dortmund university OMNeT++ Community Summit 2016 An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Benjamin Sliwa, Christoph Ide and Christian Wietfeld September 16, 2016 Faculty of Electrical Engineering & Information Technology Communication Networks Institute Prof. Dr.-Ing. Christian Wietfeld
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Problem Statement IEEE 802.11g Link Video stream Base station Mission area Application of autonomous agents for exploration of hazardous areas Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 2
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Problem Statement IEEE 802.11g Link Video stream Base station Mission area Application of autonomous agents for exploration of hazardous areas High mobility causes frequent changes of the network topology Stressed routing protocols and losses of the communication link Solution approach: using mobility control knowledge for intelligent forwarder selections Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 3
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Solution Approach Meta-model for realistic modelling of Application Exploration cooperative agents task definition Trajectory prediction for precise position estimation Mobility algorithms Controlled Mobility Mobility-aware routing approaches request reply: current mobility data Leveraging Mobility Proof of concept evaluation Prediction method Control Knowledge predicted trajectory B.A.T.Mobile Predictive routing MA-OLSR optimized neighbor selection Improved Packet Goal Delivery Ratio OLSR : Optimized Link State Routing Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 4
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Modelling the Mobility Behavior of Autonomous Agents Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 5
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Three basic Rules for the Behavior of Individual Agents Travelling in Swarms Centroid Repelling forces Plume source Attracting forces Resulting movement Cohesion Alignment Separation Collision avoidance Swarm coherence Task fulfillment Potential fields / position- Potential fields / position- Cooperative mobility based based algorithms Real-world swarming scenarios require multiple mobility algorithms to be executed in parallel [1] Reynolds, C. W., “ Flocks, herds and schools: A distributed behavioral model, ” in Proceedings of the 14th annual conference on Computer graphics and Meta-model for mobility interactive techniques, ACM Press, 1987, 25-34 Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 6
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Layers of the Reynolds Mobility Meta-model ActionSelection Steering Locomotion [2] Reynolds, C. W., “ Steering behaviors for autonomous characters, ” in Game developers conference, San Francisco, California 1999 Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 7
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Layers of the Reynolds Mobility Meta-model Task definition Exploration ActionSelection Network Provisioning Tasks Agent classes Scout Steering Relay Locomotion [2] Reynolds, C. W., “ Steering behaviors for autonomous characters, ” in Game developers conference, San Francisco, California 1999 Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 8
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Layers of the Reynolds Mobility Meta-model Fulfillment of individual tasks Path planning ActionSelection Collision avoidance Swarm coherence Logic Tasks Mobility algorithms Steering Steering Vector is the desired Steering Vector movement vector for the next iteration Locomotion [2] Reynolds, C. W., “ Steering behaviors for autonomous characters, ” in Game developers conference, San Francisco, California 1999 Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 9
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Layers of the Reynolds Mobility Meta-model Maps the desired movement vector to a travelled vector ActionSelection Physical realization of the Logic movement Tasks Simulation: calculation of the next position Steering Real vehicle: motor control Acceleration and braking Steering Vector Implementation of the vehicle Platform type Locomotion Increases portability of the mobility algorithms / steerings [2] Reynolds, C. W., “ Steering behaviors for autonomous characters, ” in Game developers conference, San Francisco, California 1999 Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 10
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Example Behavior of the Reynolds Mobility Meta-modell Mission area Step 1 : Handle the Exploration steering Step 2 : Handle the Collision Avoidance steering Step 3 : Compute the total Steering Vector with the assigned weights Steering Weight Step 4 : Handle the locomotion and limit Exploration 1 the movement vector with respect to the Collision Avoidance 2 physical capabilities of the vehicle Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 11
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Implementation of a Mobility Algorithm as a Steering module MobilityAlgorithm extends Steering class MobilityAlgorithm : public Steering { { public: parameters: MobilityAlgorithm(); } protected: SteeringVector update() ; }; MobilityAlgorithm.ned SteeringVector MobilityAlgorithm::update() { **.host*.mobilityType = "ReynoldsMobilityModel" SteeringVector vector; **.host*.numSteerings = 2 **.host*.steering[0].typename = "MobilityAlgorithm" // do fancy stuff **.host*.steering[0].weight = 2 vector = … **.host*.steering[1].typename = "AnotherMobilityAlgorithm" **.host*.steering[1].weight = 1 return vector; } **.host*.locomotion.typename = "UAVLocomotion" MobilityAlgorithm.h/.cc omnetpp.ini Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 12
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Mobility Prediction Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 13
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Leveraging Mobility Control Knowledge for Precise Position Predictions Current Steering Vector Mean vector using the last positions Current position and Waypoints position history Steering Vector Waypoint direction Extrapolation Iterative method to predict the position at a defined target iteration (here ) Usage of the most precise available mobility information in each step [3] B. Sliwa, D. Behnke, C. Ide, C. Wietfeld, "B.A.T.Mobile: Leveraging Mobility Control Knowledge for Efficient Routing in Mobile Robotic Networks", In IEEE GLOBECOM 2016 Workshop on Wireless Networking, Control and Positioning of Unmanned Autonomous Vehicles (Wi-UAV) , Washington D.C., USA, Dezember 2016, accepted for presentation. [Online]. Available: http://arxiv.org/abs/1607.01223 Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 14
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Accuracy of the Prediction Method Example evaluation with prediction width 15 Steering Vector integration decreases the prediction error Waypoints act as static orientation points Steering Vector is only beneficial if waypoints are not available W : Waypoints S : Steering Vector E : Extrapolation GNSS : Global Navigation Satellite System The prediction stability can highly be increased by integrating waypoint information Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 15
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Mobility-aware Routing Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 16
dortmund Communication Networks Institute Prof. Dr.-Ing. C. Wietfeld university Predictive Path Planning with Mobility-aware OLSR Example: What is the best path from A to E? B C B C A A E Trajectory prediction D E D Current network state Future network state Trade-off: shortest path vs stability Choose the shortest path with the highest availability Requires mobility information of all nodes high overhead Benjamin Sliwa | An OMNeT++ based Framework for Mobility-aware Routing in Mobile Robotic Networks Slide 17
Recommend
More recommend
