12th International Conference on Ambient Systems, Networks and Technologies (ANT) / 4th International Conference on Emerging Data and Industry 4.0 (EDI40), Warszawa, Poland, 23 - 26 March 2021, vol.184, pp.364-371
Geocast routing aims the transmission of a message to a community of vehicles in the same geographical area. The GPS information of vehicles, especially taxis, in a city gives reliable information about the traffic pattern of a city. With the help of this information smart models can be created to use when making routing decisions. In this paper, a smart geocasting protocol, GeoAKOM, is introduced and implemented by using the GPS data of taxis in Bursa city, Turkey. In GeoAKOM, the mobility pattern of city is analized in two levels as in GeoMobCon introduced earlier. In the macroscopic level analysis, the city is segmented into clusters using k-means clustering and optimal paths for a message is generated using these clusters. In the microscopic level, mobility based likelihood value is used to make routing decision. The message is transmitted if the encountered vehicle has a larger mobility based likelihood or contact history based likelihood value than the carrying vehicle as in GeoMobCon. Moreover, in GeoAKOM, the next cluster of a vehicle is predicted with the help of all k-order Markov model (AKOM) by using GPS records of all vehicles. Mobility based likelihood calculations are performed by using this prediction. With these features, GeoAKOM provides a robust solution for the vehicles which do not have large GPS record history and which are relatively new in the system. Simulations on real world taxi GPS data are compared for various number of clusters in terms of delivery ratio, average delay, and hop count by using the ONE simulator. Simulation results show that, GeoAKOM provides promising results when compared with GeoMobCon and the First Contact routing algorithms. (C) 2021 The Authors. Published by Elsevier B.V.