Generalized belief propagation based detector for two dimensional magnetic recording using a probabilistic voronoi discrete grain channel model
Washington State University
Master of Science (MS), Washington State University
2015
:
https://hdl.handle.net/2376/102125
The storage capacity of hard disk drives is related to its storage density, typically given in Tb/in2 . Due to physical limitations, magnetic recording technology is reaching a point where the maximum density cannot be further increased using conventional one dimensional recording. To address this problem, Two Dimensional (2D) Magnetic Recording (TDMR), a technique which writes and reads data in 2D patches on magnetic hard drives, has been proposed recently. In the 2D write-read process, and at high bit densities, bits are subject to interference and overwrite effects from adjacent bits in both dimensions. Signal processing techniques are needed to ameliorate these effects and successfully read the original written bits. Models of varying complexity, ranging from a real physical model to a simple four grain model, which simulates this new technology, have been proposed. A disadvantage of the real physical model is the increased computational complexity for the signal processing algorithms. In this thesis, a medium complexity model and a detector based on this model are proposed which can be used to explore the potential increase in areal density with TDMR. The proposed technique uses a probabilistic Voronoi discrete grain model (PVDGM) to generate the magnetic grains. The PVDGM generates convex grains with random size and shapes as well as random locations. Several model parameters can be used to control the closeness of the model to the physical model. The histogram area of the grains in the PVDGM matches the statistical distribution of a true Voronoi-Poisson process. A GBP based detector based on the PVDGM is applied to a 2x2 bit area to estimate binary LLRs. Simulation results shows proper identification of overwritten and nonoverwritten bits for real data modeled with a uniform distribution and at density of 1.2 grains-bit.
5
20
- Generalized belief propagation based detector for two dimensional magnetic recording using a probabilistic voronoi discrete grain channel model
- Guillermo Ramirez-Conejo
- Krishnamoorthy Sivakumar (Co-Chair)Benjamin Joseph Belzer (Co-Chair)Thomas Robert Fischer (Committee Member) - Washington State University, Electrical Engineering and Computer Science, School of
- Washington State University
- Electrical Engineering and Computer Science, School of
- Master of Science (MS), Washington State University
- Washington State University; [Pullman, Washington] :
- 71
- 99900525056101842
- English
- Thesis