Online Task Allocation and Scheduling in Fog IoT using Virtual Bidding

IoT applications increasingly exceed the available computing capacity at the edge. A 3-tier fog architecture may allow for low-latency compute-intensive applications. However, resource allocation in the cloud and fog, along with real-time application requirements, pose a number of challenges. In this paper, we examine the problem of auction-based resource allocation for delay-sensitive online IoT applications in the cloud-fog-edge architecture. We propose a multi-attribute double auction-based task allocation algorithm with delay-based pricing and bidding strategy. A novel virtual-bidding mechanism is designed to allocate resources to tasks that arrive in-between the bidding rounds. The price and winners are determined using McAfee and reserved price-based allocation, which also considers the perishable nature of fog/cloud resources. The proposed algorithm is implemented in NetSim and Python. We find a substantive performance boost in resource utilization when virtual bidding is used for online jobs instead of only using periodic auctions. Simulation results show an increase of 61% in resource utilization, 37% in user utility, and 79% in profit for fog service providers.