Nice post from Martin’s corner on the web about setting up IoT devices.
The typical end-user of your IoT products wants them up and running fast, with minimum hassle once they bring them home. They don’t want to be presented with initial configuration mechanisms/options they don’t understand, and their enthusiasm quickly diminishes if you make them struggle to get the product going.
A colleague once said “If you can’t explain how it works to your grandma in five minutes, you don’t understand it well enough”, I’d re-phrase that to “If the end user can’t get your product all configured and running in five minutes, you haven’t done good enough work on designing it”. That, of course, raises the bar for the developer. Ease of use comes at the cost of increased analysis/development/testing time, but all that pays off with improvement in customer experience.
When designing IoT hardware you are often balancing product cost vs ease of use. One of the old-school approaches to hardware configuration was to have DIP switches on your PCB, which of course only allows for limited configuration options (not to speak about the complexity for the end-user). Basic LCD graphical UIs with keys ease up this task, but add to the product cost. Configuration over serial connection or re-flashing the product with correct parameters is out of question for a product that claims to be user-friendly.
With the appearance of cheap, Wi-Fi enabled SoCs like the ESP8266 things rapidly changed. It is now possible to send Wi-Fi connection credentials to a non-connected ESP8266 through Espressif’s version of TI’s Smart Config, or use Wi-Fi Protected Setup (WPS) to configure your product’s connection with the simple push of a button. Alternatively one can simply have the ESP8266 start up a Wi-Fi access point with DNS captive portal configuration UI. This means you have a configured, Internet connected product within a minute of plugging it in.