Advances in coding have been relentless, particularly in the last 30 years as computation power increased to the point of making feasible much more sophisticated signal processing, fast and at affordable cost. 3 and 4G are examples of systems that leverage on the increased processing power exploiting new signal coding. 5G is not exploiting new signal coding but it leverages too on increased processing speed by making it possible to interact with multiple signal streams (on different frequencies and different coding scheme).
A similar thing has happened with optical transmission where laser beams have been modulated in different ways to increase their transport capacity.
One solution to increase the capacity of carrying signals by a light bean has been to use polarisation (both linear and circular) of the light. This is used, as an example, by television to carry 3D signals.
Now researchers at CCNY, City College of New York, are proposing a way to code signals carried by a laser beam (the one used in optical fibre communications) into "shapes" using special devices called q-plates.
They showed coding using 4 shapes but in principle one could add an unlimited number of shapes thus multiplying the capacity of the signal.
This could potentially increase the capacity of a single laser beam from the Tbps of today to the Pbps, a thousand fold increase. That would make present fibres good for several more years of traffic increase.