Communication systems, both wireless and based on optical fibres are the backbone of our digital society. Today, we transmit data from and to everywhere on the globe at blazing speeds (e.g., through undersea optical cables or with our mobile phones) and with devices that often do not even require batteries (e.g., tiny edge devices that form the internet of things). These technologies that connect the world are enabled by rapidly advancing algorithms and by extremely sophisticated integrated circuits that implement them. In the past decades, we often relied on progress in semiconductor technology to mitigate the rapidly increasing complexity of communication technologies and to hide the growing power and energy consumption associated with advanced pervasive communications. Unfortunately, as Moore’s law starts to stall, progress must rely on more careful joint optimization of communication technologies (i.e., algorithms) and the circuits that realize them.
In our SwissCHIP research we focus on the technologies that are required to enable the 6th generation of communications, which is scheduled to hit the market in 2030. 6G systems will bring a further increase in data rates, better reliability, shorter latency, and higher system capacity to cover the needs of AI and heavily data driven applications. It will also connect even more or almost all devices of our daily lives, leading to trillion connections that must be managed, supplying even the most basic devices with artificial intelligence from the cloud. The implementation challenges to realize these technologies can in many cases be traced back to few critical building blocks with rapidly increasing complexity and to new technologies introduced in 5G that have not been explored previously. These blocks and technologies are the focus of our work, targeting both high performance and energy efficient operation for sustainability.
More specifically, in the first year of our research in the SwissCHIPs program, we focused on three aspects: first, we developed new algorithms for error control coding on wireless links and on the backbone of 6G wireless systems, which relays data between base stations and continents. Building on 5G technology, we have developed new coding schemes and algorithms that are specifically tailored to be implementable despite the reduced gains available from modern technologies. Algorithm/architecture co-design is the key to be able to actually realize the ambitious goals of 6G systems and our algorithms achieve orders of magnitude better efficiency than previous technologies. After a first year of research, we are now ready to implement and demonstrate these technologies in silicon and two silicon chips are ready for tapeout thanks to SwissChips funding. A second aspect of 6G is the inclusion of ultra-low-power technologies for example for remote sensors that operate partially even without batteries. In our first year of research, we developed a new circuit design library that enables such extreme low power applications to be used for example in wakeup radios that can stay alert 24/7 with very small power consumption until a specific trigger is received to jump into action. A first test chip proving the technology is currently under way. A third research area is a new paradigm that is intended to fuel the growth of 6G beyond straightforward communications. The idea is to use communication signals also for sensing and thereby enable new applications (e.g., privacy preserving surveillance or contactless healthcare). Over the past decade we have already investigated fundamental principles and demonstrated potential applications, such as remote sensing of vital signs or intruder detection based on regular communication signals. In the last months, we have demonstrated this idea with a modified open source WiFi radio, initially on an FPGA prototype. With SwissChips it is now time to turn these demonstrators into silicon to also prove the viability in real-world scenarios (for example to integrate the technology into a smart watch that could sense your breathing even while is charging during the night on the side of the bed).
Further to our work on focusing on communications, we have also worked on improving critical building blocks for complex SoCs in general. One of the most critical building blocks are memories as they consume most of the area and power, taking advantage of our background in communications applied to a wider portfolio of systems. In our research, we have realized new technologies to make memories more reliable (using ideas from wireless communications). These works are of specific interest in non-terrestrial systems and in automotive systems, where integrated circuits are subject to stringent reliability concerns and work in harsh environments. Technologies developed by our lab mitigate these concerns without significant overhead and thereby contribute to more reliable integrated circuits in general and are an important step towards orbital electronics and the use of electronics and AI in harsh and safety-critical environments (e.g., automotive and healthcare).