23. CUDA Accelerated HAPO (C-HAPO) Algorithm for Fast Responses in Vehicular Ad Hoc Networks

With an exponential growth in the number of vehicles plying on roads in metropolitan cities, there is an urgent need of means for controlling the traffic congestion. This congestion results in increase of travel cost, travel time and pollution that have a substantial impact on both the health of people and the economy of the nation. For the reduction of congestion, one needs a routing algorithm that is able to detect the congestion in advance and helps in avoiding the congested routes. To improve the problem of congestion, many algorithms have been proposed by researchers. As the speed of vehicles is very high, the algorithm needs to compute the results in minimum possible time. Hybrid ant particle optimization (HAPO) algorithm is being used in the literature to choose the best route by leaving the optimal but congested path during peak hours in vehicular ad hoc networks (VANETs). In the paper, we are proposing a CUDA accelerated HAPO (C-HAPO) algorithm in which all the phases of HAPO algorithm are parallelized in order to provide faster computations by using the harness of compute unified device architecture (CUDA) on graphical processing units (GPUs). The proposed C-HAPO algorithm is able to speed up the computations of the existing HAPO algorithm significantly with respect to its counterparts under heavy traffic conditions and decreases the travel time for commuters. An implementation for the algorithm has been carried out on the NVIDIA architecture used with CUDA toolkit version 7.5 on a real-time northwest Delhi map obtained through Google Maps.