All the devices (TVs, fridges, washing machines, etc.) connected into Home Edge Network are considered Home Edge Devices. Some of the devices, possessing significant H/W resources and capable of running containerized applications, are called Home Edge Nodes. Tasks performed on devices of Home Edge Network are managed by Home Edge Orchestrator software. These tasks are called Home Edge Applications. They are assigned to specific Home Edge Nodes. As of now, Edge Orchestrator incessantly scans the Home Edge Network, forming lists of attached devices, and forms ratings of performance for them. Performance ratings are basis for deciding on which devices Home Edge Applications are to be run. If Home Edge Orchestrator cannot find device with rating higher than its own device rating, it will start the Home Edge Application locally. Home Edge Networks support distributed applications consisting of interacting Docker container instances. Docker containers offer quick deployment, easy management, safety and hardware independence. Applications that can run on Home Edge Network are deployed from cloud-based centralized storage as needed. Assortment of applications is determined by tasks that must be solved by Home Edge Network.

As an example, on a fig. 1 Home Edge Orchestrator software runs on AI Speaker. It must start a distributed image recognition software that obtains images from smart camera connected to Home Edge Network and processes these images on other devices connected to network. To do that, Home Edge Orchestrator checks resource availability and starts image processing application on suitable devices. To obtain image processing application, Home Edge Orchestrator connects to cloud storage, obtains needed Docker containers, deploys them to Home Edge Nodes and starts these containers, giving them input data obtained from Smart Camera. After applications finish running, Home Edge Orchestrator collects the results and presents them as needed.