TQRAM: tolerable QoS-aware resource allocation modeling in IoT based on Petri-Net

In Internet of Things (IoT), load-balancing helps enhance resources utilization through efficient and fair task allocations between computing resources. On the other hand, a lack of load balancing means that some resources are under-loaded or idle, while others are overloaded. This problem will affect resources performance and the satisfaction of users and service providers is negatively affected. Since tasks need to be served by a set of resources, load-balancing towards optimizing and managing resource allocation can improve Quality of Service (QoS) parameters and maximize the task acceptance rate. Computational tasks are allocated to resources by the controller’s decision, taking into account the load-balancing aspect to achieve QoS. Consequently, one of the critical factors for managing resources in IoT efficiently and avoiding overloaded is load-balancing. In this paper, A load-tolerance technique based on load-balancing is also proposed to increase the task acceptance rate when the resource capacity is full. It is a graphical and mathematical model in resource allocation for load-balancing to evaluate and analyze the network performance. The effectiveness of the load-tolerance technique is modeled with Petri-nets to investigate the task acceptance rate. Petri-net design, by providing a graphical model based on mathematical logic, is a valid criterion for analyzing and evaluating QoS-aware workload allocation management. This paper deals with modeling and analyzing the optimal allocation of processing resources to arrived tasks using Petri-nets. This achievement is the use of Petri-nets to model the steps of computing resource allocation in order to improve resource allocation and QoS parameters. To the best of knowledge, none of previous work has applied Petri-net to resource allocation problem in IoT. Simulation results of the task acceptance rate of the proposed approach showed that it reached 86.2%, which increased the acceptance of more tasks by about 8.4% compared to other approaches.