Roadblocks and solutions
In the very first episode of our new series ‘Becoming deep tech savvy – a corporate journey’, we sat down with the Director of R&D at Huawei, Dr. Merouane Debbah, to discuss the topic “Exploring the vast deep tech opportunity spaces: Roadblocks and solutions.”
Merouane: Algorithms are like recipes that are applied to various domains such as Wireless, Computing and AI. At Huawei we are applying deep neural networks and deep algorithms to models to improve the wireless systems. The biggest issue is that we are trying to understand how deep tech impacts corporates and our customers. It’s about thinking outside the box.
Merouane: We do not outsource our research. 50% of the employees at Huawei are in R&D, as opposed to 15-25% in other large corporations. Not just that, a huge section of marketing and business employees also have an R&D background. The survival of a corporate like us is hugely dependent on in-house R&D and inventions we can make today.
Huawei is a high-tech oriented, customer-centric company. We have the 3rd largest R&D investment portfolio, amounting to ~$60billion, and approximately one-third of this goes to fundamental research.
Merouane: There are 3 kinds of research depending on their end goals
The last two fall under the category of fundamental research. The biggest challenge for corporates to explore the vast deep tech opportunity space is to be able to focus and promote technology and vision-driven research.
Merouane: Academia & Industry speak different languages. Generally, people try to make academia speak the industry’s language. At Huawei, we have turned this on its head by trying to understand the way academia works and provide them with the necessary tools for success. The best way to work together is by presenting academia with the plethora of applied problems faced by industry and let the academics work their magic on a problem of their choice in their own turf where they are most productive. The current issue with academia is the lack of data. This can effectively be resolved by developing platforms such as Lagrange research center, where academics work in their own premises, but with continuous and open discussions with industry. Academia is the place to try & fail, repeat & succeed, and this is what sets them apart. We should look at industry as problem creators and academia as solvers.
Merouane: Our vision is what drives commercialization of research. We have the right talent whose job is to capture theorems, future applications of current fundamental research and connecting the dots of present tech with future applications. Academia moves science ahead and Industry creates tools out of it.
Eg. ADSL – Asymmetric digital subscriber line is the internet connection most of us have at home. It is based on OFDM (Orthographic frequency division multiplexing), a technology at the heart of 4G and 5G deployment that uses the basics of FFT – fast fourier transform which builds on FT (Fourier transform), a technology first developed in academia.
Another example is the development of number theory by the mathematician Galois. Joint industry academic research took the number theory to codes and implemented these codes into the tech invented today.
Image processing (wavelets to jpeg 2000) is another brilliant example of Industry/Academic collaboration and deployment of deep tech. We create patents on the tech we develop based on the principles and discoveries by academics. This is how we generate value.
Merouane: There is always a need to create new tech or adapt the existing ones to create value. The origin of this need stems from both the current market and the futuristic vision of the company.
For instance there is ‘certainty research’ that focuses on the short term market and is driven by product development, reduction of complexity, and making improvements. This is generally led by engineers. The margin of breakthrough here is small.
Then there is ‘uncertainty research’. This is where we get random surges and the 10X or 100X leap in knowledge in terms of complexity and innovation. The margin of breakthrough here is huge. The impact generated resonates beyond your own company and acts as a leap for the ecosystem. The notion of an ecosystem (growing with peers) is important here. As a company you need to push the market with your advances. Being alone in a market does not make sense and only increases the risk.
Complex, hi-tech companies like Huawei can only survive because of talents. Talents are not only sourced, but are nurtured in-house with respect to the changes happening around. We upskill our employees in 3 ways:
In the first place, we recruit a skilled workforce of PhDs and researchers. The idea is to bring people with sound fundamental background research as opposed to narrow and specific skills. Recruitment of a workforce like PhDs is necessary in such positions as by nature they understand change and the need to adapt to it.
Innovation cycle comprises theoretical breakthrough, technical readiness, technological innovation, product development and commercialization. Theoretical development is the longest and most risky zone as it comes with its own pace. This is why we look up to academia, before we can apply the Tesla model or Steve jobs model to bring innovation and products out of the research into the market. It is also true that a lot of the research never sees the light of the day, but this is not considered money down the drain. Indeed we appreciate it here, as we learn from our failures. In most of the cases we often end up repurposing the solutions, experience and skills developed for project X in project Y. The mindset of documenting everything including failures, putting it aside and revisiting it in future is extremely important here.
G here stands for “Generations for cellular”.
2G was mobile for voice, 3G was mobile for data, 4G – mobile for internet, 5G mobile for things and 6G is intended for mobile for machines. Each of them has/had challenges. For example, 4G has an IP (Internet protocol) based system. Bringing voice in 4G was a difficult task unlike in 2G. With 5G came the connectivity with things and this generated additional energy requirements. It was extremely hard to tackle – imagine a phone connected to a driving car, but for a moment this connection is disrupted leading to an accident. We needed to solve the pinpoints to avoid issues like these. So we embedded complex machine learning algorithms and tried to find the right protocol to connect intelligence between devices without having a human behind it all.
Now imagine a system that can create its own protocols. 6G is connecting these autonomous systems. 6G is about connecting AIs!
The right ecosystem is the biggest resource for growth. So I will answer this by a question : why are we here in Paris? Because Paris and Moscow are the two best hubs for mathematicians. This is the right ecosystem for us.