For those of us outside the hardware and software industries, white papers on Information Technology can be opaque and tedious.  Fog computing is different because it quickly makes sense.

Fog computing is like Cloud computing, but it takes place in the things around us, such as in sensors and controls embedded in electrical grids, or in roadways and in connected vehicles (driverless cars and trucks).

Cloud computing is a highly successful way of sharing computer hardware and software via the internet.  The Cloud computers are located in large, cost efficient datacenters that offer on-demand computer power and the most popular business and industrial software applications.  The scalable computer capacity and the near ubiquity of fast Internet allows many customer applications to operate and share efficiency.

Some of the largest Cloud operators are IBM, Amazon, Microsoft, and Cisco-Google.  They carefully keep applications up to date and secure, relieving customers of those chores.  The Cloud’s connection via Internet is suitable for some applications (e.g. a monthly customer invoicing run), but for some applications it could be slow (e.g. multiple up bound streams of real-time video), or the Internet may be subject to intolerable latency.  Latency is not about the speed at which the bits flow, rather it describes the delay before that “flow” starts.

Some applications (e.g. widespread driverless vehicles) cannot tolerate noticeable latency. A driverless car traveling at 60 mph will move 17 feet in 1/5th of a second.  Latency of 1/5th of a second could misrepresent a car’s position to the driverless control center by one car length or by a lane and a half.  Compounding that with rush hour traffic and 1/5th second latency could cause a big pileup.  Low latency is essential.

Fog computing moves the computing and software closer to the place where data is generated and captured.  That places computing power amid the local “things” that generate data and that need to receive control signals.  Shortening the path between networked things in an application can allow the system designers to eliminate some of the latency and some transmission times compared with a Cloud computing and Internet approach.  Cloud computing is still relevant for some applications, but for the most time-sensitive, Fog computing is probably the smarter choice.

Fog computing networks are unlikely to replicate the 5th generation mobile wireless.  The outermost tentacles of the 5G network can be suitable as Fog network connections, provided the 5G network does not require Fog traffic to ride all the way to an internet node. Still, there are 5G applications for edge and fog applications.

The Industrial Internet of Things (IIoT) will contain many “factories” that use specialized computers, sensors and actuators.  An IIoT factory need not be completely located in one building.  It could contain numerous sensors and computers that control industrial processes, all of which might be located in dirty, hostile environments such as steel mills or in roadway bridges.  Those environments could require specialized computers such as the fanless, completely sealed computers, something not typically found in Cloud computing centers.

The smart electric grid is an example.  Our electrical supply system will need a mixture of smart appliances that can be told to adjust their power consumption, smart meters that limit the amount of power accessible based on supply and demand conditions, and control for production of power.  Some technical and network components in the smart grid will need extra security and low latency.  Other components will be less sensitive.  All can be tied together by Fog networking and computing in a city.

Fog computing can harmonize with Cloud computing and 5G.  Consumers and businesses will benefit from its carefully adapted computing and networking.  There is no point dragging data and control signals many miles away from where they are generated and needed to trigger action.  Fog computing is just a sensible way to meet the needs of specialized computing in the local area.