| 研究実績の概要 |
In 2016, the LatticeNet Project achieved the following results, which are divided into 4 work packages (WP).
WP1. For decoding low-density lattice codes, we gave a new decoding algorithm which has the best-known performance-complexity tradeoff for such lattices of dimension 1000 and smaller; these results were published as a journal paper.
WP2. We obtained several new results for the design of power-efficient lattice codes. (a) We provided necessary and sufficient conditions to select a lattice as a shaping lattice, even if that lattice differs from the coding lattice. (b) Barnes-Wall and E8 lattices can be used to shape low-density lattice codes. (c) convolutional code lattices can be used to shape low-density lattice codes. These results are, or will be, published as journal papers.
WP3. For the relay channel, codes for data storage (write-once memory code) are effective when applied to wireless channels. For the parallel relay channel, a new mechanism increases the efficiency when applying hybrid-ARQ to the lossy-forwarding method. Both of these results were published as journal papers.
WP4. We applied lattice codes to the multiple-access relay channel, a type of Gaussian network, using compute-and-forward techniques. We showed that naive application of lattice codes results in poor performance, and solved this problem by allowing the relays to flexibly select linear equations. This work demonstrates the feasibility of nonorthogonal signaling in future wireless communications systems. (WP5 was completed in 2015.)
|
