probability ρs (z d) Pre tx = exp c where ρs is the vehicle spatial density, z is the transmission range, d is the distance between transmitter and receiver, and c is a shaping parameter FLOOD G. Ferrari IF RD 11 @ Klagenfurt SIF July 11-15, 2011 6 / 19
Our Activities: Communications in VANETs Impact of Mobility on Broadcast Dissemination (1) Highway scenario Urban scenarios Roads with traffic lights and roundabouts Multi-lane roads SUMO simulator with Krauss model Intelligent Driver Model (IDM) v max = 20 m/s v min = 30 m/s, v max = 50 m/s VanetMobiSim simulator ROI z RSU G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 7 / 19
Impact of Mobility on Broadcast Dissemination (2) 0.3 1 0.25 0.8 0.2 0.6 0.15 0.1 0.05 0.4 0.2 0 0 5 10 15 20 25 30 35 0 20 40 60 80 100 120 Mobility Throughput (communication) G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 8 / 19
Exploiting Hot Traffic Intersections 1000 900 800 700 600 2nd Transmission Domains 3rd Transmission Domain DP 500 400 1st Transmission Domain 300 DP 200 100 0 0 100 200 300 400 500 600 700 800 900 1000 Idea: extend the coverage of dissemination points (DPs) positioned at hot intersections (high traffic, great connectivity) G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 9 / 19
Our Activities: Communications in VANETs Cross-Network Effective Traffic Alert Dissemination (X-NETAD, Eureka Label E! 6252) UMTS Traffic Alert WiFi IF-based Alert Broadcast WiFi IF-based Alert Broadcast WiFi IF-based Alert Broadcast Idea: rapid diffusion of UMTS-based traffic information through local WiFi networking G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 10 / 19
From Communication to Traffic Self-organization Communication/networking applications: assume a given traffic model (e.g., SUMO: car-following models) No interest in traffic control Traffic accumulation (e.g., jams): very good for connectivity Traffic control goal: improve drivers safety Can we embed traffic control information in ever increasing vehicular communications? Ultimate goal: hidden (to drivers) self-organizing traffic mechanism G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 11 / 19
Outline 1 Our Activities: Communications in VANETs 2 3 G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 12 / 19
Traffic Self-Organization Several (mostly highly theoretical) models, inspired by empirical data: microscopic follow-the-leader; cellular automata-based; continuous Markov chain-based (statistical mechanics); macroscopic; gas-kinetic. Several universal properties emerge (e.g., transition to congested traffic at bottlenecks and ramps) What about traffic control? G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 13 / 19
A Pragmatic Per-road Approach Define the following danger function of a vehicle: d break (v) d 0 < d < d f danger (d, v) = break (v) d 0 d > d break (v) where d is the distance to the preceding vehicle, d break (v) v 2 is the breaking distance If all vehicles are moving at the same speed and given an overall space, then minimizing the overall danger function ( i f (i) danger ) leads to the intuitive solution that all vehicles should be equally spaced. What happens in an urban scenario with intersections, pedestrian crossings, etc., i.e., in a multi-road scenario? Other objectives (minimize transit times, e.g., eletrical cars) G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 14 / 19
Taking Inspiration from Ants Foraging ants: form attractive (fastest) trails to food sources through pheronome-based mechanisms (reinforcement of pheronome density) cohesive forces What happens in the presence of bottlenecks? A. Dussutour, V. Fourcassi, D. Helbing, J.-L. Deneubourg, Optimal traffic organization in ants under crowded conditions, Nature 428, 70-73 (4 March 2004). doi:10.1038/nature02345. G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 15 / 19
From Ants to Vehicle Traffic Control Ants balance cohesive and dispersive forces cohesion: pheronome-based dispersive: pushing Self-organization can be described through a nonlinear modelling approach (based on inhibitory interactions) Mimic this behaviour in realm of traffic control What is an information pheromone? What is the equivalent of pushing? Practical perspective: given that we can identify proper messages (e.g., with inertial information) to be disseminated, what is the communication overhead? Design goal: self-organize at the minimum communication cost G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 16 / 19
Outline 1 Our Activities: Communications in VANETs 2 3 G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 17 / 19
Where do we come from: vehicular communications Self-organization: several existing approaches Where we would like to go: embed information pheromones in information dissemination packets Traffic self-organization at minimum communication cost G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 18 / 19
THANK YOU FOR YOUR ATTENTION Contact gianluigi.ferrari@unipr.it G. Ferrari RD 11 @ Klagenfurt July 11-15, 2011 19 / 19