Fog computing (FC) is used to handle Internet of Things (IoT) tasks instead of cloud, improving the quality of services (QoS) required by several applications. However, one of the limitations for IoT applications is availability of constant computing resources on fog computing servers, as transferring massive amounts of data created by IoT sensors would result in a rise in computational overhead and network traffic. In this paper, proposed the improved particle swarm optimization (PSO) algorithm with simulated annealing (SA) algorithm. The suggested improved algorithm can enhance the model’s convergence speed, schedule the virtual machine (VM) task, and prevent from the dimensionality problem, and user application has been dispersed into many tasks that have identified sequential maintenance connection. The proposed model is comparing to the existing models and it achieves the better metrics in various no. of tasks are 0.06 degree of imbalance (DIL) in 1000 no of task and 76 (s) of makespan (MK) in 1000 no. of tasks. The existing models include task offloading for IoT-based applications in FC ant colony optimization (TO-ACO) and task offloading in FC for using smart ant colony optimization (TOFC-SACOA).