IEEE Future Directions
Related Articles
Communications has reached an amazing sophistication, most of the Planet is now blanketed by communications infrastructures. Technology and standards have been the enabling forces.
Till few decades ago the communications infrastructures were designed as conveyor pipes (wired and wireless) able to carry signals whose meaning was irrelevant from the point of view of the transporting infrastructure. The digitalisation of communications reinforced this aspect. A bit is a bit, from a transport point of view. Its meaning was infrastructure independent. It might code an image, a song, a voice, a temperature… Provided the infrastructure has the capability to transport it from A to B within certain time it is fine.
Of course, there are “bits” and then there are “bits”. As an example the indication of a temperature of a seaside location may well be transported in one minute and no one will object. On the other hand the increased temperature of a nuclear reactor needs to be communicated in a fraction of a millisecond to avoid catastrophic situations. Bits representing video images have much more stringent delay requirement than bits transporting still pictures and so on.
As discussion went on, on the interaction between what is being transported and the transporting infrastructure, i.e. should the infrastructure be aware of what is being transported and should behave differently, or should it be neutral, the progress in technology (and in network capacity, management of this capacity within and outside of the network) has made irrelevant the discussion. In most situations the infrastructure can do its job disregarding what is being transported.
In these last decades, and more so in the coming ones, the communications needs have been changing, due to the diversity of communicating entities. Whereas in the past it was about communications among human beings now and in the future a significant portion of communications is and will be among machines.
Communications is about sharing a meaning. This requires both the sharing of the semantics and the transport of this semantics using a syntax. Communications infrastructure support the transmission of the syntax, standards are in place to ensure that the syntax (the signal coding) is smoothly transmitted from A to B. The before mentioned “network neutrality” brings semantics into the equation but only to ensure that the syntax is taken care in an appropriate way.
The semantics is outside of the communications infrastructure, it lays inside our brain. It is our brain that decodes the sound ways and the photons generated by a screen. We are autonomous systems with our own semantic framework that makes possible the understanding.
We are now seeing the emergence of other types of autonomous systems, equipped with a growing level of “intelligence”, that is their own semantic framework and their own capabilities of applying it to “understand” communications, as well as to “generate” communications streams towards other systems.
This new scenario raises the need for an evolution of standardisation in the communications area, no longer limited to the transport but rather extending into the semantics of communications.
Industry is at work to define communications protocols among autonomous vehicles, among robots in production lines, among robots in an home environment. All of these are “closed” systems, the communications being addressed is internal communications and as such, having the control of each communicating entity can be designed, and defined, as in the past. We sit at a table and we agree on what can be communicated and how it should be communicated.
In the next decade, and more so in the following ones, new challenges will come to the fore. Autonomous systems will start to open to the world, meeting and interacting with other systems that are not part of their closed environment. How can communications be facilitated in this new scenario?
We have had some studies in this area in the past when scientists discussed how to communicate with extra-terrestrial forms, where no information is available on their semantic/syntactic underpinning of communications.
We also have some interesting facts that can be learned by observing communications among two persons coming from different culture, e.g. an Australian aboriginal and a European, and even between a person and a dog. The communications semantics and the syntactic models of the parties involved differ significantly and are unknown to each other.
In a way this is the situation confronting the communications among advanced autonomous systems.
Notice how the communication with an extra-terrestrial form differs significantly from the ones with an aboriginal or with a dog. In the former case there is not awareness on the behaviour of the other entity engaged in communications, in the latter there is (unless you are sending a letter to an aboriginal or to the dog…).
The communications among “independent” autonomous systems, i.e. not being part of a closed system where communications is designed as part of the closed system, can be approached in both ways, as if it were a communication with an extra-terrestrial system and as if it were a communication with a dog… (this latter is preferable as comparison to an aboriginal since in the case of communications with a dog the “intelligence” in the two systems is not comparable, whilst with an aboriginal the intelligence and the underlying intelligent support infrastructure is the same).
The easiest form of communication to mimic is the one with a dog, because we know that is works (!) and because we can continually adjust it based on evolving feedbacks.
This form of communications can be classified as implicit communications, whereas it is based on the interpretation made by each party of the behaviour of the other, thus leading to a change in its own behaviour that will be detected by the other party leading in turns to a change of the other party behaviour.
This form of communications requires an “understanding” by each party of the other party behaviour by interpreting external changes (wagging the tail, type of barking, hands gesture, smiling…).
Autonomous systems are developing a model, and an understanding, of their environment, and detect and interpret changes in their environment. The presence of other autonomous systems in their environment, expected or unexpected, is introducing changings and these are a form of implicit communications.
Standards are needed to simplify as much as possible the understanding of what is going on. A car starting to blink a directional light can be assumed to be about to change its trajectory, although it is not 100% guarantee, nor it is the reverse true (a car NOT blinking will not change its path).
