- On the cusp of the deployment of 5G technology in cellular telecommunications networks around the planet, the telecommunications industry is searching for methods to optimize information transmission in a “noisy channel” in order to decrease the number of errors to a minimum, as well as to save costs.
- The error-free transmission of information in a noisy channel is possible since the 1940s. The solution was to send a larger number of bits than the information bits so that they become redundant and resistant to errors. A formula dictates the optimal throughput as a function of the error rate introduced in the channel, what is called the channel capacity. However, this solution has many drawbacks, the principal one being the large amount of computation required to decode the information after its transmission through the channel. Recently Polar Codes, with a throughput arbitrarily close to the channel capacity, have addressed this problem with a decoding timeframe proportional to the quantity of information transmitted. The disadvantage is that very large Polar Codes are necessary to reach a vanishing probability of errors. The invention described herein addresses these problems.
- Arising from research in Physics at the University of Sherbrooke in Canada, the invention of Convolutional Polar Codes is a significant advance. The principle of successive cancelation decoding used is the same as for traditional Polar Codes but the algorithm and the communication protocol performing this calculation are new. The key innovation is the utilization of tensor networks in error correction, an approach recently introduced in the setting of quantum many-body physics for performing certain calculations. Recasting the decoding problem as a tensor network contraction, we arrive at a new understanding of the efficient decoding of Polar Codes. This provides a new perspective on certain problems and allows us to discern calculations that would be very complex without this framework.
- The technology of Convolutional Polar Codes is a new family of error-correction codes that generalizes Polar Codes in a natural way. Convolutional Polar Codes converge toward a vanishing error probability twice as fast as Polar Codes, and even when the quantity of information is relatively small, attain error rates substantially lower than Polar Codes.
- A program written in C++ and Julia has been designed that enables the encoding of information into a Convolutional Polar Code, the decoding of the information in a noisy channel, and the simulation of the noisy channel. Polar Codes have already been adopted by the 5G international standards thus making the extension straightforward. Furthermore, the implementation of Convolutional Polar Codes into ASIC chipsets is already at the planning stage.
- Data Storage
- Decoding of Polar Codes on channels with memory
- Satellite communications
Business Advantages of this solution
- A technology offering a major performance gain for 5G in telecoms.
- Standardization: Polar Codes, a standard-essential error-correction technology, were ratified in 2016 as part of the 5G New Radio enhanced mobile broadband (eMBB) standard, by the international standards organization 3GPP which unites the worldwide wireless communications industry. The innovation described herein allows an improved 5G!
- Error suppression rates: Superior to Polar Codes – a more reliable communication, or the utilisation of codes of smaller size.
- Competitive advantage: Higher-performing than Polar Codes – allow communication in a more reliable manner, or the use of smaller code sizes.
- A global leader in the industry confirmed implementing this code and observing the promised performance.
- Speed: Decoding time is similar than with Polar Codes.
- Extended Applications: The decoding technique can also be used for channels with memory.
- Channel capacity: Asymptotical attainment of the capacity of the channel.
- Market: As of April 2019, the Global Mobile Suppliers’ Association had identified 224 operators in 88 countries that are actively investing in 5G.
- Commercial deployments, starting in 2019, are expected to eventually reach hundreds of millions of wireless devices.
- The jump from 4G to 5G is far greater than any previous jump because 5G has the potential to be up to 100 times faster than 4G; allowing self-driving cars in the future, among many other applications.
- Reliability: Error rates substantially lower than with Polar Codes.
- Numerical proof that Convolutional Polar Codes:
- Reach channel capacity asymptotically.
- Are decodable in linear time.
- Error probability converges 2 times faster toward 0 than for conventional Polar Codes.
- Efficiency: Less redundancy for the same channel content; for a similar redundancy, a lower error probability.
- ASIC development: Polar Code encoders and decoders have already been widely implemented in high-speed and low-power VLSI chipsets, and the work is planned for extending to Convolutional Polar Codes.
Intellectual Property Status
TECHNOLOGY READINESS LEVEL
- TRL5 – Technology proven in simulation, and the integration into a VLSI-Chip is planned.
- Licenses available.
- US Provisional patent application no. 62/466,414, filed on March 3rd, 2017.
- PCT patent application no. PCT/CA2018/050259, filed on March 5th, 2018.
- Commercial partner
- Development partner
Main Inventor Contact
- David Poulin
- 819 821-8000, ext. 62054