designs

This page collects different design ideas to give you a starting point to testing different implementation of the DiPiNet. You will have to solve these issues:

• control, how to tell a display what to display and when.
• interaction, how to use the sensors to let a possible user interact with the system.
• configuration, how does your application know the physical location of the display (can you use the sensors? can use the field strength of the nodes?)
• robustness, how can you make sure that your system works without having to test as a whole? can you simulate your solution for example?

Possible designs and implementations:

standardized protocol based

Implement the Distributed Display Protocol (DDP) to control each node in the network. Try to build the sensor/event transport on top of the DDP. For configuration manually assign each node to a block in a virtual framebuffer. For robustness implement your solution in Contiki and test with Cooja. Assign a group member to each problem. Use standard-compliant DDP clients, like Cinder to control the network.

simplified roll your own

Implement a UDP multicast or TCP server on each node. Centrally control which node has to display what and when, discover by manual assignment, multicast ping or MDNS (available in Contiki and Linux). Limit your display capability to pre-stored images (this is also possible with the DDP). Implement a knock-detection on each node. Implement on Linux and test with multiple Edisons, use Wifi as the transport.

...