OK, let's sum up what we have seen so far and then we can go to our gedankenexperiment.
The network architecture we have today is the answer to a number of requirements/constraints that are no longer there. At the same time we are now faced with requirements that did not exist just twenty years ago. Of the many differences I have pointed out three stand out and are steering towards a network revolution: the abundance of -everything-, the multiplicity of players and the flexibility in the access to resources.
|Usage Frequency||3' - 5 times a day||A hundred times a day||A thousand times a day|
|Market||millions||billions||hundred of billions|
|CAPEX||centralized||30% centralised, 70% distributed||10% centralised, 20% clustered 70% distribut.|
|Players||incumbent||incumbent for the network, OTT for services||incumbent for core network, Multiproviders at the edges, OTT for services and network|
|Revenues||Direct ? services tied to the network||Direct + Indirect - fee for network access, service revenues uncoupled from network, devices||Indirect. Tied to device sale/usage ? services coupled to devices and user|
|Technology||Electromechanical||Electronics/Optics||Electronics + /Optics +|
|Access||Fixed||Fixed - Mobile||Mobile|
|Intelligence||Scarce - Centralized||Sufficient - Distributed||Abundant - Distributed|
|Architecture||hierarchical - rigid||hierarchical - flexible- for network, flat for services||flat, dynamic, cooperative and flexible for network and services|
So let's pretend we have plenty of smart terminals, were energy constraints are not an issue, each one with abundant processing and storage and able to access a broad spectrum. What kind of business would we be looking at?
The drive that led to the development of Telecommunications was the control and the expansion of the network. The network was the business. Today there is more money making in selling (smart) phones than in selling access (and network services). And today the (smart) phones are the one creating the platform to offer services, not the network. If you look for services you turn to the Android, Apple, MS Stores, not to the network.
Phones are making up some 70% of the $ spent every year on telecommunications “equipment”, they are produced in billions of pieces (2-3 billions a year) and have a lifetime of 2 years (in mature markets even less). This means that every 2 years we can look at a new generation of our service platform. Wouldn’t it make sense to start designing the network from the phones?
In the past the phones were designed based on the network whilst in the last 10 years (since the iPhone I would say) they have been designed independently of the network, some would go as far as saying “in spite of the network”. This means that the phone industry is now on a path of its own.
Hence, this is the approach I propose for our gedankenexperiment: let’s design our network architecture from the devices up.
We have billions of devices that can serve as network nodes, that have the required processing/storage power (and then more) and that can connect with one another and with a variety of network gateways.
It is not just phones. Cars represent another interesting possibility to be network nodes. They are at the same time more “mobile” than a phone and more fixed (most of the time the car is standing still, parked along a sidewalk or in a garage). Media centers (progressively including smart televisions) are also potential network nodes and may serve as data hubs.
Let’s forget the existing network (we can do that, it is a gedankenexperiment) and let’s design our new network architecture by using only these new network nodes.
Each of these (phone, car, media center) is privately owned and per sé does not provide any guarantee to be available. A phone may run out of battery or quite simply its owner may elect to not have her phone become a network node.
However, if we have hundreds of nodes that can be used indifferently at any particular location the unavailability of few of them is not going to make any significant difference.
Interestingly, when a new node joins the network the whole network increases its capacity, vs today’s situation where any new terminals accessing the network competes for the network resources, hence decreasing the “pro-capita” resource available.
In a dense device environment the traffic can be routed on several alternative paths. Even more interestingly, a dense radio space flips from being a number of overlapping and interfering pathways into a communication fabric where interference is solved by communications among the communicating entities (there is a nice theorem, that has been proven in practice, stating that if you have “n” transmitters and “n+1” receivers separated from one another at least half a wavelength then you can manage on the same spectrum n communications in parallel).
How could the routing be made? One could say: “as it is done today” in Internet. However, today the number of links is practically limited by the (hierarchical) network architecture. The mesh network created by interconnected devices is completely flat and the number of links explodes. This is good from the point of view of overall network availability but it is not good in terms of potential number of “hops” in a single session that would result in a not acceptable latency (delay). Hence the devices need to have a routing map that, depending on the kind of service, makes sure the number of hops stays below an acceptable thresholds. This should not be a problem for most devices (like smart phones) that have all the needed storage and processing capacity embedded.
This is an area where we clearly need research. The focus should be on enabling a communication based on a “communication fabric” rather than on a “communication network”. In a way it looks like gravity. Objects are not interacting one another, they change the space time and because of this change orbits change. It may be seen as stretching a bit too much the point, but I feel it gives the idea. As soon as we shift from the concept of an infrastructure to the concept of a fabric we open up new ways of looking at the problem.