Traffic Estimation AndRealTimePrediction Using AdhocNetworks
Fatima Batool, Shoab AKhan · 2005
This paper presents theprocess ofdeveloping a multilaver feedJbrward neuralnetwork combined with abackpropagation algorithm for forecasting travel timeandtraffic congestion. Prediction oftravel timeandtraffic congestion based onpastandcurrent traJJic inJbrmation is notstraightforward duetoamongothers, the highcomplexity andill predictability oftra/fic process, incorrect observations anddiffirent datasources. Howeveritappears thatneuiral networks canbeexhaustively usedtosolve these problems. Thesystem isdesigned ontopofa meshbased communication infrastructure jbr the mobile nodestocommunicate. Communication network comprises ofmuiltiple networks i.e. VHF,UHF.Themeshbasedcommunication approach enables easydeployment ofthesystem inreal world. OLSRrouiting protocol isusedjfr establishing anadhocnetwork Jbrpeer-to-peer conimmunication.