Optimizing Task Allocation for Edge Micro-Clusters in Smart Cities

Current urban technology trends like Internet-of-Things and 5G require ultra low latency compute resource to be distributed liberally at the network edge. We characterize and advocate the need for heterogeneous edge micro-clusters; these are pragmatic, low-power, low-cost, minimal footprint units that can provide sufficient resource for typical edge compute applications in smart cities. However, to make best use of heterogeneous edge micro-clusters, we require resource management techniques that are both efficient and effective. In this paper, we report on an empirical study to demonstrate that mathematical optimization (in particular, mixed integer programming) for resource management is appropriate in terms of overhead, also highly effective for executing batch-arrival workloads in smart city use cases.