Edge computing has a very bright future – it is absolutely inevitable.
If you are unconvinced, we invite you to read until the end.
For the sake of this post – we will use edge and fog computing interchangeably. This is becoming a less controversial position based on a recent exclusive interview with Chuck Byers, OpenFog CTO.
The OpenFog Consortium announced at an IoT Evolution Expo keynote in Fort Lauderdale, Florida last month, that their merger with the Industrial Internet Consortium or IIC was finalized. Both Carl Ford and Ken Briodagh with IoT Evolution have interesting insights on this merger, as does Matt Vasey, Chairman and President of OpenFog Consortium.
Suffice it to say, this combining of resources makes a great deal of sense as it helps the adoption of edge/fog computing. Here are some of the benefits and issues of fog-computing and OpenFog contributions.
Chuck discussed that apps should run in the cloud if possible but there are often impediments. Latency could be an issue – by the time a signal gets sent from a sensor to the cloud, is processed and sent back to an actuator, it could take 300-800 milliseconds.
This is not acceptable for applications such as an arc welding robot or AR and VR where anything over a seven-millisecond delay could cause nausea. Another great example is the antilock-braking system (ABS) of an autonomous vehicle – lives could be on the line and every microsecond counts.
These are all applications where edge-computing is crucial. By localizing processing power and storage, greater speeds can be achieved.
Cost of Bandwidth
In other cases, high-bandwidth sensors could cost companies a great deal in transport costs. A camera on an oil rig, for example, could cost $10,000 per day. Other applications thousands or millions of sensors in remote locations could also be cost-prohibitive.
Fog could solve this problem by processing the sensor data locally and only reaching out to the cloud when needed. For example, camera footage can be processed locally and any anomalous frames could be sent for further analysis to the cloud. Likewise, mega-sensor data could be processed nearby, looking for issues which would trigger a snapshot being uploaded to the cloud for further analysis.
IoT devices have become major targets for hackers and other malicious actors as they can be less secure than a phone or PC. Since you can’t do strong cryptographic processing locally, you need a gateway nearby to ensure the network is free of malicious activity.
As more IoT solutions become mission and life-critical, it’s import to ensure networks are free of ransomware and DDoS activity. In addition, there needs to be secure onboarding and software updates.
The OpenFog Consortium has rules allowing for a hardware root of trust which is a circuit inside the processor which enables a trusted module which in-turn enables a trusted compute environment.
Using this solution, you are also able to enable secure software updates, free of malware as well as ensuring rogue devices can’t be added to the network.
Getting back to the above example of ABS in a vehicle, there is no room for error. If a car needs to stop, the excuse can’t be, you have no connection to the cloud. In such cases, a mobile node must reside in the vehicle – or near the sensors and actuators responsible for the critical control algorithm. “This gets us closer to the five nines reliability needed,” he exclaimed.
In addition, multiple nearby nodes for compute and storage can increase redundancy if one edge-node fails. Think of it as distributed RAID.
Between OpenFog and IIC, there are 28 test beds between them which allow partners and members to take reference architectures and test them in the real-world. This allows companies to tune existing products and roadmaps.
Security Part 2
Our discussion then proceeded to DDoS attacks which have been facilitated in the past thanks to vulnerable IoT devices.
We posited that companies may not care if their devices – which are generally functioning normally, are also being used to launch DDoS attacks on other companies.
Chuck thinks over time, lawyers will begin lawsuits against companies whose devices are being used for such attacks. It is likely companies who are not adequately securing their IoT end-points will have to face legal and/or financial consequences.
This is why as Chuck explains, they take security so seriously. By protecting device onboards, they minimize or eliminate spoofing attacks. By controlling software updates, they minimize the potential for malware getting installed on cameras or sensors.
They are also working on a distributed ledger to help with these updates – not exactly blockchain but the same concept.
We asked about anomaly detection on the network and if Chuck sees this as a benefit of fog and he explained that all types of AI are better-enabled by edge-computing. He said in addition to spotting security-related anomalies, machine learning and deep learning can be used to look at vibration and temperature variations in a predictive maintenance capacity.
In other words, the AI can monitor the health of the network and the plant.
The combination of IIC and OpenFog makes both organizations much stronger. This is in-part due to the different focus of the groups. IIC focuses more on near-term implementation while OpenFog is more academic. In addition, OpenFog brings thought maturity on secure computing including management and orchestration. In addition, they have a tight relationship with IEEE meaning their concepts can and have become standards such as IEEE 1934. Chuck said, “This provides extra comfort for people, they know the architectures are solid, correct and stable.” He continued, ” They don’t have to worry about it getting pulled out from under them.”