Virtual machine placement and workload assignment for mobile edge computing

As an extension to cloud computing, Mobile Edge Computing (MEC) provides computing capability at close proximity to mobile users to reduce service latency and improve users' quality of experience. In MEC, tiny datacenters, equipped with computing and storage capabilities, are located at the edge of the mobile network. Services in these tiny datacenters are supported by application-specific software instances, which are packaged in Virtual Machines (VM). We study the problem of VM placement and workload assignment for mobile cloud applications in MEC. We formulate a mathematical model to minimize the hardware consumption required by VMs for supporting given workloads in a multi-application scenario, while meeting heterogeneous latency requirements of different applications. Numerical results show that applications' latency requirement, MEC servers' hardware capacity, and users' request load have significant influences on overall hardware consumption, and MEC server utilization can be optimized by leveraging remote VM placement and workload aggregation.