Telecommunications Networks span the world and are the result of huge investment over many decades. Someone I spoke with recently told me that the investment needed is so huge (154B$ spent by the three largest US Telecoms between 2006 and 2010) that the rate of change is really slow, so I shouldn't expect any significant changes over a 25 year horizon. And yet, I counteracted, look back 25 years, to 1989. Cellphones were basically not existing, IP traffic was basically not existing, Apps did not exist as well as names as Google, Facebook, Twitter, .... Now almost 100% of traffic in the US is IP, and the overall world traffic is by large IP, cellphones are ubiquitous and have changed the way people communicate, Apps are in the millions... Indeed, 25 years ago it was a completely different world in terms of communications infrastructures and usage.
Looking ahead, OECD Countries plus BRIC countries (based on an EU estimate) are going to see a significant increase in infrastructures spending - for water (+100%), electricity (50%), roads (+30%), rail (+20%)- with the notable exception of telecommunications infrastructure where they expect a decrease of 70%! Something is brewing! Having spent over 40 years in a Telecom Company I don't like such a trend.
The continuous increase in performances of "communications components", i.e. transport capacity, processing capacity and terminals features goes hand in hand with a dramatic cost reduction. This stimulate the faster adoption of new equipment, devices and expand the market further stimulating innovation and progress. The expected decreased investment in telecom infrastructure is not heralding a slowing down in the overall capacity increase. In twenty years time we can expect a growth of performances in the range of 100,000, and a significant decrease in cost (this is more difficult to calculate since to the decreasing cost of the electronics you need to factor increasing cost of labour per unit but less labour overall, plus increased cost of civil engineering but a decreased cost due to radio replacing wires...). An investment decrease of 70% thus is actually an "increased investment" in terms of capacity deployment. Clearly, the decreased volume of investment will have its impact on manufacturers and telcos...
Hence, expect in the coming twenty years technology evolution addressing several areas, as shown in the picture with reference to 5G, something that is going to have a profound impact in the next decade and in the following one.
I would like to offer another side-thought.
Whenever I hear the word "network", because of my background, I tend to associate it to wires and Nodes (with wires sometimes being in the shape of radio frequencies....) and to technology. But if I were asked to look 20 years ahead other kinds of networks may come to mind, networks that we find in Nature and that have evolved over hundreds of millions of years: metabolic and neural networks.
Methabolic networks are an amazing example of order merging from chaos. Out of billions of molecules swarming in a cell you get an ordinate and coherent set of proteins making our machine work. There is a sort of reference manual, the DNA, but the magic of entropy decrease is made by the ordered chaos inside the cell. Each molecule is an independent entity, as each cell in a multicellular body is basically an independent entity that reacts to its boundary condition. It seems impossible and yet it works. Relatively few basic rules ensure a coordinated ensemble even though each component is basically independent of one another. The same situation occurs in the movement of school of fish or flocks of starlings.
I wonder if we will not have to resort to these models/paradigms of autonomous systems in the evolution of our network"ing" infrastructures as more and more players are playing the game and as terminals (smartphones, robots, IoTs) become themselves an integral part of the infrastructure, communication nodes rather than terminals. Their sheer number requires, and allows, quite different communications architectures and paradigms and I feel that 5G makes some steps in this direction.
Neural networks, the ones cramming our skull, have an amazing property: they process signals and they create emerging information, something we call "emotions" and "thoughts". As humans we surely experience them and based on experiments we can tell that many animals perceive these emerging information, although probably just a subset of them. The quantity of nodes (neurones) and connections (dendrites and axons) seems to affect this capability. Our communications networks along with the data carried and stored are reaching a thresholds where we can start seeing something similar to awareness and information emerging. Our inner neural networks behaves in such a way that it is not possibile, nor accurate, to separate communications from processing and storage. They are one and the same. Something similar may become true for the future communications infrastructures where processing at the edges may morph into storage at the edges and where communications may actually become a state of the overall system that will behave accordingly. Again, we are seeing some hints of this paradigm evolution in the 5G.
The work going on in projects like "the human brain" and "connectomics" may influence our views of networks and shape their evolution. My bet is that we are going to see by 2035 clouds of clouds that will be at the same time connection, processing and storage domains.
All of this, and more, is something we will discuss at the coming IEEE Technology Time Machine on October 21st and 22nd in San Jose, California.