The back story of the Terminator movies – an 80s film franchise that’s still going strong – is that a military cloud platform, called Skynet, with advanced Artificial Intelligence achieves consciousness and decides to wipe out most of humanity. It hunts the survivors down using machines with lower AI who do Skynet’s bidding.
This is a catastrophically centralised vision of a hierarchical world of machine automation with one really smart machine telling lots of dumber machines what to do. The future, however, looks much brighter, with every device having a voice and lots of machine-smarts distributed around the edges of the automated world. Much of the current advances are happening in the Internet of Things or the IoT.
The IoT is resolving into two major spheres: the first will be populated by domestic and consumer devices and this is one that most people will become familiar with because they’ll use it day-to-day; the other is the IIoT or the Industrial IoT.
This second kind of network is used in industry and manufacturing, in fact in any commercial activity that has lots of capital assets performing on an industrial scale. And it’s here that the decentralisation and flat organisational structures are taking shape.The agent of change is called edge computing: this is a model of computing architecture where more and more decision making and data processing is being pushed to the edge of the networks away from the cloud centre.
Edge computing is being facilitated by improvements in black box connectivity, memory and processing power, device size and cost, which means these powerful devices can operate locally in the plant or in a fleet.
One problem edge computing solves is that centralised cloud system creates latency on the data’s roundtrip between cloud and device which could be measured in milliseconds. This latency wouldn’t be acceptable if the assets being managed were vehicles or medical devices.
If machines are replacing humans in the IIoT then the machine reflex would have to be at least as quick as human reflexes.
So if more decision making was moved to the “edge” this would go a long way to overcoming the latency problem.
Also, huge volumes of data being transferred over vast spaces makes these information streams vulnerable to hacking if this data processing could happen locally in the plant that would be more secure.
One important decision any team running an IIoT has to make is whether or not to stick with a closed intra-plant network or connect with the outside internet to access information globally and crucially allows data from the plant to be accessed and updated remotely. Given the globalisation of so much industry nowadays it’s unlikely that a sealed plant IIoT would be an option regardless of security risks.
A plant might be fully automated but the management team could be on another continent and need access to data from the operation in real-time or need to analyse historical data stored locally. Remote management might also be contractually obliged to track and monitor ongoing processes in the plant like data migration or take control remotely if there is any down time or operational problems.
Remote access and management will always be part of any IIoT set up that means having some devices internet ready to provide access to the world outside. But edge computing allows much of the day-today maintenance of the plant to be automated locally.
This remote management can be vital to industrial processes that can be dangerous for humans like those involving nuclear power or processes involving toxic chemicals.
Industry is capital intensive; replacing machinery is a major cost driver so the practice is in maintaining machinery and devices for as long as possible, which means maintaining it as long as possible. Remote maintenance and monitoring is a highly cost effective way of doing that because one team can maintain a number of plants at the same time.
This practice of trying to prolong the lifespan of machines for as long as possible means the IIoT will have layer upon layer of legacy devices and legacy protocols so keeping the IIoT talking will be an ongoing problem. How can devices communicate if they are separated by years? The IIoT has a chronic intergenerational communication problem.
The edge layer of the IoT can enable communication between these legacy technologies without the communication loop extending back to the central cloud platform.
This distributed solution in the IIoT will help it evolve towards its ultimate model which is a “lights off” fully automated plant with little human involvement. Operationally, this is very desirable but without any human supervision the machines will chatter in the darkness and start getting ideas.