The Proto-6G Device: A Parody that May be Useful
The primordial soup of academic and industrial visions on 6G is already revealing some stable shapes, likely to be featured in the next generation of wireless systems. A representative, but not exhaustive, list of this shapes starts with Artificial Intelligence (AI) and Machine Learning (ML) techniques, which are going to play prominent roles in multiple layers of the communication system, starting from the application layers and down to the design oh physical waveforms for communication. For instance, AI is seen as something that will be natively included in the radio access protocols, rather than being added as an afterthought.
Another shape in the 6G primordial soup is the one of semantic communication, motivated by the observation that the technical communication problem of correctly restoring data at the receiver does not consider the meaning or the effectiveness of the data. Having the communication system to focus on transmitting the meaningful data it can, potentially, more effectively achieve some higher-level goal or save energy by not spending it on irrelevant messages. This brings the next feature that is required from 6G, which is the low energy consumption. This feature, besides following the wave of wishing and washing green technologies, is motivated by the observed high power consumption of 5G systems. One reason is that, even if the wireless systems are becoming more energy-efficient, the total number of bits sent through these systems increases, thus pushing the overall power consumption up. Reducing the number of transmitted bits to the most relevant ones is the point of connection to semantic communication.
Then there is the requirement, sometimes a chatter, for low latency, even lower compared to the one in 5G. This is something I do not entirely agree with, as latency is a type of specific timing requirement and the timing requirements of various applications can be met even if the latency is not extremely low. Ideally, one would like to bring the communication latency to the level that is determined solely by the speed of light, reducing the delays introduced by various circuits for computing and communication. This can be, in principle, achieved by joint design across multiple layers of communication system, something known as cross-layer optimization or, in the information-theoretic literature, joint source channel coding. As disussed in this article, joint source channel coding is in itself an important approach towards attaining semantic communication.
The last shape from the primordial soup that will be mentioned here is resilient communication. While in 5G there was a lot of talk about reliability and ultra-reliability (which is not only difficult to achieve but also difficult to measure and validate) resilience is about the capability of the system to adapt to unanticipated conditions and offer certain minimal service, which is of degraded quality compared to the operation under the boundaries of the normal conditions. One way to think about it is that the system may be operational 99.999% of the time, or the communication attempts, but is suffciently elastic to adapt to the bad conditions (e.g. jamming or very poor coverage) in the remaining 0.001 %.
How would, then, a 6G device or service that has all these features look like? Using slight overfitting, we can land on the analog phone, with analog voice channel, as a proto-6G device that has the features discussed above. It applies joint source-channel coding, which leads to low latency, practically determined by the propagation time through the phone line or radio trunks. The communication is semantic and goal-oriented, as at both ends there is natural intelligence that is capable to interpret the meaning. In a sense, the communication was energy efficient, as there was no excessive signaling during the call and no energy consumption at the end nodes; the analog phones did not need to get recharged, they got power supply through the phone line. Of course, this does not consider the energy consumption of the brains of the interlocutors, but the brain is the most energy efficient computing engine we know of. If the quality of the line degrades, then a lot of the information can still be interpreted correctly. This is true even if the amount of interference and noise on the channel gets excessive to levels that have not been anticipated; that is, the analog voice channel is resilient. Finally, not that the powering of analog phones through the phone line was also part of the resiliency features, as the phone was supposed to work even under power cuts.
This is, of course, a parody of a 6G device, but there are some important lessons to be learned:
- The overall communication process is a complex interplay of the computing/intelligence and the protocols for data transmission.
- The interplay of computing and communication can decrease latency or, more generally, meet the target timing requirement, since a significant number of the latency contributors are at the end nodes (coding, compression, etc.).
- Achieving semantic and effective/pragmatic communication is a result of this interplay between intelligence and communication, eventually resulting in energy-efficient and resilient connectivity.
- Analog communication is an ultimate form of joint source-channel coding and can exhibit resilience under unanticipated/extreme communication conditions. First, this is done at the expense of more energy spent for communication (higher power), but also more energy spent for computation at the end nodes to interpret the messages. Second, the resilience of analog communication is also due to the fact that the decoding of data is not conditioned on decoding a control information (metadata), but everything is done jointly: by listening to the voice we get both the data (what is being spoken) and the metadata (who speaks, based on the tone of the voice).
Paradoxically, when we move towards making the advanced digital 6G system, we may need to go to the drawing board and find some inspiration in analog communication.
