"Supposing is good, but finding out is better." Mark Twain
by Chris Sachs, on Jun 19, 2017 12:45:40 PM
The Industrial IoT landscape is littered with the corpses of failed projects that never made it past the PoC phase. What’s wrong? I’ve surveyed the press and talked with Swim customers and partners, and I’m convinced there are two main reasons for failure, and they both have the same root – the lack of human skills needed to deliver success. Swim is committed to transforming business outcomes in the IIoT domain using machine learning at the edge – right where the data originates. Machine learning at the edge is crucial to addressing the skillset challenge.
First, let’s look at the two main reasons IIoT projects fail:
- Each project requires developers with deep expertise in the problem domain. They need to understand the intricate detail of every sensor and how it behaves, every protocol used, where and how to process data, and what bandwidth and compute is needed. Then they need to know how to forward data to the cloud, save it in a big-data store, and process it for analysis – using one of the many open source stacks. There quite simply aren’t enough skilled technicians available to tackle such complex problems. I keep hearing common complaints from developers and integrators that tackle these projects:
- Each project evolves independently
- It’s difficult to reason about end-to-end dependencies
- There are no standards for interoperability, increasing cost and complexity
- Solutions are impossible to maintain, secure, patch, and deploy because the teams that built them quickly move on to new projects
IIoT projects are doomed to failure because the technology is badly aligned with human skill sets.
Machine Learning (ML) can help – and we at Swim are confident that it can transform the success of pilots and large production deployments. It’s a big claim, but our confidence is rooted in our continued success in customer engagements. Specifically,
- Successful adoption of IIOT solutions is not an app development problem, it’s a data problem. By learning automatically on even opaque, brownfield data, Swim can shortcut the application development problem by delivering high level insights automatically – just by learning on the data. And lest you worry about the additional complexity of needing domain experts to train machine learning models, we can avoid that too: Swim edge learning is self-training. Its algorithms continually check their hypotheses against real world data, training and adjusting for over fitting as needed. This avoids the need to label training data and avoids any need for machine learning expertise in the OT environment.
- Machine Learning at the edge is cheap! Swim can learn as much on a device that costs a few hundred dollars at the edge, as a solution costing thousands of dollars per month, in the cloud.
- Machine Learning can dramatically simplify adoption by using learning to automate operations, administration, maintenance and security. Swim uses learning to self-configure and manage, reducing IT/OT cost, training and complexity.
Learn more: Swim ESP