A Custom Hardware CCA Engine for Real-time SSVEP-based BCI Applications
| Autor: | Reza Karkon, Seyed Mohammad Reza Shahshahani, Hamid Reza Mahdiani |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
010302 applied physics
Signal processing Computer science Interface (computing) 02 engineering and technology 01 natural sciences 020202 computer hardware & architecture Computer engineering 0103 physical sciences 0202 electrical engineering electronic engineering information engineering Decomposition (computer science) Multiplication Canonical correlation Field-programmable gate array Eigenvalues and eigenvectors Brain–computer interface |
| Zdroj: | 2020 20th International Symposium on Computer Architecture and Digital Systems (CADS). |
| DOI: | 10.1109/cads50570.2020.9211863 |
| Popis: | Canonical Correlation Analysis is widely exploited in different signal processing applications such as detection of the brain internal activity in an SSVEP based Brain-Computer Interface system. The main disadvantage of the CCA algorithm is its computationally intensive nature which demands numerous complex matrix transformations including inversion, eigenvalue and eigenvector extraction, multiplication, and decomposition. These complexities prevent its efficient hardware implementation while its software implementation also is not so fast to be exploited in real-time applications. In this paper, some simplifications are applied to an 8-channel CCA algorithm to make its hardware implementation feasible. The effects of these simplifications on the output accuracy of the algorithm are also demonstrated in the paper. A custom fully hardware fixed-point CCA engine is then developed based on the simplified algorithm to achieve higher performance and less power consumption. The FPGA synthesis results of the design are also included in the paper. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|