A Pessimist's Guide to 6G

Being at the IEEE Globecom conference is always an opportunity to get a live snapshot of the state of communication engineering. This edition was impeccably organized in beautiful Kuala Lumpur. Conferences are great places to gather insights from colleagues that inspire reflection, encouraging one to take a step back and critically examine research on communication and wireless technologies. It also motivated me to revisit what I had been writing in the past — after the events in 2016, 2017, and 2019 — and self-critically assess what turned out to be less correct and what is still valid (left as an exercise for the reader :)

The optimism at the conference is reserved for the papers, presentations, and most of the panels. Papers and presentations are, by definition, optimistic, as they usually showcase the gains and excellence of a method or a product; there is no focus on the negative results. Nevertheless, some papers are presented in a way that does not radiate optimism; this could be due to language barriers or a lack of focus on presentation skills, but it is something that universities and companies should work on. The optimism was evident in some of the excellent keynotes, which discussed redefining communication standards, the symbiosis of Large Language Models (LLMs) and telecommunications, integration of sensing and communication, or the role of visible light communication. Panels are places that can accommodate some of the conference skeptics, playing an important role in taming the hypes or downplaying the significance of specific technologies, such as AI or Reconfigurable Intelligent Surfaces (RIS).

The true pessimism is reserved for the informal discussions at the breaks and conference dinners. This is the place for the adult-in-the-room prophetic statements that "X will never work" or "Y is doomed". Here are some of those: (1) Machine Learning (ML) will never be able to help in designing new communication protocols, due to large complexity; (2) Open RAN will never work due to high computational requirements; (3) Nobody will ever use RIS as there are too many problems with it, such as affecting the bands of multiple operators; (4) Integrated sensing and communication (ISAC) will have a spectrum problem, as companies do not want to give up on a pure spectrum used for radar and accommodate unpredictable communications.

These are valid observations, and instead of succumbing to despair, one can view them as a set of relevant research problems to be addressed. In (1), what is the combination of human-based design and ML that can lead to new communication protocols? In (2), what are the use cases where Open RAN shines, whether in security, privacy, or adaptation to terminals with variable performance? For instance, overall latency or energy consumption is affected by the terminals, and we lack control over the implementation of the terminals attached to the network, as long as they adhere to the prescribed standardization. Thus, it could be the radio network that adapts to the current set of attached terminals. Regarding (3), the pushback against RIS is partially due to the extreme research production in the area. Nevertheless, there is justified criticism of the overall energy consumption, while skepticism about it being an integral part of the communication infrastructure is related to the fact that relays never gained mainstream adoption in cellular networks. Yet, there are clear use cases with respect to localisation and sensing, and the problems of bandwidth and overall energy consumption are relevant research topics. Finally, (4) is justified by business interests and the inertia of the spectrum regulation process. However, in the upcoming fusion of the physical and digital world, I believe that Base Stations will play a significant role as sources of trustworthy sensing information, and, since they already play a role in communication, Base Stations with ISAC present a highly plausible future in wireless systems.

However, the set of relevant research problems should not merely be defined as a reaction to the 6G pessimism, but there should be a clear purpose within the bigger socio-economic-academic picture. For instance, one reason for why we should develop 6G is because the odd mobile generations fail (that is why 5G fails or will fail), while the even ones succeed. This is rather a statistical overfitting over a very few data points, which becomes almost equivalent to superstition. (As a side remark on the occult and superstition, 3GPP adopted the term "5G numerology", while the traditional use of the term numerology is the belief in an occult, divine or mystical relationship between a number and one or more coinciding events.)

Where should we, then, look for the optimism and the right reasons to develop 6G? I think it should be at the intersection of the following five axes, ordered in decreasing order with respect to their scope:

1. Societal: Defining a clear big-picture purpose by asking the question: What are the societal problems that can be solved by a better digital infrastructure and how should 6G contribute to them? There has been an attempt to address this question by relating 6G to sustainability and values. The recipe for bringing in sustainability is to relate whatever we are doing to the 17 Sustainable Development Goals and immediately feel optimistic and relevant. Those 17 goals are undeniably highly important and it would be great for 6G to contribute to them, but I have yet to see a convincing way to relate them. To the critics — yes, I have read articles about that and seen multiple talks, including now at IEEE Globecom. Nevertheless, most of the time those papers and talks can be described as follows. Let us say that I am invited to give a talk on the Theory of Relativity, for which I know very little. Then my talk goes as follows: "Special Theory of Relativity and the effects on time dilation; time is related to latency and I have worked with latency and Age of Information (AoI) in communication systems, so let me tell you something about that."
2. Digital technology: What are the problems in the digital technologies that can be solved by 6G? For instance, the public focus at the moment is on capable AI models, such as LLMs. Are there problems that these technologies will create, such that we need new breakthroughs in communication to address them? I have put some thoughts about it here: when generative AI becomes accessible at a large scale, we need an infrastructure that will help in supporting the objective truth of the combined physical-digital worlds. This is where communication and, specifically, Integrated Communication and Sensing, can play an important role.
3. Communication Infrastructure: Are there new elements in the communication infrastructure that need further research? I have already mentioned Open RAN and Base Stations equipped with ISAC. Non-terrestrial and satellite communications represent a major upgrade to the infrastructure: offering truly global connectivity by connecting the unconnected areas and people, resilience (recall the cutting of optical fibers during the current military conflicts) and futuristic large-scale control infrastructure (for example, flying vehicles). It does imply that these networks are an important element of 6G. However, when talking about 6G, the elephant in the room is the wide use of unlicensed technologies, such as Wi-Fi and Bluetooth. Even more, it is always sobering to have someone that points out to the elephant. I have recently attended an excellent symposium organized by ELLIIT in Sweden, with deep and focused technology presentations. One of the academics that is outside the wireless area told us that they expect us to "develop all the elements of the 6G technology, such as mobile, Wi-Fi, Bluetooth, and IoT standards"; this is not a mistaken statement, but a nice revelation that sometimes we choose not to see some facts. Ironically, most of the Zoom talks about the greatness of 5G and 6G are done over a laptop that uses a Wi-Fi connection.
4. Communication industry: Which are the areas of growth for the communication industry and who are the new market players interested in the 6G technology? When assessing the need for 6G, is always instructive to consult the stock market and see whether the 6G optimism of a company is proportional to its stock market performance. 5G brought the vertical industries as stakeholders in mobile communication standards and this continues to hold in 6G. This set of stakeholders is expanded by the companies dealing with satellite communications and computation/AI. Nevertheless, some of the incumbent wireless players may not have an immense interest in new 6G technologies, judging from their 2016-flavored talks at IEEE Globecom.
5. Academic research. What are the intellectually important problems to be addressed by communication theory and communication engineering? While this can be a source of immense inspiration and breakthroughs, there is also a danger of “l’art pour l’art” research that has no relevance for real-world systems. An example of this is the set of "unrejectable papers": defining a problem in a popular area, showing its mathematical structure (e.g. non-convexity), throwing some theorems, and getting at least 30 % gain. If we consider a sequence of 50 papers that build on top of each other with 30% gains (which is rather conservative), then the gain with respect to the original benchmark is 1.3⁵⁰=497929, that is, improvement of around 500 000 times (!). Since we are not seeing that in reality, it must be that some of the assumptions are wrong, or not relevant, or not taking into account some cost in the system. It would thus be great to direct the intellectual curiosity by thinking about the assumptions and the purpose and broader context, outlined by the four axes above.

As a concluding remark, in embracing skepticism and acknowledging pessimism, we find not only challenges but a compass pointing towards the most relevant problems in the pursuit of advancing communication technologies for the next generation.