Building the Real Skynet
Skynet, the self-aware artificial intelligence system in the “Terminator” movie series is a metaphor for this age of automation, and the real fear and fascination that we have with systems that will free us or make us slaves to their commands.
Skynet, which we wrote about earlier this year, is also the original name for Chris Matthieu’s company, Octoblu, which is developing a machine-to-machine network, or mesh of IoT networks, that share a common API or communications protocol. The original Skynet open source project is now managed by Matthieu and is independent of Octoblu.
At the Defrag conference earlier this month, Matthieu discussed how using Octoblu, devices can discover, query, and message other devices on the network his company has developed. The New Stack Founder Alex Williams attended the conference and made this recording.
For other recordings, visit the podcast section of The New Stack.
Matthieu is developing what he describes as an unstoppable learning mesh network that speaks many protocols where disparate devices communicate via variety of protocols, something like a machine-to-machine instant messaging (M2M IM) neural network.
Here’s a bit more about the project he is building:
- The single Skynet API supports the following IoT protocols: HTTP, REST, WebSockets, MQTT (Message Queue Telemetry Transport), and CoAP (Constrained Application Protocol).
- Skynet is open source and both a multi-protocol and cross-protocol platform where any device that speaks the language is be able to communicate with each other.
- Every single resource on the mesh network has a UUID — a 36 character device credential that is associated with that mesh network which has its own UUID, eventually making it a network of UUIDs.
- The platform also needed to be able to support discovery and message ability of UUIDs to other UUIDs, where it should be able to search for an array of UUIDs based on the given search criteria.
- Security is flexible enough to be controlled by the UUID that can discover and send messages, and also indicate which UUID can subscribe or be configured in a whitelist and blacklist fashion — all integrated into the platform.
- An API is positioned on top of every native protocol known, where the same style of the REST API was applied to every protocol to share the same communications platform and interact through mesh networks.
- Neural Processing Units (NPUs) are the chips inside every Skynet device that allows Skynet to control the device and connect with it as required.
- The Octoblu Skynet operating system runs on Arduinos, Spark devices and Pinocios where by adding more microcontrollers that is basically a sketch with 32K RAM enables the Arduino to come online and request a UUID from the mesh network that is multicasting and connects to it.
- Through a mobile Skynet network anyone who can connect to the mesh network can also control it.
- Skynet has about 30 APIs built into the platform including enterprise API’s with a browser based designer to drag and drop it. As events and messages flow through Skynet they pulse in real time talking to APIs that also provides a login user with a UUID.
- This has resulted in a platform that brings everything together with simple drag and drop browser and APIs on top of different protocols.
- If there is a device that doesn’t speak the language, there is a gateway with a plugin to include the device into the mix of protocols.
- Cisco is estimating that by the year 2020 there will be 50 billion connected devices to the Internet which is an exponential growth in connected devices.
Alex Williams edited and contributed to this story.