Automated Routing Protocol Selection in Mobile Ad Hoc Networks Taesoo Jun and Christine Julien March 13, 2007 Presented by Taesoo Jun The Mobile and Pervasive Computing Group Electrical and Computer Engineering The University of Texas at Austin
Outline Challenges and Goal Motivating Scenarios Protocol Selection Process Conclusion 2
Challenges Challenges of Mobile Network Deployers Multitude of choices for communication protocols Different characteristics dependent on operational environment Various applications and goals to be achieved Different application requirements Consideration of operational environment and application requirements Select the most appropriate protocol for a particular situation 3
Goal Create an Automated Design Tool Takes user input Physical characteristics of network: mobility degree, node density, etc. Characteristics of application: traffic rate, application goal, etc. Recommends the most appropriate protocol Based on quantitative evaluation of candidate protocols 4
Disaster Recovery Destroyed infrastructure EMT, police moving with various tasks Conference Motivating Scenarios Temporary meeting Moving presenter, stationary attendee with data sharing application Solution Mobile Ad Hoc Network Which routing protocol is best for each situation? 5
Protocol Selection Process: Overview Pre-design Time Processes Simulation / Measurement Behavior Extraction Design Time Processes Requirement Analysis Evaluation Input Environmental Information Application Requirement Output Recommended Routing Protocol 6
Protocol Selection Process Pre-design Time Simulation/Measurement Gather information about performance characteristics of candidate protocols Affected by parameters Topological parameters Influence of node movement in a topology: mobility model e.g., Random Waypoint Mobility Model (pause time, max. speed, node density) Traffic parameters Effect of varying data traffic e.g., avg. payload size, data sending rate, src. density Simulate extensively over parameters Provide the tool with the results as DB of input 7
Protocol Selection Process Pre-design Time Behavior Extraction Estimate generalized tendency for protocol performance Abstract the results of simulation/experiment Least square data fitting method Protocol behavior model Analytic model on protocols performance metrics over various parameters Present in tool before design time 8
Prototype Protocol Behavior Model Pre-design Time 3 Performance Metrics over 2 Parameters End-to-end delay, Throughput, Packet delivery ratio # of traffic sources, pause time with RWP mobility model Multiple Regression on Simulation Results 9
Protocol Behavior Model in the Scenario pre-design time Multiple Regression Result e.g., AODV PDR Analytic Model 1 2 6 2 7 3 10 = 9.27 10 + 2.78 10 p 1.46 10 p + 2.49 10 p 1.37 10 p 1 2 3 3 5 4 + 2.84 s 2.74 10 s + 3.12 10 s + 4.35 10 s ( p : pause time, s :# of traffic sources ) Estimate a protocol s performance with given parameters 4 10
Protocol Selection Process Design Time Requirement Analysis Requirement Analysis Takes inputs Environmental information: description of target environment Application requirements: conditions for achieving application s goal Generates outputs Environmental parameters: operational parameters, traffic parameters, mobility model type, node movement parameters Performance metric requirements: quantity of performance metrics with minimum or maximum Priority information: relative priority on application and performance metric Maps or interprets Mapping or Interpretation of inputs to pertinent parameters Effective interface required 11
Scenario Requirements Design Time Disaster Recovery Scenario Environmental Information 50 first responders following the random waypoint model Communication with each other maintaining 20 connections Application Requirements 4 applications operated in the network Voice communications: codec for voice Command dispatch: highest priority, delivery guarantee Location information exchange: lowest priority among applications Snapshot transfer: lowest priority on delay Other Information 512 bytes UDP packet Candidate Protocols Reactive: DSR, AODV Proactive: DSDV 12
Scenario Protocol Selection Design Time Environmental Parameters Performance Metric Requirements # of mobile hosts: 50 Traffic type: Constant Bit Rate Packet size: 512 bytes # of traffic sources: 20 Movement model: Random Waypoint Model Priority Information Mobile hosts Max. speed: 1 m/s Pause time: 100 ms 13
Protocol Selection Process Design Time Evaluation Determine the most appropriate protocol Compare expected performance of each protocol 1) Look up protocol behavior model with parameters 2) Translate application requirement into preference values Preference value: normalized deviation between simulated performance and given performance requirement 3) Combine the priority of performance metrics for each application 4) Combine the priority of applications for each protocol 5) Calculate total sum of weighted preference values 14
Scenario Protocol Selection Design Time Evaluation Result Recommended protocol for disaster recovery scenario: AODV 15
Conclusion Novel design tool to support a mobile application deployer Selection of the most appropriate routing protocol for a target deployment Protocol behavior model with several parameters given by user Future work Rapid Deployment of MANET applications Analytical protocol behavior model Free from time-consuming simulations Effective requirement description Efficient combination of requirements with protocol behavior model 16
Questions? E-mail: hopelist@mail.utexas.edu MPC URL: http://mpc.ece.utexas.edu 17
Simulation Simulation Environment Ns-2 ver. 2.29 10 times for each setting Parameters 18
Protocol Behavior Model Pre-design Time Meaning Analytical capture of S/W designer s concern about protocol characteristics under an environment Input Operational environment information: topological parameters Traffic information: traffic parameters Output Protocol characteristics: performance metrics Method Empirical study followed by regression method Analytical analysis of protocols 19
Protocol Behavior Model in the Scenario AODV behavior model over pausetime, # of traffic sources 20
Protocol Behavior Model in the Scenario DSR behavior model over pausetime, # of traffic sources 21
Protocol Behavior Model in the Scenario DSDV behavior model over pausetime, # of traffic sources 22
Protocol Selection Process Design Time Evaluation Determine the most appropriate protocol Compare expected performance with application requirements Preference value: how favorable w.r.t. performance metrics Preference value table Weighed sum for j-th application with k-th protocol Total sum for the system with k-th protocol 23