2021 Fiscal Year Annual Research Report
Lattice Codes for Gaussian Wireless Networks Beyond 5G
| Project/Area Number |
19H02137
|
|---|
| Research Institution | Japan Advanced Institute of Science and Technology |
|---|
Principal Investigator |
KURKOSKI Brian 北陸先端科学技術大学院大学, 先端科学技術研究科, 教授 (80444123)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
落合 秀樹 横浜国立大学, 大学院工学研究院, 教授 (20334576)
|
|---|
| Project Period (FY) |
2019-04-01 – 2023-03-31
|
| Keywords | lattices / information theory / coding theory / wireless networks / wireless communications |
|---|
| Outline of Annual Research Achievements |
We designed new polar code lattices with short block length for low-latency applications.We optimized the design by semi-analytically evaluating the probability of error of each Construction D layer. The resulting design has error-rate performance within 0.2 dB of the best-known code, but with significantly lower decoding complexity. For complex-valued wireless communications channels, we lowered the complexity of decoding low-density lattice codes by proposing a reliability-based decoding algorithm. Lattice shaping is a well-known technique to reduce transmit power, and our new lattice code construction uses convolutional codes for shaping and low-density parity-check code for coding. Our construction has the highest-known shaping gain of 1.25 dB of any practical construction.
|
| Current Status of Research Progress |
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
Reason
In FY2021, we published one journal paper in IEEE Transactions on Communications and five conference papers at IEEE Information Theory Workshop and IEEE International Symposium on Information Theory.
|
| Strategy for Future Research Activity |
This year, we plan to develop CRC-enabled lattice codes and show they can improve the decoding performance on the AWGN channel when lattice expansion is used. We design lattice codes that possess the new key property of group isomorphism, in addition to the other properties of coding gain, shaping gain and efficient encoding/decoding. QC-LDPC codes for coding gain and convolutional code lattices for shaping gain are the most promising approach. CRC-enabled lattice codes are applied to multi-terminal networks, with the goal of improving the throughput of multiple-access scenarios enabled by the compute-forward framework. CRC-enabled lattice codes allow the compute-forward receiver to retry decoding without requesting re-transmission.
|
