LAMEN: Towards orchestrating the growing intelligence on the edge
Ahmed Salem, Tamer Nadeem · 2016
Exploiting the network's edge is trending nowadays due to the evolution end devices, e.g., IoTs and smartphones. In this work, we present LAMEN1, an initial attempt to execute services at the network's edge closer to data sources. Unlike crowdsensing approaches, which use end devices as data collectors wasting the bandwidth in transmission to cloud servers for processing. LAMEN proposes a layered architecture where devices are grouped by their location. Each group elects a head node, i.e., mediator, to control service delivery and execution. Further, mediators names and organizes devices within a group in terms of resources they hold, this enables efficiently locating the right resources for service execution. We envision pushing light weight services that end devices can execute over their generated data to have the following benefits: First, avoid sharing user data with cloud server, which is a privacy concern for many of them; Second, introduce mediators as local cloud closer to the edge; Third, hide the user's identity behind mediators; and Finally, minimize communication overhead from sending raw data for processing to transmitting processed information. Our evaluation shows LAMEN to have an optimized resource lookup and service assignment techniques. Moreover, scalability in handling networks with large number of devices.