English | EN 
This series of articles talks about CBIoTS and Cognitive technologies. They are produced directly from the event IBM World of Watson 2016.
Introduction
In early 2014, I wrote an article here entitled “Managing New Complexity”, where I promoted reflexivity among semantic and type theory as a candidate “mechanism” to better understand and govern complex systems, and more precisely socio-technical systems, the ones we are dealing with every day as IT professionals.
Today, I’m proud to announce to the SogetiLabs community that we succeeded in building an IoT platform available in the Cloud, On-Premise or Hybrid that allows reification (structural) and local introspection (behavioral) aspects of reflexivity. The main goal for our framework is the control/command of Ultra Large Scale System (ULSS). For us, this scale of ULSS is above a million machines, and we named our platform Cockpit for Big Systems (CBS). Today we are at the scale of Very Large Scale Systems (tens of thousands machines), already experienced with an on-premise architecture. We know that the scale of ULSS is reachable by multiplication of instances of our platform and some optimizations. To reach this scale we decided to make an extension of CBS to IoT. We call it CBIoTS (pronounce it like cybiots).
Since June 2015, we are working with a French-tech start-up named ‘Drotek’ on an ambitious project and very interesting use case for CBIoTS about smart farming. Drotek apecializes in robotics, UAVs and embedded systems in general.
We both were in Las Vegas to present CBIoTS and ISPF (Intelligent System for Precision Farming) at the IBM World of Watson 2016 event that took place from October 24th to October 27th, and spoke on the subject: https://myibm.ibm.com/events/wow/all-sessions/session/2052A.
Problems
When you try to face complexity, you will come across on your way plurality, variety, hierarchy, feedback phenomenon, speed of changes, and the most terrible one: emergence of behavior. Any combination of those is possible, and results in an unpredictable state for the system you are observing. That means no pattern, no model, and no optimization or calculation is possible on complex systems with an exhaustive and exact perspective.
CBIoTS Architecture
Technically, our platform is based on the Multi-Agents Systems (MAS) paradigm. Each instance of our CBSIoTS platform (Cockpit for Big IoT Systems), is capable of managing 30 000 machines. Devices like Desktop/Server, Laptop, Nanocomputer, Tablet or Smartphone, with almost all common Operating Systems (OS): Unix, Linux, Windows, Android, Raspbian, Yocto, and soon iOS and MacOS. This high level of interoperability and abstraction of what is an OS was obtained by developing our agent in C. That gives it also high performance and low footprint memory (~300KB on linux, ~800KB on Windows). Our agent can even be used on real-time machine without disruption. The CBIoTS platform is now in the cloud, so you can multiply instances of the core engine, but you can also store your results on-premise, in common relational, document or graph databases; depending on your use cases. To deal with very high stream of data from million of IoTs, we are currently implementing a high streaming entry point for CBIoTS, based on open source components.
Coming up this week…
- More details about CBIoTS
- How useful the multi-agent paradigm can be in dealing with complex systems
- Some use cases : done and in development
- MAS and Cognitive technologies : the winning couple
- Some perspectives
Image courtesy: http://siliconangle.com/
