Wireless-Fog Mesh: A framework for in-network computing of microservices in semipermanent smart environments

Large gathering events such as open air rock concerts and religious gatherings require a semipermanent communication and computation infrastructure for providing IoT-based smart solutions. In these places, the installation of wired network is not a cost-effective solution to provide connectivity. Wireless mesh network (WMN) presents itself as a quickly deployable and dismantable solution to provide wireless connectivity to IoT devices. In this paper, we have proposed a "wireless-fog mesh" computing framework for installation of microservices in network devices of WMN. This requires continuous monitoring of residual resources of devices and associated network characteristics. Our proposed fog controller exploits underutilized resources of mesh devices and network metrics for mapping a microservice to a fog node. This controller uses a DHT-based overlay of fog nodes (also acting as distributed MQTT broker) for disseminating resource profiles of fog nodes. The network metrics like path cost, routing loads, number of stations, and delay are used to implement a policy-based FNS algorithm for electing a fog node to run microservice. We have designed and implemented a testbed that consists of 10 OpenWrt-based Raspberry Pi3 node to form a WMN. We have deployed various microservices such as MQTT broker, air quality, web servers, annd video surveillance in fog nodes (mesh routers and access points) to test the FNS algorithm. The results establish that the FNS algorithm significantly outperforms the random node selection (RNS) algorithm and advocates the applicability of our wireless-fog mesh framework for semipermanent use cases.