TRAP: task-resource adaptive pairing for efficient scheduling in fog computing

摘要

The proliferation of smart services and devices leads to connection delay and high traffic load in networks connecting cloud computing to end users. Fog computing resolve these issues by bringing cloud services closer to end users and consequently delivers better service quality to requested tasks. However, assigning resources to tasks is challenging due to complex and strict quality of service requirements. Moreover, concurrently optimizing multiple objectives such as network usage, energy consumption and delay increases complexity of the scheduling process. In this regard, we investigate optimal task-resource pairing for efficient scheduling to simultaneously minimize delay, cost and energy consumption. The problem is modeled as a multi-objective optimization problem to efficiently schedule latency-sensitive tasks on fog resources. The proposed solution consists of three main key components, viz a batch system, a ranking, and a priority method. The batch system exploits ranking and priority methods to optimally pair tasks and fog nodes. The significant advantage of the presented approach is the reduction of the search space through batches. The proposed mechanism is implemented on iFogSim simulator and results show that the proposed system significantly reduces delay and energy.