Greedy algorithm for dynamic allocation of intelligent services in vehicular edge computing

In vehicular edge computing, real-time processing of requests along with the mobility of vehicles between cells necessitate dynamic allocation of requests to servers so as to maximize the number of timely responses to the service requests. The best existing methods still have high execution times in large-scale scenarios, and thus they cannot be executed in short intervals, resulting in a significant deviation from the optimal allocation in such cases. We propose a greedy algorithm that, while maintaining high execution speed, can provide close to optimal solutions. Evaluation results of the algorithm, conducted through simulation, demonstrate that compared to the best competing method, our approach provides 49 percent more timely responses to vehicle requests. Additionally, its deviation from the best possible allocation obtained through MILP is only 6.8 percent.