Efficient Welfare Maximization in Fog-Edge Computing Environment

A Fog network comprises a group of peer nodes where all the peer nodes take part in a collaborative network for sharing computing and storage resources. In a collaborative FOG network, the major challenge for a task node is to efficiently select and offload the tasks to helper nodes. The collaborative scheme in general works with a reasonable incentive policy for sharing the load of others or helping others where revenue and profit are involved in task execution. We formulate the problem of welfare maximization which is profit maximization and minimization of disparity in profit between sharer nodes, and formulate the said problem in integer linear program for understanding the problem clearly and solved using standard LP solver. We propose three welfare scheme heuristics for load sharing considering the local and centralized solution approaches and compared the performance of these for different data sets. The proposed welfare schemes that use task selection for load sharing which consider by properly balancing the effects of execution time, revenue and data size of the task for selection in sharing the tasks with others for welfare maximization. We heuristically calculate the weight of the task parameters in selecting tasks for sharing with other sharer nodes. Based on the experimental result, the proposed weight-based task selection in load sharing achieves an average of 96.23% profit as compared to the optimal amount of total profit using ILP based profit maximization. Also, weight-based local reservation scheme for load sharing achieve significantly better results in both maximizing total profit and minimizing the disparity in profits of the sharer nodes.