Cognitive IoT Systems via Adaptive Swarm Intelligence

We present a novel paradigm called Adaptive Swarm Intelligence ASI where heterogeneous devices or agents engage in collaborative swarm computing for robust and adaptive real-time operation. ASI, a paradigm inspired by the collaborative and decentralized behavior of some systems in nature, finds application in a myriad of scenarios, in domains like the IoT, mobile computing and distributed systems. Examples include network cybersecurity, connectedautonomous cars, and other types of unmanned vehicles, like intelligent drone swarms. This is by no means an exhaustive list but it gives an indication of the many and diverse domains that can benefit from this paradigm. This paper presents a specific ASI case study for cooperative sensor fusion in prospective connectedautonomous vehicles, which constitutes the driving application of the IBM-led Efficient Programmability of Cognitive Heterogeneous Systems EPOCHS project under the DARPA DSSoC program. Due to the magnitude of EPOCHS, we focus on one specific piece of our project the EPOCHS Reference Application ERA for multi-vehicle sensor fusion. We show characterization results on a x86 system that allow us to draft preliminary conclusions about ERAs performance characteristics and real-time needs. The paper briefly describes EPOCHS roadmap and future work.