What's The Best Way To Speed Up IoT Product Development?

RE: What's The Best Way To Speed Up IoT Product Development?

eshein — Tue, 13 Dec 2016 20:56:17 GMT

Hello, I am a freelance writer for a new publication called enterprise.nxt. I’m working on a story looking at strategies for speeding up IoT product development. I know this a big concern for enterprises and municipalities. I read these comments and wondered if anyone would allow me to quote them? If so, I'd like to hear about examples of rapid IoT projects you have worked on and how you have accomplished a quick turnaround. Please contact me @eshein@shein.net. My deadline is mid next week. Thanks, Esther

RE: What's The Best Way To Speed Up IoT Product Development?

Former Member — Sat, 10 Sep 2016 03:03:17 GMT

I agree with [mention:4966e6ad97f64dac9ea8919abb220e1d:e9ed411860ed4f2ba0265705b8793d05] to reuse as much as possible. And don't skimp on security because you never know how much privacy, authentication, and auditing that you'll need in the future - start your IoT apps with security in mind from the beginning.

In a github project, I have included some Python-language SSL communicating partners ("alice" and "bob") that might have interest for you. All included code runs on a variety of machines.

Feel free to reuse: Do-It-Yourself Certificate Authority on a Raspberry Pi (link updated 2016-09-10).

RE: What's The Best Way To Speed Up IoT Product Development?

DAB — Thu, 08 Sep 2016 18:00:46 GMT

The best time savers are to reuse as many things as possible.

That means find and IOT implementation that is robust enough to be used with many different sensors and devices.

After that, you should only need minor software changes to use different data.

End result, fast to market, safe operation, low cost.

DAB

RE: What's The Best Way To Speed Up IoT Product Development?

Robert Peter Oakes — Thu, 08 Sep 2016 16:00:54 GMT

One of the things I find interesting is that many products try to do way too much on the edge. There is no need for this, If IoT devices are to be adopted, cheap and easy to deploy they need to be mostly dumb. simply collect data and report it up to a higher node where perhaps some control logic may exist or at a minimum would concentrate the data for further upload to a more intelligent Business processing system

With this in mind, having an edge node that is based on a simple platform, an ESP8266, Sierra Wireless, Atmel sub Ghz etc. This would have all the basic capabilities to gather data and upload it to a hub of sorts. these nodes don't even need to know what that data is, an node could be collecting flow, temperature, humidity, light, pressure... who cares, to it this is simply data and as such bears little security risk. Another edge node can be used for control rather than monitoring and as such is also basically unaware of its true purpose (Sometimes it needs more knowledge) so for instance it can be controlling a relay, a DAC, or simple logic outputs.

The communications can be SSL based to ensure some sense of security or leverage other encryption techniques so easily built into many MPUs these days.

The point is there cheap and easily adapted to different sensors or outputs. and can use WIFI, BLE, Sub Ghz etc. imo it is easier if there able to leverage a TCPIP stack or be close enough to a hub to leverage subGhz

Next there would be the hub, this can be a bit more advanced but still no more complex than a Raspberry PI or Beagle Bone Black. Running open products like Node-Red and MQTT with perhaps a DB like sqlite would allow historical data to be locally collected and retrieved on a schedule.

This hub can choose what to share (Programmed/configured) and what to not share, Running Node Red it can also easily manage local workflows for security, process control, office Automation etc with ease. It would make sense of the data and be able to understand that the data is a temperature or pressure or a door status etc and take appropriate actions.

The MQTT part would allow the node to be used by all the edge devices to either publish or subscribe to appropriate data streams and act according to received notifications.

This is becoming a post all of its own rather than a simple comment so ill stop there and perhaps make a blog / video all about this.

I guess my primary point here is that I tend to see most IoT project being made way too complex at the edge and having everything including the kitchen sink thrown in. This immediately makes it slow to adapt and slow to design and deploy. Keep it simple and relatively dumb at the edge and increase intelligence as you work toward the center