Integrative, Hybrid and Complex Systems Jaekyun Moon, University of Minnesota-Twin Cities Tong Zhang, Rensselaer Polytechnic Institute COLLABORATIVE: GOALI: New Algorithm and Architecture Solutions for Ultra-High-Throughput Single Carrier Communication System Geared to High Capacity Storage
Intellectual Merit:
High-capacity, high-throughput disk drives are a critical component of many modern computing systems. This research aims to devise, integrate, and jointly optimize soft-output equalization and error control algorithms for application to high-capacity storage. The project will also lead to a very-large-scale-integrated (VLSI) architecture solution for enabling high-capacity, high-throughput storage of digital information, taking into account the constraints and characteristics of the disk drive channel. In particular, a "read channel" solution with a throughput exceeding 5 gigabits per second (Gbps) is targeted for disk drives approaching a storage density of 1 terabit per square inch. To achieve this throughput while maintaining capacity requires close and active interaction between research in algorithms and VLSI architecture. Innovations are necessary in both areas. The project will first apply algorithm and VLSI implementation co-design methods to a soft-output trellis detector and error correction decoder, the two most computationally-intensive and, hence, energy and silicon-hungry functional blocks. The project will then pursue integrated design and optimization of the signal detection sub-system that includes an equalizer and trellis detector and of the coding sub-system that includes inner error control coding and outer error correction coding. Third, system optimization is applied to determine the best configuration of individual functional blocks that balances performance and implementation complexity of the entire data path. Advanced soft-information processing through signal detection and decoder blocks serves as the technical cornerstone, with a focus on forward-only trellis-based soft-output equalizer and low-density parity-check (LDPC) error correction coding combined with high-rate inner error control coding.
Broader Impact:
This research has the potential to facilitate the development of extremely high-capacity disk drives for data storage, with a wide array of applications including portable mass storage devices for the "mobile Internet" and wireless multimedia broadcasting. Close collaboration with industry will facilitate technology transfer to industry. The project will also educate students on read channel algorithms and implementation, naturally providing industry with potential employees with knowledge of both theory and practical system constraints. Women and other under-represented groups will be actively recruited as research assistants. Curriculum development includes a new graduate-level course on signal processing, enhancements to one undergraduate course, senior and honor's design projects, and incorporating findings on the coding/signal processing algorithms into existing graduate courses on digital communication, information theory and error control coding.
