Motion-Aware Routing in Vehicular Ad-hoc Networks Christopher Cardé March 13, 2003
Introduction & Outline Mobile Ad-hoc Networks (MANETs) Overview Vehicular applications Motion-aware Routing Concept Competing ideas Simulation March 13, 2003 Christopher Cardé, Motion-Aware Routing
Mobile Ad-hoc Networks Characteristics: Non-infrastructure network of nodes Moving nodes → changing topology Develop routes proactively or on-demand Goals (in general, vary by application) Global availability Low power / high efficiency Low latency High bandwidth March 13, 2003 Christopher Cardé, Motion-Aware Routing
Mobile Ad-Hoc Networks Many protocols & approaches: AODV (Ad-hoc On-demand Distance Vector) DSR (Dynamic Source Routing) Research areas, more “exotic” approaches: LUNAR Geocasting March 13, 2003 Christopher Cardé, Motion-Aware Routing
Mobile Ad-Hoc Networks Applications: Air-dropped sensor network Battlefield communications Search & rescue operations Inter-vehicular networking March 13, 2003 Christopher Cardé, Motion-Aware Routing
Christopher Cardé, Motion-Aware Routing Vehicular Networks Why network vehicles? Share traffic, safety hazard information Distributed traffic statistic generation Extend reach of infrastructure networks Lower infrastructure costs! Provide information / entertainment services to passengers. March 13, 2003 Christopher Cardé, Motion-Aware Routing
Christopher Cardé, Motion-Aware Routing Vehicular Networks Road network has special properties At microscopic scale, generalizes to line with bidirectional traffic flow. Exploit this to reduce topology “churn” by choosing peers moving with you instead of against? March 13, 2003 Christopher Cardé, Motion-Aware Routing
Motion-Vector Routing Proposal: Include motion vector in neighbor advertisements. Prefer neighbors with similar motion vectors Goal: Reduce rate of topology change Reduce frequency of route changes Increase efficiency & availability March 13, 2003 Christopher Cardé, Motion-Aware Routing
Motion-Vector Routing Goal: Improve this: March 13, 2003 Christopher Cardé, Motion-Aware Routing
Motion-Vector Routing Goal: To this: Route will change less frequently! March 13, 2003 Christopher Cardé, Motion-Aware Routing
Motion-Vector Routing Competing / similar idea: Location-based routing Chooses neighbors / routes based on location Requires GPS / other absolute positioning Increased processing overhead Single position vector does not show trends Motion-vector routing, if it works, would be cheaper and simpler to implement. March 13, 2003 Christopher Cardé, Motion-Aware Routing
Christopher Cardé, Motion-Aware Routing Simulation Validating the concept: Home-brew Java time-stepped simulation OO architecture allows pluggable: Motion models Routing strategies March 13, 2003 Christopher Cardé, Motion-Aware Routing
Christopher Cardé, Motion-Aware Routing Simulation The plan: Assume average duration of “in-range” with neighbor ≈ average rate of topology change. Use as metric to evaluate algorithm’s effectiveness Use several motion models “Random” motion model (used as baseline in many MANET papers) Grid motion model More complex road networks On each motion model, try: Standard (range-based) neighbor selection Motion-aware routing March 13, 2003 Christopher Cardé, Motion-Aware Routing
Christopher Cardé, Motion-Aware Routing Simulation Results not yet ready Simulation not fully tested Some unreasonable numbers being generated Can’t tell how well it works yet! March 13, 2003 Christopher Cardé, Motion-Aware Routing
Christopher Cardé, Motion-Aware Routing Conclusion An area of important future research Special properties might be exploited to improve performance In paper: Simulation results Further technical details of simulator Any questions? March 13, 2003 Christopher Cardé, Motion-Aware Routing